shader/track: Search inside of conditional nodes

Some games search conditionally use global memory instructions. This
allows the heuristic to search inside conditional nodes for the source
constant buffer.

.travis/common/travis-ci.env

@@ -6,6 +6,8 @@ TRAVIS_BRANCH
 TRAVIS_BUILD_ID
 TRAVIS_BUILD_NUMBER
 TRAVIS_COMMIT
+TRAVIS_COMMIT_RANGE
+TRAVIS_EVENT_TYPE
 TRAVIS_JOB_ID
 TRAVIS_JOB_NUMBER
 TRAVIS_REPO_SLUG

src/audio_core/stream.cpp

@@ -68,7 +68,7 @@ static void VolumeAdjustSamples(std::vector<s16>& samples) {
     }
 
     // Implementation of a volume slider with a dynamic range of 60 dB
-    const float volume_scale_factor{std::exp(6.90775f * volume) * 0.001f};
+    const float volume_scale_factor = volume == 0 ? 0 : std::exp(6.90775f * volume) * 0.001f;
     for (auto& sample : samples) {
         sample = static_cast<s16>(sample * volume_scale_factor);
     }

src/core/CMakeLists.txt

@@ -140,8 +140,6 @@ add_library(core STATIC
     hle/kernel/svc_wrap.h
     hle/kernel/thread.cpp
     hle/kernel/thread.h
-    hle/kernel/timer.cpp
-    hle/kernel/timer.h
     hle/kernel/vm_manager.cpp
     hle/kernel/vm_manager.h
     hle/kernel/wait_object.cpp

src/core/hle/ipc_helpers.h

@@ -216,6 +216,11 @@ private:
 
 /// Push ///
 
+template <>
+inline void ResponseBuilder::Push(s32 value) {
+    cmdbuf[index++] = static_cast<u32>(value);
+}
+
 template <>
 inline void ResponseBuilder::Push(u32 value) {
     cmdbuf[index++] = value;
@@ -234,6 +239,22 @@ inline void ResponseBuilder::Push(ResultCode value) {
     Push<u32>(0);
 }
 
+template <>
+inline void ResponseBuilder::Push(s8 value) {
+    PushRaw(value);
+}
+
+template <>
+inline void ResponseBuilder::Push(s16 value) {
+    PushRaw(value);
+}
+
+template <>
+inline void ResponseBuilder::Push(s64 value) {
+    Push(static_cast<u32>(value));
+    Push(static_cast<u32>(value >> 32));
+}
+
 template <>
 inline void ResponseBuilder::Push(u8 value) {
     PushRaw(value);

src/core/hle/kernel/kernel.cpp

@@ -18,7 +18,6 @@
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
 #include "core/hle/kernel/thread.h"
-#include "core/hle/kernel/timer.h"
 #include "core/hle/lock.h"
 #include "core/hle/result.h"
 
@@ -86,27 +85,12 @@ static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] int cycles_
     }
 }
 
-/// The timer callback event, called when a timer is fired
-static void TimerCallback(u64 timer_handle, int cycles_late) {
-    const auto proper_handle = static_cast<Handle>(timer_handle);
-    const auto& system = Core::System::GetInstance();
-    SharedPtr<Timer> timer = system.Kernel().RetrieveTimerFromCallbackHandleTable(proper_handle);
-
-    if (timer == nullptr) {
-        LOG_CRITICAL(Kernel, "Callback fired for invalid timer {:016X}", timer_handle);
-        return;
-    }
-
-    timer->Signal(cycles_late);
-}
-
 struct KernelCore::Impl {
     void Initialize(KernelCore& kernel) {
         Shutdown();
 
         InitializeSystemResourceLimit(kernel);
         InitializeThreads();
-        InitializeTimers();
     }
 
     void Shutdown() {
@@ -122,9 +106,6 @@ struct KernelCore::Impl {
         thread_wakeup_callback_handle_table.Clear();
         thread_wakeup_event_type = nullptr;
 
-        timer_callback_handle_table.Clear();
-        timer_callback_event_type = nullptr;
-
         named_ports.clear();
     }
 
@@ -146,11 +127,6 @@ struct KernelCore::Impl {
             CoreTiming::RegisterEvent("ThreadWakeupCallback", ThreadWakeupCallback);
     }
 
-    void InitializeTimers() {
-        timer_callback_handle_table.Clear();
-        timer_callback_event_type = CoreTiming::RegisterEvent("TimerCallback", TimerCallback);
-    }
-
     std::atomic<u32> next_object_id{0};
     std::atomic<u64> next_process_id{Process::ProcessIDMin};
     std::atomic<u64> next_thread_id{1};
@@ -161,12 +137,6 @@ struct KernelCore::Impl {
 
     SharedPtr<ResourceLimit> system_resource_limit;
 
-    /// The event type of the generic timer callback event
-    CoreTiming::EventType* timer_callback_event_type = nullptr;
-    // TODO(yuriks): This can be removed if Timer objects are explicitly pooled in the future,
-    // allowing us to simply use a pool index or similar.
-    Kernel::HandleTable timer_callback_handle_table;
-
     CoreTiming::EventType* thread_wakeup_event_type = nullptr;
     // TODO(yuriks): This can be removed if Thread objects are explicitly pooled in the future,
     // allowing us to simply use a pool index or similar.
@@ -198,10 +168,6 @@ SharedPtr<Thread> KernelCore::RetrieveThreadFromWakeupCallbackHandleTable(Handle
     return impl->thread_wakeup_callback_handle_table.Get<Thread>(handle);
 }
 
-SharedPtr<Timer> KernelCore::RetrieveTimerFromCallbackHandleTable(Handle handle) const {
-    return impl->timer_callback_handle_table.Get<Timer>(handle);
-}
-
 void KernelCore::AppendNewProcess(SharedPtr<Process> process) {
     impl->process_list.push_back(std::move(process));
 }
@@ -247,18 +213,10 @@ u64 KernelCore::CreateNewProcessID() {
     return impl->next_process_id++;
 }
 
-ResultVal<Handle> KernelCore::CreateTimerCallbackHandle(const SharedPtr<Timer>& timer) {
-    return impl->timer_callback_handle_table.Create(timer);
-}
-
 CoreTiming::EventType* KernelCore::ThreadWakeupCallbackEventType() const {
     return impl->thread_wakeup_event_type;
 }
 
-CoreTiming::EventType* KernelCore::TimerCallbackEventType() const {
-    return impl->timer_callback_event_type;
-}
-
 Kernel::HandleTable& KernelCore::ThreadWakeupCallbackHandleTable() {
     return impl->thread_wakeup_callback_handle_table;
 }

src/core/hle/kernel/kernel.h

@@ -22,7 +22,6 @@ class HandleTable;
 class Process;
 class ResourceLimit;
 class Thread;
-class Timer;
 
 /// Represents a single instance of the kernel.
 class KernelCore {
@@ -51,9 +50,6 @@ public:
     /// Retrieves a shared pointer to a Thread instance within the thread wakeup handle table.
     SharedPtr<Thread> RetrieveThreadFromWakeupCallbackHandleTable(Handle handle) const;
 
-    /// Retrieves a shared pointer to a Timer instance within the timer callback handle table.
-    SharedPtr<Timer> RetrieveTimerFromCallbackHandleTable(Handle handle) const;
-
     /// Adds the given shared pointer to an internal list of active processes.
     void AppendNewProcess(SharedPtr<Process> process);
 
@@ -82,7 +78,6 @@ private:
     friend class Object;
     friend class Process;
     friend class Thread;
-    friend class Timer;
 
     /// Creates a new object ID, incrementing the internal object ID counter.
     u32 CreateNewObjectID();
@@ -93,15 +88,9 @@ private:
     /// Creates a new thread ID, incrementing the internal thread ID counter.
     u64 CreateNewThreadID();
 
-    /// Creates a timer callback handle for the given timer.
-    ResultVal<Handle> CreateTimerCallbackHandle(const SharedPtr<Timer>& timer);
-
     /// Retrieves the event type used for thread wakeup callbacks.
     CoreTiming::EventType* ThreadWakeupCallbackEventType() const;
 
-    /// Retrieves the event type used for timer callbacks.
-    CoreTiming::EventType* TimerCallbackEventType() const;
-
     /// Provides a reference to the thread wakeup callback handle table.
     Kernel::HandleTable& ThreadWakeupCallbackHandleTable();
 

src/core/hle/kernel/object.cpp

@@ -16,7 +16,6 @@ bool Object::IsWaitable() const {
     case HandleType::ReadableEvent:
     case HandleType::Thread:
     case HandleType::Process:
-    case HandleType::Timer:
     case HandleType::ServerPort:
     case HandleType::ServerSession:
         return true;

src/core/hle/kernel/object.h

@@ -25,7 +25,6 @@ enum class HandleType : u32 {
     Thread,
     Process,
     AddressArbiter,
-    Timer,
     ResourceLimit,
     ClientPort,
     ServerPort,

src/core/hle/kernel/readable_event.cpp

@@ -44,8 +44,4 @@ ResultCode ReadableEvent::Reset() {
     return RESULT_SUCCESS;
 }
 
-void ReadableEvent::WakeupAllWaitingThreads() {
-    WaitObject::WakeupAllWaitingThreads();
-}
-
 } // namespace Kernel

src/core/hle/kernel/readable_event.h

@@ -39,8 +39,6 @@ public:
     bool ShouldWait(Thread* thread) const override;
     void Acquire(Thread* thread) override;
 
-    void WakeupAllWaitingThreads() override;
-
     /// Unconditionally clears the readable event's state.
     void Clear();
 

src/core/hle/kernel/timer.cpp

@@ -1,88 +0,0 @@
-// Copyright 2015 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include "common/assert.h"
-#include "common/logging/log.h"
-#include "core/core.h"
-#include "core/core_timing.h"
-#include "core/core_timing_util.h"
-#include "core/hle/kernel/handle_table.h"
-#include "core/hle/kernel/kernel.h"
-#include "core/hle/kernel/object.h"
-#include "core/hle/kernel/thread.h"
-#include "core/hle/kernel/timer.h"
-
-namespace Kernel {
-
-Timer::Timer(KernelCore& kernel) : WaitObject{kernel} {}
-Timer::~Timer() = default;
-
-SharedPtr<Timer> Timer::Create(KernelCore& kernel, ResetType reset_type, std::string name) {
-    SharedPtr<Timer> timer(new Timer(kernel));
-
-    timer->reset_type = reset_type;
-    timer->signaled = false;
-    timer->name = std::move(name);
-    timer->initial_delay = 0;
-    timer->interval_delay = 0;
-    timer->callback_handle = kernel.CreateTimerCallbackHandle(timer).Unwrap();
-
-    return timer;
-}
-
-bool Timer::ShouldWait(Thread* thread) const {
-    return !signaled;
-}
-
-void Timer::Acquire(Thread* thread) {
-    ASSERT_MSG(!ShouldWait(thread), "object unavailable!");
-
-    if (reset_type == ResetType::OneShot)
-        signaled = false;
-}
-
-void Timer::Set(s64 initial, s64 interval) {
-    // Ensure we get rid of any previous scheduled event
-    Cancel();
-
-    initial_delay = initial;
-    interval_delay = interval;
-
-    if (initial == 0) {
-        // Immediately invoke the callback
-        Signal(0);
-    } else {
-        CoreTiming::ScheduleEvent(CoreTiming::nsToCycles(initial), kernel.TimerCallbackEventType(),
-                                  callback_handle);
-    }
-}
-
-void Timer::Cancel() {
-    CoreTiming::UnscheduleEvent(kernel.TimerCallbackEventType(), callback_handle);
-}
-
-void Timer::Clear() {
-    signaled = false;
-}
-
-void Timer::WakeupAllWaitingThreads() {
-    WaitObject::WakeupAllWaitingThreads();
-}
-
-void Timer::Signal(int cycles_late) {
-    LOG_TRACE(Kernel, "Timer {} fired", GetObjectId());
-
-    signaled = true;
-
-    // Resume all waiting threads
-    WakeupAllWaitingThreads();
-
-    if (interval_delay != 0) {
-        // Reschedule the timer with the interval delay
-        CoreTiming::ScheduleEvent(CoreTiming::nsToCycles(interval_delay) - cycles_late,
-                                  kernel.TimerCallbackEventType(), callback_handle);
-    }
-}
-
-} // namespace Kernel

src/core/hle/kernel/timer.h

@@ -1,90 +0,0 @@
-// Copyright 2015 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include "common/common_types.h"
-#include "core/hle/kernel/object.h"
-#include "core/hle/kernel/wait_object.h"
-
-namespace Kernel {
-
-class KernelCore;
-
-class Timer final : public WaitObject {
-public:
-    /**
-     * Creates a timer
-     * @param kernel The kernel instance to create the timer callback handle for.
-     * @param reset_type ResetType describing how to create the timer
-     * @param name Optional name of timer
-     * @return The created Timer
-     */
-    static SharedPtr<Timer> Create(KernelCore& kernel, ResetType reset_type,
-                                   std::string name = "Unknown");
-
-    std::string GetTypeName() const override {
-        return "Timer";
-    }
-    std::string GetName() const override {
-        return name;
-    }
-
-    static const HandleType HANDLE_TYPE = HandleType::Timer;
-    HandleType GetHandleType() const override {
-        return HANDLE_TYPE;
-    }
-
-    ResetType GetResetType() const {
-        return reset_type;
-    }
-
-    u64 GetInitialDelay() const {
-        return initial_delay;
-    }
-
-    u64 GetIntervalDelay() const {
-        return interval_delay;
-    }
-
-    bool ShouldWait(Thread* thread) const override;
-    void Acquire(Thread* thread) override;
-
-    void WakeupAllWaitingThreads() override;
-
-    /**
-     * Starts the timer, with the specified initial delay and interval.
-     * @param initial Delay until the timer is first fired
-     * @param interval Delay until the timer is fired after the first time
-     */
-    void Set(s64 initial, s64 interval);
-
-    void Cancel();
-    void Clear();
-
-    /**
-     * Signals the timer, waking up any waiting threads and rescheduling it
-     * for the next interval.
-     * This method should not be called from outside the timer callback handler,
-     * lest multiple callback events get scheduled.
-     */
-    void Signal(int cycles_late);
-
-private:
-    explicit Timer(KernelCore& kernel);
-    ~Timer() override;
-
-    ResetType reset_type; ///< The ResetType of this timer
-
-    u64 initial_delay;  ///< The delay until the timer fires for the first time
-    u64 interval_delay; ///< The delay until the timer fires after the first time
-
-    bool signaled;    ///< Whether the timer has been signaled or not
-    std::string name; ///< Name of timer (optional)
-
-    /// Handle used as userdata to reference this object when inserting into the CoreTiming queue.
-    Handle callback_handle;
-};
-
-} // namespace Kernel

src/core/hle/kernel/wait_object.h

@@ -33,19 +33,19 @@ public:
      * Add a thread to wait on this object
      * @param thread Pointer to thread to add
      */
-    virtual void AddWaitingThread(SharedPtr<Thread> thread);
+    void AddWaitingThread(SharedPtr<Thread> thread);
 
     /**
      * Removes a thread from waiting on this object (e.g. if it was resumed already)
      * @param thread Pointer to thread to remove
      */
-    virtual void RemoveWaitingThread(Thread* thread);
+    void RemoveWaitingThread(Thread* thread);
 
     /**
      * Wake up all threads waiting on this object that can be awoken, in priority order,
      * and set the synchronization result and output of the thread.
      */
-    virtual void WakeupAllWaitingThreads();
+    void WakeupAllWaitingThreads();
 
     /**
      * Wakes up a single thread waiting on this object.

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

@@ -249,7 +249,8 @@ AppletAE::AppletAE(std::shared_ptr<NVFlinger::NVFlinger> nvflinger,
         {300, nullptr, "OpenOverlayAppletProxy"},
         {350, nullptr, "OpenSystemApplicationProxy"},
         {400, nullptr, "CreateSelfLibraryAppletCreatorForDevelop"},
-        {401, nullptr, "GetSystemAppletControllerForDebug"},
+        {410, nullptr, "GetSystemAppletControllerForDebug"},
+        {1000, nullptr, "GetDebugFunctions"},
     };
     // clang-format on
 

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

@@ -12,6 +12,7 @@ namespace Service::Audio {
 class IAudioIn final : public ServiceFramework<IAudioIn> {
 public:
     IAudioIn() : ServiceFramework("IAudioIn") {
+        // clang-format off
         static const FunctionInfo functions[] = {
             {0, nullptr, "GetAudioInState"},
             {1, nullptr, "StartAudioIn"},
@@ -28,16 +29,24 @@ public:
             {12, nullptr, "SetAudioInDeviceGain"},
             {13, nullptr, "GetAudioInDeviceGain"},
         };
+        // clang-format on
+
         RegisterHandlers(functions);
     }
     ~IAudioIn() = default;
 };
 
 AudInU::AudInU() : ServiceFramework("audin:u") {
+    // clang-format off
     static const FunctionInfo functions[] = {
-        {0, nullptr, "ListAudioIns"},    {1, nullptr, "OpenAudioIn"},      {2, nullptr, "Unknown"},
-        {3, nullptr, "OpenAudioInAuto"}, {4, nullptr, "ListAudioInsAuto"},
+        {0, nullptr, "ListAudioIns"},
+        {1, nullptr, "OpenAudioIn"},
+        {2, nullptr, "Unknown"},
+        {3, nullptr, "OpenAudioInAuto"},
+        {4, nullptr, "ListAudioInsAuto"},
     };
+    // clang-format on
+
     RegisterHandlers(functions);
 }
 

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

@@ -12,6 +12,7 @@ namespace Service::Audio {
 class IFinalOutputRecorder final : public ServiceFramework<IFinalOutputRecorder> {
 public:
     IFinalOutputRecorder() : ServiceFramework("IFinalOutputRecorder") {
+        // clang-format off
         static const FunctionInfo functions[] = {
             {0, nullptr, "GetFinalOutputRecorderState"},
             {1, nullptr, "StartFinalOutputRecorder"},
@@ -20,10 +21,13 @@ public:
             {4, nullptr, "RegisterBufferEvent"},
             {5, nullptr, "GetReleasedFinalOutputRecorderBuffer"},
             {6, nullptr, "ContainsFinalOutputRecorderBuffer"},
-            {7, nullptr, "Unknown"},
+            {7, nullptr, "GetFinalOutputRecorderBufferEndTime"},
             {8, nullptr, "AppendFinalOutputRecorderBufferAuto"},
             {9, nullptr, "GetReleasedFinalOutputRecorderBufferAuto"},
+            {10, nullptr, "FlushFinalOutputRecorderBuffers"},
         };
+        // clang-format on
+
         RegisterHandlers(functions);
     }
     ~IFinalOutputRecorder() = default;

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

@@ -229,14 +229,16 @@ private:
 }; // namespace Audio
 
 AudRenU::AudRenU() : ServiceFramework("audren:u") {
+    // clang-format off
     static const FunctionInfo functions[] = {
         {0, &AudRenU::OpenAudioRenderer, "OpenAudioRenderer"},
         {1, &AudRenU::GetAudioRendererWorkBufferSize, "GetAudioRendererWorkBufferSize"},
-        {2, &AudRenU::GetAudioDevice, "GetAudioDevice"},
+        {2, &AudRenU::GetAudioDeviceService, "GetAudioDeviceService"},
         {3, nullptr, "OpenAudioRendererAuto"},
-        {4, &AudRenU::GetAudioDeviceServiceWithRevisionInfo,
-         "GetAudioDeviceServiceWithRevisionInfo"},
+        {4, &AudRenU::GetAudioDeviceServiceWithRevisionInfo, "GetAudioDeviceServiceWithRevisionInfo"},
     };
+    // clang-format on
+
     RegisterHandlers(functions);
 }
 
@@ -313,7 +315,7 @@ void AudRenU::GetAudioRendererWorkBufferSize(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_Audio, "buffer_size=0x{:X}", output_sz);
 }
 
-void AudRenU::GetAudioDevice(Kernel::HLERequestContext& ctx) {
+void AudRenU::GetAudioDeviceService(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_Audio, "called");
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};

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

@@ -20,7 +20,7 @@ public:
 private:
     void OpenAudioRenderer(Kernel::HLERequestContext& ctx);
     void GetAudioRendererWorkBufferSize(Kernel::HLERequestContext& ctx);
-    void GetAudioDevice(Kernel::HLERequestContext& ctx);
+    void GetAudioDeviceService(Kernel::HLERequestContext& ctx);
     void GetAudioDeviceServiceWithRevisionInfo(Kernel::HLERequestContext& ctx);
 
     enum class AudioFeatures : u32 {

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

@@ -5,7 +5,6 @@
 #include <chrono>
 #include <cstring>
 #include <memory>
-#include <optional>
 #include <vector>
 
 #include <opus.h>
@@ -30,48 +29,66 @@ public:
                                 u32 channel_count)
         : ServiceFramework("IHardwareOpusDecoderManager"), decoder(std::move(decoder)),
           sample_rate(sample_rate), channel_count(channel_count) {
+        // clang-format off
         static const FunctionInfo functions[] = {
-            {0, &IHardwareOpusDecoderManager::DecodeInterleaved, "DecodeInterleaved"},
+            {0, &IHardwareOpusDecoderManager::DecodeInterleavedOld, "DecodeInterleavedOld"},
             {1, nullptr, "SetContext"},
-            {2, nullptr, "DecodeInterleavedForMultiStream"},
+            {2, nullptr, "DecodeInterleavedForMultiStreamOld"},
             {3, nullptr, "SetContextForMultiStream"},
-            {4, &IHardwareOpusDecoderManager::DecodeInterleavedWithPerformance,
-             "DecodeInterleavedWithPerformance"},
-            {5, nullptr, "Unknown5"},
-            {6, nullptr, "Unknown6"},
-            {7, nullptr, "Unknown7"},
+            {4, &IHardwareOpusDecoderManager::DecodeInterleavedWithPerfOld, "DecodeInterleavedWithPerfOld"},
+            {5, nullptr, "DecodeInterleavedForMultiStreamWithPerfOld"},
+            {6, &IHardwareOpusDecoderManager::DecodeInterleaved, "DecodeInterleaved"},
+            {7, nullptr, "DecodeInterleavedForMultiStream"},
         };
+        // clang-format on
+
         RegisterHandlers(functions);
     }
 
 private:
+    /// Describes extra behavior that may be asked of the decoding context.
+    enum class ExtraBehavior {
+        /// No extra behavior.
+        None,
+
+        /// Resets the decoder context back to a freshly initialized state.
+        ResetContext,
+    };
+
+    void DecodeInterleavedOld(Kernel::HLERequestContext& ctx) {
+        LOG_DEBUG(Audio, "called");
+
+        DecodeInterleavedHelper(ctx, nullptr, ExtraBehavior::None);
+    }
+
+    void DecodeInterleavedWithPerfOld(Kernel::HLERequestContext& ctx) {
+        LOG_DEBUG(Audio, "called");
+
+        u64 performance = 0;
+        DecodeInterleavedHelper(ctx, &performance, ExtraBehavior::None);
+    }
+
     void DecodeInterleaved(Kernel::HLERequestContext& ctx) {
         LOG_DEBUG(Audio, "called");
 
-        u32 consumed = 0;
-        u32 sample_count = 0;
-        std::vector<opus_int16> samples(ctx.GetWriteBufferSize() / sizeof(opus_int16));
-        if (!Decoder_DecodeInterleaved(consumed, sample_count, ctx.ReadBuffer(), samples)) {
-            LOG_ERROR(Audio, "Failed to decode opus data");
-            IPC::ResponseBuilder rb{ctx, 2};
-            // TODO(ogniK): Use correct error code
-            rb.Push(ResultCode(-1));
-            return;
-        }
-        IPC::ResponseBuilder rb{ctx, 4};
-        rb.Push(RESULT_SUCCESS);
-        rb.Push<u32>(consumed);
-        rb.Push<u32>(sample_count);
-        ctx.WriteBuffer(samples.data(), samples.size() * sizeof(s16));
+        IPC::RequestParser rp{ctx};
+        const auto extra_behavior =
+            rp.Pop<bool>() ? ExtraBehavior::ResetContext : ExtraBehavior::None;
+
+        u64 performance = 0;
+        DecodeInterleavedHelper(ctx, &performance, extra_behavior);
     }
 
-    void DecodeInterleavedWithPerformance(Kernel::HLERequestContext& ctx) {
-        LOG_DEBUG(Audio, "called");
-
+    void DecodeInterleavedHelper(Kernel::HLERequestContext& ctx, u64* performance,
+                                 ExtraBehavior extra_behavior) {
         u32 consumed = 0;
         u32 sample_count = 0;
-        u64 performance = 0;
         std::vector<opus_int16> samples(ctx.GetWriteBufferSize() / sizeof(opus_int16));
+
+        if (extra_behavior == ExtraBehavior::ResetContext) {
+            ResetDecoderContext();
+        }
+
         if (!Decoder_DecodeInterleaved(consumed, sample_count, ctx.ReadBuffer(), samples,
                                        performance)) {
             LOG_ERROR(Audio, "Failed to decode opus data");
@@ -80,25 +97,28 @@ private:
             rb.Push(ResultCode(-1));
             return;
         }
-        IPC::ResponseBuilder rb{ctx, 6};
+
+        const u32 param_size = performance != nullptr ? 6 : 4;
+        IPC::ResponseBuilder rb{ctx, param_size};
         rb.Push(RESULT_SUCCESS);
         rb.Push<u32>(consumed);
         rb.Push<u32>(sample_count);
-        rb.Push<u64>(performance);
+        if (performance) {
+            rb.Push<u64>(*performance);
+        }
         ctx.WriteBuffer(samples.data(), samples.size() * sizeof(s16));
     }
 
-    bool Decoder_DecodeInterleaved(
-        u32& consumed, u32& sample_count, const std::vector<u8>& input,
-        std::vector<opus_int16>& output,
-        std::optional<std::reference_wrapper<u64>> performance_time = std::nullopt) {
+    bool Decoder_DecodeInterleaved(u32& consumed, u32& sample_count, const std::vector<u8>& input,
+                                   std::vector<opus_int16>& output, u64* out_performance_time) {
         const auto start_time = std::chrono::high_resolution_clock::now();
-        std::size_t raw_output_sz = output.size() * sizeof(opus_int16);
+        const std::size_t raw_output_sz = output.size() * sizeof(opus_int16);
         if (sizeof(OpusHeader) > input.size()) {
             LOG_ERROR(Audio, "Input is smaller than the header size, header_sz={}, input_sz={}",
                       sizeof(OpusHeader), input.size());
             return false;
         }
+
         OpusHeader hdr{};
         std::memcpy(&hdr, input.data(), sizeof(OpusHeader));
         if (sizeof(OpusHeader) + static_cast<u32>(hdr.sz) > input.size()) {
@@ -106,8 +126,9 @@ private:
                       sizeof(OpusHeader) + static_cast<u32>(hdr.sz), input.size());
             return false;
         }
-        auto frame = input.data() + sizeof(OpusHeader);
-        auto decoded_sample_count = opus_packet_get_nb_samples(
+
+        const auto frame = input.data() + sizeof(OpusHeader);
+        const auto decoded_sample_count = opus_packet_get_nb_samples(
             frame, static_cast<opus_int32>(input.size() - sizeof(OpusHeader)),
             static_cast<opus_int32>(sample_rate));
         if (decoded_sample_count * channel_count * sizeof(u16) > raw_output_sz) {
@@ -117,8 +138,9 @@ private:
                 decoded_sample_count * channel_count * sizeof(u16), raw_output_sz);
             return false;
         }
+
         const int frame_size = (static_cast<int>(raw_output_sz / sizeof(s16) / channel_count));
-        auto out_sample_count =
+        const auto out_sample_count =
             opus_decode(decoder.get(), frame, hdr.sz, output.data(), frame_size, 0);
         if (out_sample_count < 0) {
             LOG_ERROR(Audio,
@@ -127,16 +149,24 @@ private:
                       out_sample_count, frame_size, static_cast<u32>(hdr.sz));
             return false;
         }
+
         const auto end_time = std::chrono::high_resolution_clock::now() - start_time;
         sample_count = out_sample_count;
         consumed = static_cast<u32>(sizeof(OpusHeader) + hdr.sz);
-        if (performance_time.has_value()) {
-            performance_time->get() =
+        if (out_performance_time != nullptr) {
+            *out_performance_time =
                 std::chrono::duration_cast<std::chrono::milliseconds>(end_time).count();
         }
+
         return true;
     }
 
+    void ResetDecoderContext() {
+        ASSERT(decoder != nullptr);
+
+        opus_decoder_ctl(decoder.get(), OPUS_RESET_STATE);
+    }
+
     struct OpusHeader {
         u32_be sz; // Needs to be BE for some odd reason
         INSERT_PADDING_WORDS(1);
@@ -157,6 +187,7 @@ void HwOpus::GetWorkBufferSize(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
     const auto sample_rate = rp.Pop<u32>();
     const auto channel_count = rp.Pop<u32>();
+
     LOG_DEBUG(Audio, "called with sample_rate={}, channel_count={}", sample_rate, channel_count);
 
     ASSERT_MSG(sample_rate == 48000 || sample_rate == 24000 || sample_rate == 16000 ||
@@ -174,9 +205,10 @@ void HwOpus::GetWorkBufferSize(Kernel::HLERequestContext& ctx) {
 
 void HwOpus::OpenOpusDecoder(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
-    auto sample_rate = rp.Pop<u32>();
-    auto channel_count = rp.Pop<u32>();
-    auto buffer_sz = rp.Pop<u32>();
+    const auto sample_rate = rp.Pop<u32>();
+    const auto channel_count = rp.Pop<u32>();
+    const auto buffer_sz = rp.Pop<u32>();
+
     LOG_DEBUG(Audio, "called sample_rate={}, channel_count={}, buffer_size={}", sample_rate,
               channel_count, buffer_sz);
 
@@ -185,8 +217,9 @@ void HwOpus::OpenOpusDecoder(Kernel::HLERequestContext& ctx) {
                "Invalid sample rate");
     ASSERT_MSG(channel_count == 1 || channel_count == 2, "Invalid channel count");
 
-    std::size_t worker_sz = WorkerBufferSize(channel_count);
+    const std::size_t worker_sz = WorkerBufferSize(channel_count);
     ASSERT_MSG(buffer_sz >= worker_sz, "Worker buffer too large");
+
     std::unique_ptr<OpusDecoder, OpusDeleter> decoder{
         static_cast<OpusDecoder*>(operator new(worker_sz))};
     if (const int err = opus_decoder_init(decoder.get(), sample_rate, channel_count)) {

src/core/hle/service/btdrv/btdrv.cpp

@@ -19,16 +19,16 @@ public:
     explicit Bt() : ServiceFramework{"bt"} {
         // clang-format off
         static const FunctionInfo functions[] = {
-            {0, nullptr, "Unknown0"},
-            {1, nullptr, "Unknown1"},
-            {2, nullptr, "Unknown2"},
-            {3, nullptr, "Unknown3"},
-            {4, nullptr, "Unknown4"},
-            {5, nullptr, "Unknown5"},
-            {6, nullptr, "Unknown6"},
-            {7, nullptr, "Unknown7"},
-            {8, nullptr, "Unknown8"},
-            {9, &Bt::RegisterEvent, "RegisterEvent"},
+            {0, nullptr, "LeClientReadCharacteristic"},
+            {1, nullptr, "LeClientReadDescriptor"},
+            {2, nullptr, "LeClientWriteCharacteristic"},
+            {3, nullptr, "LeClientWriteDescriptor"},
+            {4, nullptr, "LeClientRegisterNotification"},
+            {5, nullptr, "LeClientDeregisterNotification"},
+            {6, nullptr, "SetLeResponse"},
+            {7, nullptr, "LeSendIndication"},
+            {8, nullptr, "GetLeEventInfo"},
+            {9, &Bt::RegisterBleEvent, "RegisterBleEvent"},
         };
         // clang-format on
         RegisterHandlers(functions);
@@ -39,7 +39,7 @@ public:
     }
 
 private:
-    void RegisterEvent(Kernel::HLERequestContext& ctx) {
+    void RegisterBleEvent(Kernel::HLERequestContext& ctx) {
         LOG_WARNING(Service_BTM, "(STUBBED) called");
 
         IPC::ResponseBuilder rb{ctx, 2, 1};
@@ -55,11 +55,11 @@ public:
     explicit BtDrv() : ServiceFramework{"btdrv"} {
         // clang-format off
         static const FunctionInfo functions[] = {
-            {0, nullptr, "Unknown"},
-            {1, nullptr, "Init"},
-            {2, nullptr, "Enable"},
-            {3, nullptr, "Disable"},
-            {4, nullptr, "CleanupAndShutdown"},
+            {0, nullptr, "InitializeBluetoothDriver"},
+            {1, nullptr, "InitializeBluetooth"},
+            {2, nullptr, "EnableBluetooth"},
+            {3, nullptr, "DisableBluetooth"},
+            {4, nullptr, "CleanupBluetooth"},
             {5, nullptr, "GetAdapterProperties"},
             {6, nullptr, "GetAdapterProperty"},
             {7, nullptr, "SetAdapterProperty"},
@@ -70,36 +70,91 @@ public:
             {12, nullptr, "CancelBond"},
             {13, nullptr, "PinReply"},
             {14, nullptr, "SspReply"},
-            {15, nullptr, "Unknown2"},
-            {16, nullptr, "InitInterfaces"},
-            {17, nullptr, "HidHostInterface_Connect"},
-            {18, nullptr, "HidHostInterface_Disconnect"},
-            {19, nullptr, "HidHostInterface_SendData"},
-            {20, nullptr, "HidHostInterface_SendData2"},
-            {21, nullptr, "HidHostInterface_SetReport"},
-            {22, nullptr, "HidHostInterface_GetReport"},
-            {23, nullptr, "HidHostInterface_WakeController"},
-            {24, nullptr, "HidHostInterface_AddPairedDevice"},
-            {25, nullptr, "HidHostInterface_GetPairedDevice"},
-            {26, nullptr, "HidHostInterface_CleanupAndShutdown"},
-            {27, nullptr, "Unknown3"},
-            {28, nullptr, "ExtInterface_SetTSI"},
-            {29, nullptr, "ExtInterface_SetBurstMode"},
-            {30, nullptr, "ExtInterface_SetZeroRetran"},
-            {31, nullptr, "ExtInterface_SetMcMode"},
-            {32, nullptr, "ExtInterface_StartLlrMode"},
-            {33, nullptr, "ExtInterface_ExitLlrMode"},
-            {34, nullptr, "ExtInterface_SetRadio"},
-            {35, nullptr, "ExtInterface_SetVisibility"},
-            {36, nullptr, "Unknown4"},
-            {37, nullptr, "Unknown5"},
-            {38, nullptr, "HidHostInterface_GetLatestPlr"},
-            {39, nullptr, "ExtInterface_GetPendingConnections"},
-            {40, nullptr, "HidHostInterface_GetChannelMap"},
-            {41, nullptr, "SetIsBluetoothBoostEnabled"},
-            {42, nullptr, "GetIsBluetoothBoostEnabled"},
-            {43, nullptr, "SetIsBluetoothAfhEnabled"},
-            {44, nullptr, "GetIsBluetoothAfhEnabled"},
+            {15, nullptr, "GetEventInfo"},
+            {16, nullptr, "InitializeHid"},
+            {17, nullptr, "HidConnect"},
+            {18, nullptr, "HidDisconnect"},
+            {19, nullptr, "HidSendData"},
+            {20, nullptr, "HidSendData2"},
+            {21, nullptr, "HidSetReport"},
+            {22, nullptr, "HidGetReport"},
+            {23, nullptr, "HidWakeController"},
+            {24, nullptr, "HidAddPairedDevice"},
+            {25, nullptr, "HidGetPairedDevice"},
+            {26, nullptr, "CleanupHid"},
+            {27, nullptr, "HidGetEventInfo"},
+            {28, nullptr, "ExtSetTsi"},
+            {29, nullptr, "ExtSetBurstMode"},
+            {30, nullptr, "ExtSetZeroRetran"},
+            {31, nullptr, "ExtSetMcMode"},
+            {32, nullptr, "ExtStartLlrMode"},
+            {33, nullptr, "ExtExitLlrMode"},
+            {34, nullptr, "ExtSetRadio"},
+            {35, nullptr, "ExtSetVisibility"},
+            {36, nullptr, "ExtSetTbfcScan"},
+            {37, nullptr, "RegisterHidReportEvent"},
+            {38, nullptr, "HidGetReportEventInfo"},
+            {39, nullptr, "GetLatestPlr"},
+            {40, nullptr, "ExtGetPendingConnections"},
+            {41, nullptr, "GetChannelMap"},
+            {42, nullptr, "EnableBluetoothBoostSetting"},
+            {43, nullptr, "IsBluetoothBoostSettingEnabled"},
+            {44, nullptr, "EnableBluetoothAfhSetting"},
+            {45, nullptr, "IsBluetoothAfhSettingEnabled"},
+            {46, nullptr, "InitializeBluetoothLe"},
+            {47, nullptr, "EnableBluetoothLe"},
+            {48, nullptr, "DisableBluetoothLe"},
+            {49, nullptr, "CleanupBluetoothLe"},
+            {50, nullptr, "SetLeVisibility"},
+            {51, nullptr, "SetLeConnectionParameter"},
+            {52, nullptr, "SetLeDefaultConnectionParameter"},
+            {53, nullptr, "SetLeAdvertiseData"},
+            {54, nullptr, "SetLeAdvertiseParameter"},
+            {55, nullptr, "StartLeScan"},
+            {56, nullptr, "StopLeScan"},
+            {57, nullptr, "AddLeScanFilterCondition"},
+            {58, nullptr, "DeleteLeScanFilterCondition"},
+            {59, nullptr, "DeleteLeScanFilter"},
+            {60, nullptr, "ClearLeScanFilters"},
+            {61, nullptr, "EnableLeScanFilter"},
+            {62, nullptr, "RegisterLeClient"},
+            {63, nullptr, "UnregisterLeClient"},
+            {64, nullptr, "UnregisterLeClientAll"},
+            {65, nullptr, "LeClientConnect"},
+            {66, nullptr, "LeClientCancelConnection"},
+            {67, nullptr, "LeClientDisconnect"},
+            {68, nullptr, "LeClientGetAttributes"},
+            {69, nullptr, "LeClientDiscoverService"},
+            {70, nullptr, "LeClientConfigureMtu"},
+            {71, nullptr, "RegisterLeServer"},
+            {72, nullptr, "UnregisterLeServer"},
+            {73, nullptr, "LeServerConnect"},
+            {74, nullptr, "LeServerDisconnect"},
+            {75, nullptr, "CreateLeService"},
+            {76, nullptr, "StartLeService"},
+            {77, nullptr, "AddLeCharacteristic"},
+            {78, nullptr, "AddLeDescriptor"},
+            {79, nullptr, "GetLeCoreEventInfo"},
+            {80, nullptr, "LeGetFirstCharacteristic"},
+            {81, nullptr, "LeGetNextCharacteristic"},
+            {82, nullptr, "LeGetFirstDescriptor"},
+            {83, nullptr, "LeGetNextDescriptor"},
+            {84, nullptr, "RegisterLeCoreDataPath"},
+            {85, nullptr, "UnregisterLeCoreDataPath"},
+            {86, nullptr, "RegisterLeHidDataPath"},
+            {87, nullptr, "UnregisterLeHidDataPath"},
+            {88, nullptr, "RegisterLeDataPath"},
+            {89, nullptr, "UnregisterLeDataPath"},
+            {90, nullptr, "LeClientReadCharacteristic"},
+            {91, nullptr, "LeClientReadDescriptor"},
+            {92, nullptr, "LeClientWriteCharacteristic"},
+            {93, nullptr, "LeClientWriteDescriptor"},
+            {94, nullptr, "LeClientRegisterNotification"},
+            {95, nullptr, "LeClientDeregisterNotification"},
+            {96, nullptr, "GetLeHidEventInfo"},
+            {97, nullptr, "RegisterBleHidEvent"},
+            {98, nullptr, "SetLeScanParameter"},
+            {256, nullptr, "GetIsManufacturingMode"}
         };
         // clang-format on
 

src/core/hle/service/btm/btm.cpp

@@ -20,38 +20,38 @@ public:
     explicit IBtmUserCore() : ServiceFramework{"IBtmUserCore"} {
         // clang-format off
         static const FunctionInfo functions[] = {
-            {0, &IBtmUserCore::GetScanEvent, "GetScanEvent"},
-            {1, nullptr, "Unknown1"},
-            {2, nullptr, "Unknown2"},
-            {3, nullptr, "Unknown3"},
-            {4, nullptr, "Unknown4"},
-            {5, nullptr, "Unknown5"},
-            {6, nullptr, "Unknown6"},
-            {7, nullptr, "Unknown7"},
-            {8, nullptr, "Unknown8"},
-            {9, nullptr, "Unknown9"},
-            {10, nullptr, "Unknown10"},
-            {17, &IBtmUserCore::GetConnectionEvent, "GetConnectionEvent"},
-            {18, nullptr, "Unknown18"},
-            {19, nullptr, "Unknown19"},
-            {20, nullptr, "Unknown20"},
-            {21, nullptr, "Unknown21"},
-            {22, nullptr, "Unknown22"},
-            {23, nullptr, "Unknown23"},
-            {24, nullptr, "Unknown24"},
-            {25, nullptr, "Unknown25"},
-            {26, &IBtmUserCore::GetDiscoveryEvent, "AcquireBleServiceDiscoveryEventImpl"},
-            {27, nullptr, "Unknown27"},
-            {28, nullptr, "Unknown28"},
-            {29, nullptr, "Unknown29"},
-            {30, nullptr, "Unknown30"},
-            {31, nullptr, "Unknown31"},
-            {32, nullptr, "Unknown32"},
-            {33, &IBtmUserCore::GetConfigEvent, "GetConfigEvent"},
-            {34, nullptr, "Unknown34"},
-            {35, nullptr, "Unknown35"},
-            {36, nullptr, "Unknown36"},
-            {37, nullptr, "Unknown37"},
+            {0, &IBtmUserCore::AcquireBleScanEvent, "AcquireBleScanEvent"},
+            {1, nullptr, "GetBleScanFilterParameter"},
+            {2, nullptr, "GetBleScanFilterParameter2"},
+            {3, nullptr, "StartBleScanForGeneral"},
+            {4, nullptr, "StopBleScanForGeneral"},
+            {5, nullptr, "GetBleScanResultsForGeneral"},
+            {6, nullptr, "StartBleScanForPaired"},
+            {7, nullptr, "StopBleScanForPaired"},
+            {8, nullptr, "StartBleScanForSmartDevice"},
+            {9, nullptr, "StopBleScanForSmartDevice"},
+            {10, nullptr, "GetBleScanResultsForSmartDevice"},
+            {17, &IBtmUserCore::AcquireBleConnectionEvent, "AcquireBleConnectionEvent"},
+            {18, nullptr, "BleConnect"},
+            {19, nullptr, "BleDisconnect"},
+            {20, nullptr, "BleGetConnectionState"},
+            {21, nullptr, "AcquireBlePairingEvent"},
+            {22, nullptr, "BlePairDevice"},
+            {23, nullptr, "BleUnPairDevice"},
+            {24, nullptr, "BleUnPairDevice2"},
+            {25, nullptr, "BleGetPairedDevices"},
+            {26, &IBtmUserCore::AcquireBleServiceDiscoveryEvent, "AcquireBleServiceDiscoveryEvent"},
+            {27, nullptr, "GetGattServices"},
+            {28, nullptr, "GetGattService"},
+            {29, nullptr, "GetGattIncludedServices"},
+            {30, nullptr, "GetBelongingGattService"},
+            {31, nullptr, "GetGattCharacteristics"},
+            {32, nullptr, "GetGattDescriptors"},
+            {33, &IBtmUserCore::AcquireBleMtuConfigEvent, "AcquireBleMtuConfigEvent"},
+            {34, nullptr, "ConfigureBleMtu"},
+            {35, nullptr, "GetBleMtu"},
+            {36, nullptr, "RegisterBleGattDataPath"},
+            {37, nullptr, "UnregisterBleGattDataPath"},
         };
         // clang-format on
         RegisterHandlers(functions);
@@ -68,7 +68,7 @@ public:
     }
 
 private:
-    void GetScanEvent(Kernel::HLERequestContext& ctx) {
+    void AcquireBleScanEvent(Kernel::HLERequestContext& ctx) {
         LOG_WARNING(Service_BTM, "(STUBBED) called");
 
         IPC::ResponseBuilder rb{ctx, 2, 1};
@@ -76,7 +76,7 @@ private:
         rb.PushCopyObjects(scan_event.readable);
     }
 
-    void GetConnectionEvent(Kernel::HLERequestContext& ctx) {
+    void AcquireBleConnectionEvent(Kernel::HLERequestContext& ctx) {
         LOG_WARNING(Service_BTM, "(STUBBED) called");
 
         IPC::ResponseBuilder rb{ctx, 2, 1};
@@ -84,7 +84,7 @@ private:
         rb.PushCopyObjects(connection_event.readable);
     }
 
-    void GetDiscoveryEvent(Kernel::HLERequestContext& ctx) {
+    void AcquireBleServiceDiscoveryEvent(Kernel::HLERequestContext& ctx) {
         LOG_WARNING(Service_BTM, "(STUBBED) called");
 
         IPC::ResponseBuilder rb{ctx, 2, 1};
@@ -92,7 +92,7 @@ private:
         rb.PushCopyObjects(service_discovery.readable);
     }
 
-    void GetConfigEvent(Kernel::HLERequestContext& ctx) {
+    void AcquireBleMtuConfigEvent(Kernel::HLERequestContext& ctx) {
         LOG_WARNING(Service_BTM, "(STUBBED) called");
 
         IPC::ResponseBuilder rb{ctx, 2, 1};
@@ -111,14 +111,14 @@ public:
     explicit BTM_USR() : ServiceFramework{"btm:u"} {
         // clang-format off
         static const FunctionInfo functions[] = {
-            {0, &BTM_USR::GetCoreImpl, "GetCoreImpl"},
+            {0, &BTM_USR::GetCore, "GetCore"},
         };
         // clang-format on
         RegisterHandlers(functions);
     }
 
 private:
-    void GetCoreImpl(Kernel::HLERequestContext& ctx) {
+    void GetCore(Kernel::HLERequestContext& ctx) {
         LOG_DEBUG(Service_BTM, "called");
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
@@ -134,26 +134,64 @@ public:
         static const FunctionInfo functions[] = {
             {0, nullptr, "Unknown1"},
             {1, nullptr, "Unknown2"},
-            {2, nullptr, "RegisterSystemEventForConnectedDeviceConditionImpl"},
+            {2, nullptr, "RegisterSystemEventForConnectedDeviceCondition"},
             {3, nullptr, "Unknown3"},
             {4, nullptr, "Unknown4"},
             {5, nullptr, "Unknown5"},
             {6, nullptr, "Unknown6"},
             {7, nullptr, "Unknown7"},
-            {8, nullptr, "RegisterSystemEventForRegisteredDeviceInfoImpl"},
+            {8, nullptr, "RegisterSystemEventForRegisteredDeviceInfo"},
             {9, nullptr, "Unknown8"},
             {10, nullptr, "Unknown9"},
             {11, nullptr, "Unknown10"},
             {12, nullptr, "Unknown11"},
             {13, nullptr, "Unknown12"},
-            {14, nullptr, "EnableRadioImpl"},
-            {15, nullptr, "DisableRadioImpl"},
+            {14, nullptr, "EnableRadio"},
+            {15, nullptr, "DisableRadio"},
             {16, nullptr, "Unknown13"},
             {17, nullptr, "Unknown14"},
             {18, nullptr, "Unknown15"},
             {19, nullptr, "Unknown16"},
             {20, nullptr, "Unknown17"},
             {21, nullptr, "Unknown18"},
+            {22, nullptr, "Unknown19"},
+            {23, nullptr, "Unknown20"},
+            {24, nullptr, "Unknown21"},
+            {25, nullptr, "Unknown22"},
+            {26, nullptr, "Unknown23"},
+            {27, nullptr, "Unknown24"},
+            {28, nullptr, "Unknown25"},
+            {29, nullptr, "Unknown26"},
+            {30, nullptr, "Unknown27"},
+            {31, nullptr, "Unknown28"},
+            {32, nullptr, "Unknown29"},
+            {33, nullptr, "Unknown30"},
+            {34, nullptr, "Unknown31"},
+            {35, nullptr, "Unknown32"},
+            {36, nullptr, "Unknown33"},
+            {37, nullptr, "Unknown34"},
+            {38, nullptr, "Unknown35"},
+            {39, nullptr, "Unknown36"},
+            {40, nullptr, "Unknown37"},
+            {41, nullptr, "Unknown38"},
+            {42, nullptr, "Unknown39"},
+            {43, nullptr, "Unknown40"},
+            {44, nullptr, "Unknown41"},
+            {45, nullptr, "Unknown42"},
+            {46, nullptr, "Unknown43"},
+            {47, nullptr, "Unknown44"},
+            {48, nullptr, "Unknown45"},
+            {49, nullptr, "Unknown46"},
+            {50, nullptr, "Unknown47"},
+            {51, nullptr, "Unknown48"},
+            {52, nullptr, "Unknown49"},
+            {53, nullptr, "Unknown50"},
+            {54, nullptr, "Unknown51"},
+            {55, nullptr, "Unknown52"},
+            {56, nullptr, "Unknown53"},
+            {57, nullptr, "Unknown54"},
+            {58, nullptr, "Unknown55"},
+            {59, nullptr, "Unknown56"},
         };
         // clang-format on
 
@@ -166,7 +204,7 @@ public:
     explicit BTM_DBG() : ServiceFramework{"btm:dbg"} {
         // clang-format off
         static const FunctionInfo functions[] = {
-            {0, nullptr, "RegisterSystemEventForDiscoveryImpl"},
+            {0, nullptr, "RegisterSystemEventForDiscovery"},
             {1, nullptr, "Unknown1"},
             {2, nullptr, "Unknown2"},
             {3, nullptr, "Unknown3"},
@@ -175,6 +213,10 @@ public:
             {6, nullptr, "Unknown6"},
             {7, nullptr, "Unknown7"},
             {8, nullptr, "Unknown8"},
+            {9, nullptr, "Unknown9"},
+            {10, nullptr, "Unknown10"},
+            {11, nullptr, "Unknown11"},
+            {12, nullptr, "Unknown11"},
         };
         // clang-format on
 
@@ -187,16 +229,16 @@ public:
     explicit IBtmSystemCore() : ServiceFramework{"IBtmSystemCore"} {
         // clang-format off
         static const FunctionInfo functions[] = {
-            {0, nullptr, "StartGamepadPairingImpl"},
-            {1, nullptr, "CancelGamepadPairingImpl"},
-            {2, nullptr, "ClearGamepadPairingDatabaseImpl"},
-            {3, nullptr, "GetPairedGamepadCountImpl"},
-            {4, nullptr, "EnableRadioImpl"},
-            {5, nullptr, "DisableRadioImpl"},
-            {6, nullptr, "GetRadioOnOffImpl"},
-            {7, nullptr, "AcquireRadioEventImpl"},
-            {8, nullptr, "AcquireGamepadPairingEventImpl"},
-            {9, nullptr, "IsGamepadPairingStartedImpl"},
+            {0, nullptr, "StartGamepadPairing"},
+            {1, nullptr, "CancelGamepadPairing"},
+            {2, nullptr, "ClearGamepadPairingDatabase"},
+            {3, nullptr, "GetPairedGamepadCount"},
+            {4, nullptr, "EnableRadio"},
+            {5, nullptr, "DisableRadio"},
+            {6, nullptr, "GetRadioOnOff"},
+            {7, nullptr, "AcquireRadioEvent"},
+            {8, nullptr, "AcquireGamepadPairingEvent"},
+            {9, nullptr, "IsGamepadPairingStarted"},
         };
         // clang-format on
 
@@ -209,7 +251,7 @@ public:
     explicit BTM_SYS() : ServiceFramework{"btm:sys"} {
         // clang-format off
         static const FunctionInfo functions[] = {
-            {0, &BTM_SYS::GetCoreImpl, "GetCoreImpl"},
+            {0, &BTM_SYS::GetCore, "GetCore"},
         };
         // clang-format on
 
@@ -217,7 +259,7 @@ public:
     }
 
 private:
-    void GetCoreImpl(Kernel::HLERequestContext& ctx) {
+    void GetCore(Kernel::HLERequestContext& ctx) {
         LOG_DEBUG(Service_BTM, "called");
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};

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

@@ -627,8 +627,8 @@ private:
 FSP_SRV::FSP_SRV() : ServiceFramework("fsp-srv") {
     // clang-format off
     static const FunctionInfo functions[] = {
-        {0, nullptr, "MountContent"},
-        {1, &FSP_SRV::Initialize, "Initialize"},
+        {0, nullptr, "OpenFileSystem"},
+        {1, &FSP_SRV::SetCurrentProcess, "SetCurrentProcess"},
         {2, nullptr, "OpenDataFileSystemByCurrentProcess"},
         {7, &FSP_SRV::OpenFileSystemWithPatch, "OpenFileSystemWithPatch"},
         {8, nullptr, "OpenFileSystemWithId"},
@@ -637,10 +637,10 @@ FSP_SRV::FSP_SRV() : ServiceFramework("fsp-srv") {
         {12, nullptr, "OpenBisStorage"},
         {13, nullptr, "InvalidateBisCache"},
         {17, nullptr, "OpenHostFileSystem"},
-        {18, &FSP_SRV::MountSdCard, "MountSdCard"},
+        {18, &FSP_SRV::OpenSdCardFileSystem, "OpenSdCardFileSystem"},
         {19, nullptr, "FormatSdCardFileSystem"},
         {21, nullptr, "DeleteSaveDataFileSystem"},
-        {22, &FSP_SRV::CreateSaveData, "CreateSaveData"},
+        {22, &FSP_SRV::CreateSaveDataFileSystem, "CreateSaveDataFileSystem"},
         {23, nullptr, "CreateSaveDataFileSystemBySystemSaveDataId"},
         {24, nullptr, "RegisterSaveDataFileSystemAtomicDeletion"},
         {25, nullptr, "DeleteSaveDataFileSystemBySaveDataSpaceId"},
@@ -652,7 +652,8 @@ FSP_SRV::FSP_SRV() : ServiceFramework("fsp-srv") {
         {32, nullptr, "ExtendSaveDataFileSystem"},
         {33, nullptr, "DeleteCacheStorage"},
         {34, nullptr, "GetCacheStorageSize"},
-        {51, &FSP_SRV::MountSaveData, "MountSaveData"},
+        {35, nullptr, "CreateSaveDataFileSystemByHashSalt"},
+        {51, &FSP_SRV::OpenSaveDataFileSystem, "OpenSaveDataFileSystem"},
         {52, nullptr, "OpenSaveDataFileSystemBySystemSaveDataId"},
         {53, &FSP_SRV::OpenReadOnlySaveDataFileSystem, "OpenReadOnlySaveDataFileSystem"},
         {57, nullptr, "ReadSaveDataFileSystemExtraDataBySaveDataSpaceId"},
@@ -664,21 +665,26 @@ FSP_SRV::FSP_SRV() : ServiceFramework("fsp-srv") {
         {64, nullptr, "OpenSaveDataInternalStorageFileSystem"},
         {65, nullptr, "UpdateSaveDataMacForDebug"},
         {66, nullptr, "WriteSaveDataFileSystemExtraData2"},
+        {67, nullptr, "FindSaveDataWithFilter"},
+        {68, nullptr, "OpenSaveDataInfoReaderBySaveDataFilter"},
         {80, nullptr, "OpenSaveDataMetaFile"},
         {81, nullptr, "OpenSaveDataTransferManager"},
         {82, nullptr, "OpenSaveDataTransferManagerVersion2"},
         {83, nullptr, "OpenSaveDataTransferProhibiterForCloudBackUp"},
+        {84, nullptr, "ListApplicationAccessibleSaveDataOwnerId"},
         {100, nullptr, "OpenImageDirectoryFileSystem"},
         {110, nullptr, "OpenContentStorageFileSystem"},
+        {120, nullptr, "OpenCloudBackupWorkStorageFileSystem"},
         {200, &FSP_SRV::OpenDataStorageByCurrentProcess, "OpenDataStorageByCurrentProcess"},
         {201, nullptr, "OpenDataStorageByProgramId"},
         {202, &FSP_SRV::OpenDataStorageByDataId, "OpenDataStorageByDataId"},
-        {203, &FSP_SRV::OpenRomStorage, "OpenRomStorage"},
+        {203, &FSP_SRV::OpenPatchDataStorageByCurrentProcess, "OpenPatchDataStorageByCurrentProcess"},
         {400, nullptr, "OpenDeviceOperator"},
         {500, nullptr, "OpenSdCardDetectionEventNotifier"},
         {501, nullptr, "OpenGameCardDetectionEventNotifier"},
         {510, nullptr, "OpenSystemDataUpdateEventNotifier"},
         {511, nullptr, "NotifySystemDataUpdateEvent"},
+        {520, nullptr, "SimulateGameCardDetectionEvent"},
         {600, nullptr, "SetCurrentPosixTime"},
         {601, nullptr, "QuerySaveDataTotalSize"},
         {602, nullptr, "VerifySaveDataFileSystem"},
@@ -717,6 +723,8 @@ FSP_SRV::FSP_SRV() : ServiceFramework("fsp-srv") {
         {1008, nullptr, "OpenRegisteredUpdatePartition"},
         {1009, nullptr, "GetAndClearMemoryReportInfo"},
         {1100, nullptr, "OverrideSaveDataTransferTokenSignVerificationKey"},
+        {1110, nullptr, "CorruptSaveDataFileSystemBySaveDataSpaceId2"},
+        {1200, nullptr, "OpenMultiCommitManager"},
     };
     // clang-format on
     RegisterHandlers(functions);
@@ -724,7 +732,7 @@ FSP_SRV::FSP_SRV() : ServiceFramework("fsp-srv") {
 
 FSP_SRV::~FSP_SRV() = default;
 
-void FSP_SRV::Initialize(Kernel::HLERequestContext& ctx) {
+void FSP_SRV::SetCurrentProcess(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service_FS, "(STUBBED) called");
 
     IPC::ResponseBuilder rb{ctx, 2};
@@ -743,7 +751,7 @@ void FSP_SRV::OpenFileSystemWithPatch(Kernel::HLERequestContext& ctx) {
     rb.Push(ResultCode(-1));
 }
 
-void FSP_SRV::MountSdCard(Kernel::HLERequestContext& ctx) {
+void FSP_SRV::OpenSdCardFileSystem(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_FS, "called");
 
     IFileSystem filesystem(OpenSDMC().Unwrap());
@@ -753,7 +761,7 @@ void FSP_SRV::MountSdCard(Kernel::HLERequestContext& ctx) {
     rb.PushIpcInterface<IFileSystem>(std::move(filesystem));
 }
 
-void FSP_SRV::CreateSaveData(Kernel::HLERequestContext& ctx) {
+void FSP_SRV::CreateSaveDataFileSystem(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
 
     auto save_struct = rp.PopRaw<FileSys::SaveDataDescriptor>();
@@ -767,7 +775,7 @@ void FSP_SRV::CreateSaveData(Kernel::HLERequestContext& ctx) {
     rb.Push(RESULT_SUCCESS);
 }
 
-void FSP_SRV::MountSaveData(Kernel::HLERequestContext& ctx) {
+void FSP_SRV::OpenSaveDataFileSystem(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
 
     auto space_id = rp.PopRaw<FileSys::SaveDataSpaceId>();
@@ -793,7 +801,7 @@ void FSP_SRV::MountSaveData(Kernel::HLERequestContext& ctx) {
 
 void FSP_SRV::OpenReadOnlySaveDataFileSystem(Kernel::HLERequestContext& ctx) {
     LOG_WARNING(Service_FS, "(STUBBED) called, delegating to 51 OpenSaveDataFilesystem");
-    MountSaveData(ctx);
+    OpenSaveDataFileSystem(ctx);
 }
 
 void FSP_SRV::OpenSaveDataInfoReaderBySaveDataSpaceId(Kernel::HLERequestContext& ctx) {
@@ -881,7 +889,7 @@ void FSP_SRV::OpenDataStorageByDataId(Kernel::HLERequestContext& ctx) {
     rb.PushIpcInterface<IStorage>(std::move(storage));
 }
 
-void FSP_SRV::OpenRomStorage(Kernel::HLERequestContext& ctx) {
+void FSP_SRV::OpenPatchDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx) {
     IPC::RequestParser rp{ctx};
 
     auto storage_id = rp.PopRaw<FileSys::StorageId>();

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

@@ -19,17 +19,17 @@ public:
     ~FSP_SRV() override;
 
 private:
-    void Initialize(Kernel::HLERequestContext& ctx);
+    void SetCurrentProcess(Kernel::HLERequestContext& ctx);
     void OpenFileSystemWithPatch(Kernel::HLERequestContext& ctx);
-    void MountSdCard(Kernel::HLERequestContext& ctx);
-    void CreateSaveData(Kernel::HLERequestContext& ctx);
-    void MountSaveData(Kernel::HLERequestContext& ctx);
+    void OpenSdCardFileSystem(Kernel::HLERequestContext& ctx);
+    void CreateSaveDataFileSystem(Kernel::HLERequestContext& ctx);
+    void OpenSaveDataFileSystem(Kernel::HLERequestContext& ctx);
     void OpenReadOnlySaveDataFileSystem(Kernel::HLERequestContext& ctx);
     void OpenSaveDataInfoReaderBySaveDataSpaceId(Kernel::HLERequestContext& ctx);
     void GetGlobalAccessLogMode(Kernel::HLERequestContext& ctx);
     void OpenDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx);
     void OpenDataStorageByDataId(Kernel::HLERequestContext& ctx);
-    void OpenRomStorage(Kernel::HLERequestContext& ctx);
+    void OpenPatchDataStorageByCurrentProcess(Kernel::HLERequestContext& ctx);
 
     FileSys::VirtualFile romfs;
 };

src/core/hle/service/ncm/ncm.cpp

@@ -40,10 +40,10 @@ public:
             {6, nullptr, "CloseContentStorageForcibly"},
             {7, nullptr, "CloseContentMetaDatabaseForcibly"},
             {8, nullptr, "CleanupContentMetaDatabase"},
-            {9, nullptr, "OpenContentStorage2"},
-            {10, nullptr, "CloseContentStorage"},
-            {11, nullptr, "OpenContentMetaDatabase2"},
-            {12, nullptr, "CloseContentMetaDatabase"},
+            {9, nullptr, "ActivateContentStorage"},
+            {10, nullptr, "InactivateContentStorage"},
+            {11, nullptr, "ActivateContentMetaDatabase"},
+            {12, nullptr, "InactivateContentMetaDatabase"},
         };
         // clang-format on
 

src/core/hle/service/ns/ns.cpp

@@ -43,7 +43,7 @@ public:
             {11, nullptr, "CalculateApplicationOccupiedSize"},
             {16, nullptr, "PushApplicationRecord"},
             {17, nullptr, "ListApplicationRecordContentMeta"},
-            {19, nullptr, "LaunchApplication"},
+            {19, nullptr, "LaunchApplicationOld"},
             {21, nullptr, "GetApplicationContentPath"},
             {22, nullptr, "TerminateApplication"},
             {23, nullptr, "ResolveApplicationContentPath"},
@@ -96,10 +96,10 @@ public:
             {86, nullptr, "EnableApplicationCrashReport"},
             {87, nullptr, "IsApplicationCrashReportEnabled"},
             {90, nullptr, "BoostSystemMemoryResourceLimit"},
-            {91, nullptr, "Unknown1"},
-            {92, nullptr, "Unknown2"},
+            {91, nullptr, "DeprecatedLaunchApplication"},
+            {92, nullptr, "GetRunningApplicationProgramId"},
             {93, nullptr, "GetMainApplicationProgramIndex"},
-            {94, nullptr, "LaunchApplication2"},
+            {94, nullptr, "LaunchApplication"},
             {95, nullptr, "GetApplicationLaunchInfo"},
             {96, nullptr, "AcquireApplicationLaunchInfo"},
             {97, nullptr, "GetMainApplicationProgramIndex2"},
@@ -163,7 +163,7 @@ public:
             {907, nullptr, "WithdrawApplicationUpdateRequest"},
             {908, nullptr, "ListApplicationRecordInstalledContentMeta"},
             {909, nullptr, "WithdrawCleanupAddOnContentsWithNoRightsRecommendation"},
-            {910, nullptr, "Unknown3"},
+            {910, nullptr, "HasApplicationRecord"},
             {911, nullptr, "SetPreInstalledApplication"},
             {912, nullptr, "ClearPreInstalledApplicationFlag"},
             {1000, nullptr, "RequestVerifyApplicationDeprecated"},
@@ -219,10 +219,10 @@ public:
             {2015, nullptr, "CompareSystemDeliveryInfo"},
             {2016, nullptr, "ListNotCommittedContentMeta"},
             {2017, nullptr, "CreateDownloadTask"},
-            {2018, nullptr, "Unknown4"},
-            {2050, nullptr, "Unknown5"},
-            {2100, nullptr, "Unknown6"},
-            {2101, nullptr, "Unknown7"},
+            {2018, nullptr, "GetApplicationDeliveryInfoHash"},
+            {2050, nullptr, "GetApplicationRightsOnClient"},
+            {2100, nullptr, "GetApplicationTerminateResult"},
+            {2101, nullptr, "GetRawApplicationTerminateResult"},
             {2150, nullptr, "CreateRightsEnvironment"},
             {2151, nullptr, "DestroyRightsEnvironment"},
             {2152, nullptr, "ActivateRightsEnvironment"},
@@ -237,10 +237,10 @@ public:
             {2182, nullptr, "SetActiveRightsContextUsingStateToRightsEnvironment"},
             {2190, nullptr, "GetRightsEnvironmentHandleForApplication"},
             {2199, nullptr, "GetRightsEnvironmentCountForDebug"},
-            {2200, nullptr, "Unknown8"},
-            {2201, nullptr, "Unknown9"},
-            {2250, nullptr, "Unknown10"},
-            {2300, nullptr, "Unknown11"},
+            {2200, nullptr, "GetGameCardApplicationCopyIdentifier"},
+            {2201, nullptr, "GetInstalledApplicationCopyIdentifier"},
+            {2250, nullptr, "RequestReportActiveELicence"},
+            {2300, nullptr, "ListEventLog"},
         };
         // clang-format on
 
@@ -355,6 +355,7 @@ public:
         static const FunctionInfo functions[] = {
             {21, nullptr, "GetApplicationContentPath"},
             {23, nullptr, "ResolveApplicationContentPath"},
+            {93, nullptr, "GetRunningApplicationProgramId"},
         };
         // clang-format on
 
@@ -389,6 +390,11 @@ public:
         // clang-format off
         static const FunctionInfo functions[] = {
             {0, nullptr, "RequestLinkDevice"},
+            {1, nullptr, "RequestCleanupAllPreInstalledApplications"},
+            {2, nullptr, "RequestCleanupPreInstalledApplication"},
+            {3, nullptr, "RequestSyncRights"},
+            {4, nullptr, "RequestUnlinkDevice"},
+            {5, nullptr, "RequestRevokeAllELicense"},
         };
         // clang-format on
 
@@ -403,7 +409,7 @@ public:
         static const FunctionInfo functions[] = {
             {100, nullptr, "ResetToFactorySettings"},
             {101, nullptr, "ResetToFactorySettingsWithoutUserSaveData"},
-            {102, nullptr, "ResetToFactorySettingsForRefurbishment "},
+            {102, nullptr, "ResetToFactorySettingsForRefurbishment"},
         };
         // clang-format on
 

src/core/hle/service/nvflinger/nvflinger.cpp

@@ -5,7 +5,6 @@
 #include <algorithm>
 #include <optional>
 
-#include "common/alignment.h"
 #include "common/assert.h"
 #include "common/logging/log.h"
 #include "common/microprofile.h"
@@ -22,7 +21,6 @@
 #include "core/hle/service/nvflinger/nvflinger.h"
 #include "core/perf_stats.h"
 #include "video_core/renderer_base.h"
-#include "video_core/video_core.h"
 
 namespace Service::NVFlinger {
 
@@ -30,12 +28,6 @@ constexpr std::size_t SCREEN_REFRESH_RATE = 60;
 constexpr u64 frame_ticks = static_cast<u64>(CoreTiming::BASE_CLOCK_RATE / SCREEN_REFRESH_RATE);
 
 NVFlinger::NVFlinger() {
-    // Add the different displays to the list of displays.
-    displays.emplace_back(0, "Default");
-    displays.emplace_back(1, "External");
-    displays.emplace_back(2, "Edid");
-    displays.emplace_back(3, "Internal");
-
     // Schedule the screen composition events
     composition_event =
         CoreTiming::RegisterEvent("ScreenComposition", [this](u64 userdata, int cycles_late) {
@@ -55,13 +47,13 @@ void NVFlinger::SetNVDrvInstance(std::shared_ptr<Nvidia::Module> instance) {
 }
 
 u64 NVFlinger::OpenDisplay(std::string_view name) {
-    LOG_WARNING(Service, "Opening display {}", name);
+    LOG_DEBUG(Service, "Opening \"{}\" display", name);
 
     // TODO(Subv): Currently we only support the Default display.
     ASSERT(name == "Default");
 
-    auto itr = std::find_if(displays.begin(), displays.end(),
-                            [&](const Display& display) { return display.name == name; });
+    const auto itr = std::find_if(displays.begin(), displays.end(),
+                                  [&](const Display& display) { return display.name == name; });
 
     ASSERT(itr != displays.end());
 
@@ -69,48 +61,66 @@ u64 NVFlinger::OpenDisplay(std::string_view name) {
 }
 
 u64 NVFlinger::CreateLayer(u64 display_id) {
-    auto& display = GetDisplay(display_id);
+    auto& display = FindDisplay(display_id);
 
     ASSERT_MSG(display.layers.empty(), "Only one layer is supported per display at the moment");
 
-    u64 layer_id = next_layer_id++;
-    u32 buffer_queue_id = next_buffer_queue_id++;
+    const u64 layer_id = next_layer_id++;
+    const u32 buffer_queue_id = next_buffer_queue_id++;
     auto buffer_queue = std::make_shared<BufferQueue>(buffer_queue_id, layer_id);
     display.layers.emplace_back(layer_id, buffer_queue);
     buffer_queues.emplace_back(std::move(buffer_queue));
     return layer_id;
 }
 
-u32 NVFlinger::GetBufferQueueId(u64 display_id, u64 layer_id) {
-    const auto& layer = GetLayer(display_id, layer_id);
+u32 NVFlinger::FindBufferQueueId(u64 display_id, u64 layer_id) const {
+    const auto& layer = FindLayer(display_id, layer_id);
     return layer.buffer_queue->GetId();
 }
 
 Kernel::SharedPtr<Kernel::ReadableEvent> NVFlinger::GetVsyncEvent(u64 display_id) {
-    return GetDisplay(display_id).vsync_event.readable;
+    return FindDisplay(display_id).vsync_event.readable;
 }
 
-std::shared_ptr<BufferQueue> NVFlinger::GetBufferQueue(u32 id) const {
-    auto itr = std::find_if(buffer_queues.begin(), buffer_queues.end(),
-                            [&](const auto& queue) { return queue->GetId() == id; });
+std::shared_ptr<BufferQueue> NVFlinger::FindBufferQueue(u32 id) const {
+    const auto itr = std::find_if(buffer_queues.begin(), buffer_queues.end(),
+                                  [&](const auto& queue) { return queue->GetId() == id; });
 
     ASSERT(itr != buffer_queues.end());
     return *itr;
 }
 
-Display& NVFlinger::GetDisplay(u64 display_id) {
-    auto itr = std::find_if(displays.begin(), displays.end(),
-                            [&](const Display& display) { return display.id == display_id; });
+Display& NVFlinger::FindDisplay(u64 display_id) {
+    const auto itr = std::find_if(displays.begin(), displays.end(),
+                                  [&](const Display& display) { return display.id == display_id; });
 
     ASSERT(itr != displays.end());
     return *itr;
 }
 
-Layer& NVFlinger::GetLayer(u64 display_id, u64 layer_id) {
-    auto& display = GetDisplay(display_id);
+const Display& NVFlinger::FindDisplay(u64 display_id) const {
+    const auto itr = std::find_if(displays.begin(), displays.end(),
+                                  [&](const Display& display) { return display.id == display_id; });
 
-    auto itr = std::find_if(display.layers.begin(), display.layers.end(),
-                            [&](const Layer& layer) { return layer.id == layer_id; });
+    ASSERT(itr != displays.end());
+    return *itr;
+}
+
+Layer& NVFlinger::FindLayer(u64 display_id, u64 layer_id) {
+    auto& display = FindDisplay(display_id);
+
+    const auto itr = std::find_if(display.layers.begin(), display.layers.end(),
+                                  [&](const Layer& layer) { return layer.id == layer_id; });
+
+    ASSERT(itr != display.layers.end());
+    return *itr;
+}
+
+const Layer& NVFlinger::FindLayer(u64 display_id, u64 layer_id) const {
+    const auto& display = FindDisplay(display_id);
+
+    const auto itr = std::find_if(display.layers.begin(), display.layers.end(),
+                                  [&](const Layer& layer) { return layer.id == layer_id; });
 
     ASSERT(itr != display.layers.end());
     return *itr;
@@ -145,7 +155,7 @@ void NVFlinger::Compose() {
             continue;
         }
 
-        auto& igbp_buffer = buffer->get().igbp_buffer;
+        const auto& igbp_buffer = buffer->get().igbp_buffer;
 
         // Now send the buffer to the GPU for drawing.
         // TODO(Subv): Support more than just disp0. The display device selection is probably based

src/core/hle/service/nvflinger/nvflinger.h

@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include <array>
 #include <memory>
 #include <string>
 #include <string_view>
@@ -56,35 +57,47 @@ public:
     /// Sets the NVDrv module instance to use to send buffers to the GPU.
     void SetNVDrvInstance(std::shared_ptr<Nvidia::Module> instance);
 
-    /// Opens the specified display and returns the id.
+    /// Opens the specified display and returns the ID.
     u64 OpenDisplay(std::string_view name);
 
-    /// Creates a layer on the specified display and returns the layer id.
+    /// Creates a layer on the specified display and returns the layer ID.
     u64 CreateLayer(u64 display_id);
 
-    /// Gets the buffer queue id of the specified layer in the specified display.
-    u32 GetBufferQueueId(u64 display_id, u64 layer_id);
+    /// Finds the buffer queue ID of the specified layer in the specified display.
+    u32 FindBufferQueueId(u64 display_id, u64 layer_id) const;
 
     /// Gets the vsync event for the specified display.
     Kernel::SharedPtr<Kernel::ReadableEvent> GetVsyncEvent(u64 display_id);
 
-    /// Obtains a buffer queue identified by the id.
-    std::shared_ptr<BufferQueue> GetBufferQueue(u32 id) const;
+    /// Obtains a buffer queue identified by the ID.
+    std::shared_ptr<BufferQueue> FindBufferQueue(u32 id) const;
 
     /// Performs a composition request to the emulated nvidia GPU and triggers the vsync events when
     /// finished.
     void Compose();
 
 private:
-    /// Returns the display identified by the specified id.
-    Display& GetDisplay(u64 display_id);
+    /// Finds the display identified by the specified ID.
+    Display& FindDisplay(u64 display_id);
 
-    /// Returns the layer identified by the specified id in the desired display.
-    Layer& GetLayer(u64 display_id, u64 layer_id);
+    /// Finds the display identified by the specified ID.
+    const Display& FindDisplay(u64 display_id) const;
+
+    /// Finds the layer identified by the specified ID in the desired display.
+    Layer& FindLayer(u64 display_id, u64 layer_id);
+
+    /// Finds the layer identified by the specified ID in the desired display.
+    const Layer& FindLayer(u64 display_id, u64 layer_id) const;
 
     std::shared_ptr<Nvidia::Module> nvdrv;
 
-    std::vector<Display> displays;
+    std::array<Display, 5> displays{{
+        {0, "Default"},
+        {1, "External"},
+        {2, "Edid"},
+        {3, "Internal"},
+        {4, "Null"},
+    }};
     std::vector<std::shared_ptr<BufferQueue>> buffer_queues;
 
     /// Id to use for the next layer that is created, this counter is shared among all displays.

src/core/hle/service/pm/pm.cpp

@@ -13,7 +13,7 @@ public:
     explicit BootMode() : ServiceFramework{"pm:bm"} {
         static const FunctionInfo functions[] = {
             {0, &BootMode::GetBootMode, "GetBootMode"},
-            {1, nullptr, "SetMaintenanceBoot"},
+            {1, &BootMode::SetMaintenanceBoot, "SetMaintenanceBoot"},
         };
         RegisterHandlers(functions);
     }
@@ -24,8 +24,19 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 3};
         rb.Push(RESULT_SUCCESS);
-        rb.Push<u32>(static_cast<u32>(SystemBootMode::Normal)); // Normal boot mode
+        rb.PushEnum(boot_mode);
     }
+
+    void SetMaintenanceBoot(Kernel::HLERequestContext& ctx) {
+        LOG_DEBUG(Service_PM, "called");
+
+        boot_mode = SystemBootMode::Maintenance;
+
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+    }
+
+    SystemBootMode boot_mode = SystemBootMode::Normal;
 };
 
 class DebugMonitor final : public ServiceFramework<DebugMonitor> {

src/core/hle/service/pm/pm.h

@@ -9,7 +9,12 @@ class ServiceManager;
 }
 
 namespace Service::PM {
-enum class SystemBootMode : u32 { Normal = 0, Maintenance = 1 };
+
+enum class SystemBootMode {
+    Normal,
+    Maintenance,
+};
+
 /// Registers all PM services with the specified service manager.
 void InstallInterfaces(SM::ServiceManager& service_manager);
 

src/core/hle/service/psc/psc.cpp

@@ -17,13 +17,13 @@ public:
     explicit PSC_C() : ServiceFramework{"psc:c"} {
         // clang-format off
         static const FunctionInfo functions[] = {
-            {0, nullptr, "Unknown1"},
-            {1, nullptr, "Unknown2"},
-            {2, nullptr, "Unknown3"},
-            {3, nullptr, "Unknown4"},
-            {4, nullptr, "Unknown5"},
-            {5, nullptr, "Unknown6"},
-            {6, nullptr, "Unknown7"},
+            {0, nullptr, "Initialize"},
+            {1, nullptr, "DispatchRequest"},
+            {2, nullptr, "GetResult"},
+            {3, nullptr, "GetState"},
+            {4, nullptr, "Cancel"},
+            {5, nullptr, "PrintModuleInformation"},
+            {6, nullptr, "GetModuleInformation"},
         };
         // clang-format on
 
@@ -39,7 +39,8 @@ public:
             {0, nullptr, "Initialize"},
             {1, nullptr, "GetRequest"},
             {2, nullptr, "Acknowledge"},
-            {3, nullptr, "Unknown1"},
+            {3, nullptr, "Finalize"},
+            {4, nullptr, "AcknowledgeEx"},
         };
         // clang-format on
 

src/core/hle/service/vi/vi.cpp

@@ -524,7 +524,7 @@ private:
         LOG_DEBUG(Service_VI, "called. id=0x{:08X} transaction={:X}, flags=0x{:08X}", id,
                   static_cast<u32>(transaction), flags);
 
-        auto buffer_queue = nv_flinger->GetBufferQueue(id);
+        auto buffer_queue = nv_flinger->FindBufferQueue(id);
 
         if (transaction == TransactionId::Connect) {
             IGBPConnectRequestParcel request{ctx.ReadBuffer()};
@@ -558,7 +558,7 @@ private:
                     [=](Kernel::SharedPtr<Kernel::Thread> thread, Kernel::HLERequestContext& ctx,
                         Kernel::ThreadWakeupReason reason) {
                         // Repeat TransactParcel DequeueBuffer when a buffer is available
-                        auto buffer_queue = nv_flinger->GetBufferQueue(id);
+                        auto buffer_queue = nv_flinger->FindBufferQueue(id);
                         std::optional<u32> slot = buffer_queue->DequeueBuffer(width, height);
                         ASSERT_MSG(slot != std::nullopt, "Could not dequeue buffer.");
 
@@ -628,7 +628,7 @@ private:
 
         LOG_WARNING(Service_VI, "(STUBBED) called id={}, unknown={:08X}", id, unknown);
 
-        const auto buffer_queue = nv_flinger->GetBufferQueue(id);
+        const auto buffer_queue = nv_flinger->FindBufferQueue(id);
 
         // TODO(Subv): Find out what this actually is.
         IPC::ResponseBuilder rb{ctx, 2, 1};
@@ -704,13 +704,14 @@ private:
         rb.Push(RESULT_SUCCESS);
     }
 
+    // This function currently does nothing but return a success error code in
+    // the vi library itself, so do the same thing, but log out the passed in values.
     void SetLayerVisibility(Kernel::HLERequestContext& ctx) {
         IPC::RequestParser rp{ctx};
         const u64 layer_id = rp.Pop<u64>();
         const bool visibility = rp.Pop<bool>();
 
-        LOG_WARNING(Service_VI, "(STUBBED) called, layer_id=0x{:08X}, visibility={}", layer_id,
-                    visibility);
+        LOG_DEBUG(Service_VI, "called, layer_id=0x{:08X}, visibility={}", layer_id, visibility);
 
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(RESULT_SUCCESS);
@@ -1043,7 +1044,7 @@ private:
         LOG_DEBUG(Service_VI, "called. layer_id=0x{:016X}, aruid=0x{:016X}", layer_id, aruid);
 
         const u64 display_id = nv_flinger->OpenDisplay(display_name);
-        const u32 buffer_queue_id = nv_flinger->GetBufferQueueId(display_id, layer_id);
+        const u32 buffer_queue_id = nv_flinger->FindBufferQueueId(display_id, layer_id);
 
         NativeWindow native_window{buffer_queue_id};
         IPC::ResponseBuilder rb{ctx, 4};
@@ -1062,7 +1063,7 @@ private:
         // TODO(Subv): What's the difference between a Stray and a Managed layer?
 
         const u64 layer_id = nv_flinger->CreateLayer(display_id);
-        const u32 buffer_queue_id = nv_flinger->GetBufferQueueId(display_id, layer_id);
+        const u32 buffer_queue_id = nv_flinger->FindBufferQueueId(display_id, layer_id);
 
         NativeWindow native_window{buffer_queue_id};
         IPC::ResponseBuilder rb{ctx, 6};

src/video_core/CMakeLists.txt

@@ -87,6 +87,7 @@ add_library(video_core STATIC
     shader/decode.cpp
     shader/shader_ir.cpp
     shader/shader_ir.h
+    shader/track.cpp
     surface.cpp
     surface.h
     textures/astc.cpp

src/video_core/dma_pusher.h

@@ -83,7 +83,7 @@ private:
         u32 subchannel;        ///< Current subchannel
         u32 method_count;      ///< Current method count
         u32 length_pending;    ///< Large NI command length pending
-        bool non_incrementing; ///< Current command’s NI flag
+        bool non_incrementing; ///< Current command's NI flag
     };
 
     DmaState dma_state{};

src/video_core/engines/shader_bytecode.h

@@ -208,6 +208,8 @@ enum class UniformType : u64 {
     SignedShort = 3,
     Single = 4,
     Double = 5,
+    Quad = 6,
+    UnsignedQuad = 7,
 };
 
 enum class StoreType : u64 {
@@ -784,6 +786,12 @@ union Instruction {
         BitField<44, 2, u64> unknown;
     } st_l;
 
+    union {
+        BitField<48, 3, UniformType> type;
+        BitField<46, 2, u64> cache_mode;
+        BitField<20, 24, s64> immediate_offset;
+    } ldg;
+
     union {
         BitField<0, 3, u64> pred0;
         BitField<3, 3, u64> pred3;
@@ -1240,11 +1248,19 @@ union Instruction {
     union {
         BitField<20, 14, u64> offset;
         BitField<34, 5, u64> index;
+
+        u64 GetOffset() const {
+            return offset * 4;
+        }
     } cbuf34;
 
     union {
         BitField<20, 16, s64> offset;
         BitField<36, 5, u64> index;
+
+        s64 GetOffset() const {
+            return offset;
+        }
     } cbuf36;
 
     // Unsure about the size of this one.

src/video_core/gpu.cpp

@@ -3,6 +3,8 @@
 // Refer to the license.txt file included.
 
 #include "common/assert.h"
+#include "core/core_timing.h"
+#include "core/memory.h"
 #include "video_core/engines/fermi_2d.h"
 #include "video_core/engines/kepler_memory.h"
 #include "video_core/engines/maxwell_3d.h"
@@ -124,9 +126,36 @@ u32 DepthFormatBytesPerPixel(DepthFormat format) {
     }
 }
 
+// Note that, traditionally, methods are treated as 4-byte addressable locations, and hence
+// their numbers are written down multiplied by 4 in Docs. Here we are not multiply by 4.
+// So the values you see in docs might be multiplied by 4.
 enum class BufferMethods {
-    BindObject = 0,
-    CountBufferMethods = 0x40,
+    BindObject = 0x0,
+    Nop = 0x2,
+    SemaphoreAddressHigh = 0x4,
+    SemaphoreAddressLow = 0x5,
+    SemaphoreSequence = 0x6,
+    SemaphoreTrigger = 0x7,
+    NotifyIntr = 0x8,
+    WrcacheFlush = 0x9,
+    Unk28 = 0xA,
+    Unk2c = 0xB,
+    RefCnt = 0x14,
+    SemaphoreAcquire = 0x1A,
+    SemaphoreRelease = 0x1B,
+    Unk70 = 0x1C,
+    Unk74 = 0x1D,
+    Unk78 = 0x1E,
+    Unk7c = 0x1F,
+    Yield = 0x20,
+    NonPullerMethods = 0x40,
+};
+
+enum class GpuSemaphoreOperation {
+    AcquireEqual = 0x1,
+    WriteLong = 0x2,
+    AcquireGequal = 0x4,
+    AcquireMask = 0x8,
 };
 
 void GPU::CallMethod(const MethodCall& method_call) {
@@ -135,20 +164,78 @@ void GPU::CallMethod(const MethodCall& method_call) {
 
     ASSERT(method_call.subchannel < bound_engines.size());
 
-    if (method_call.method == static_cast<u32>(BufferMethods::BindObject)) {
-        // Bind the current subchannel to the desired engine id.
-        LOG_DEBUG(HW_GPU, "Binding subchannel {} to engine {}", method_call.subchannel,
-                  method_call.argument);
-        bound_engines[method_call.subchannel] = static_cast<EngineID>(method_call.argument);
-        return;
+    if (ExecuteMethodOnEngine(method_call)) {
+        CallEngineMethod(method_call);
+    } else {
+        CallPullerMethod(method_call);
     }
+}
 
-    if (method_call.method < static_cast<u32>(BufferMethods::CountBufferMethods)) {
-        // TODO(Subv): Research and implement these methods.
-        LOG_ERROR(HW_GPU, "Special buffer methods other than Bind are not implemented");
-        return;
+bool GPU::ExecuteMethodOnEngine(const MethodCall& method_call) {
+    const auto method = static_cast<BufferMethods>(method_call.method);
+    return method >= BufferMethods::NonPullerMethods;
+}
+
+void GPU::CallPullerMethod(const MethodCall& method_call) {
+    regs.reg_array[method_call.method] = method_call.argument;
+    const auto method = static_cast<BufferMethods>(method_call.method);
+
+    switch (method) {
+    case BufferMethods::BindObject: {
+        ProcessBindMethod(method_call);
+        break;
     }
+    case BufferMethods::Nop:
+    case BufferMethods::SemaphoreAddressHigh:
+    case BufferMethods::SemaphoreAddressLow:
+    case BufferMethods::SemaphoreSequence:
+    case BufferMethods::RefCnt:
+        break;
+    case BufferMethods::SemaphoreTrigger: {
+        ProcessSemaphoreTriggerMethod();
+        break;
+    }
+    case BufferMethods::NotifyIntr: {
+        // TODO(Kmather73): Research and implement this method.
+        LOG_ERROR(HW_GPU, "Special puller engine method NotifyIntr not implemented");
+        break;
+    }
+    case BufferMethods::WrcacheFlush: {
+        // TODO(Kmather73): Research and implement this method.
+        LOG_ERROR(HW_GPU, "Special puller engine method WrcacheFlush not implemented");
+        break;
+    }
+    case BufferMethods::Unk28: {
+        // TODO(Kmather73): Research and implement this method.
+        LOG_ERROR(HW_GPU, "Special puller engine method Unk28 not implemented");
+        break;
+    }
+    case BufferMethods::Unk2c: {
+        // TODO(Kmather73): Research and implement this method.
+        LOG_ERROR(HW_GPU, "Special puller engine method Unk2c not implemented");
+        break;
+    }
+    case BufferMethods::SemaphoreAcquire: {
+        ProcessSemaphoreAcquire();
+        break;
+    }
+    case BufferMethods::SemaphoreRelease: {
+        ProcessSemaphoreRelease();
+        break;
+    }
+    case BufferMethods::Yield: {
+        // TODO(Kmather73): Research and implement this method.
+        LOG_ERROR(HW_GPU, "Special puller engine method Yield not implemented");
+        break;
+    }
+    default:
+        LOG_ERROR(HW_GPU, "Special puller engine method {:X} not implemented",
+                  static_cast<u32>(method));
+        break;
+    }
+}
 
+void GPU::CallEngineMethod(const MethodCall& method_call) {
     const EngineID engine = bound_engines[method_call.subchannel];
 
     switch (engine) {
@@ -172,4 +259,76 @@ void GPU::CallMethod(const MethodCall& method_call) {
     }
 }
 
+void GPU::ProcessBindMethod(const MethodCall& method_call) {
+    // Bind the current subchannel to the desired engine id.
+    LOG_DEBUG(HW_GPU, "Binding subchannel {} to engine {}", method_call.subchannel,
+              method_call.argument);
+    bound_engines[method_call.subchannel] = static_cast<EngineID>(method_call.argument);
+}
+
+void GPU::ProcessSemaphoreTriggerMethod() {
+    const auto semaphoreOperationMask = 0xF;
+    const auto op =
+        static_cast<GpuSemaphoreOperation>(regs.semaphore_trigger & semaphoreOperationMask);
+    if (op == GpuSemaphoreOperation::WriteLong) {
+        auto address = memory_manager->GpuToCpuAddress(regs.smaphore_address.SmaphoreAddress());
+        struct Block {
+            u32 sequence;
+            u32 zeros = 0;
+            u64 timestamp;
+        };
+
+        Block block{};
+        block.sequence = regs.semaphore_sequence;
+        // TODO(Kmather73): Generate a real GPU timestamp and write it here instead of
+        // CoreTiming
+        block.timestamp = CoreTiming::GetTicks();
+        Memory::WriteBlock(*address, &block, sizeof(block));
+    } else {
+        const auto address =
+            memory_manager->GpuToCpuAddress(regs.smaphore_address.SmaphoreAddress());
+        const u32 word = Memory::Read32(*address);
+        if ((op == GpuSemaphoreOperation::AcquireEqual && word == regs.semaphore_sequence) ||
+            (op == GpuSemaphoreOperation::AcquireGequal &&
+             static_cast<s32>(word - regs.semaphore_sequence) > 0) ||
+            (op == GpuSemaphoreOperation::AcquireMask && (word & regs.semaphore_sequence))) {
+            // Nothing to do in this case
+        } else {
+            regs.acquire_source = true;
+            regs.acquire_value = regs.semaphore_sequence;
+            if (op == GpuSemaphoreOperation::AcquireEqual) {
+                regs.acquire_active = true;
+                regs.acquire_mode = false;
+            } else if (op == GpuSemaphoreOperation::AcquireGequal) {
+                regs.acquire_active = true;
+                regs.acquire_mode = true;
+            } else if (op == GpuSemaphoreOperation::AcquireMask) {
+                // TODO(kemathe) The acquire mask operation waits for a value that, ANDed with
+                // semaphore_sequence, gives a non-0 result
+                LOG_ERROR(HW_GPU, "Invalid semaphore operation AcquireMask not implemented");
+            } else {
+                LOG_ERROR(HW_GPU, "Invalid semaphore operation");
+            }
+        }
+    }
+}
+
+void GPU::ProcessSemaphoreRelease() {
+    const auto address = memory_manager->GpuToCpuAddress(regs.smaphore_address.SmaphoreAddress());
+    Memory::Write32(*address, regs.semaphore_release);
+}
+
+void GPU::ProcessSemaphoreAcquire() {
+    const auto address = memory_manager->GpuToCpuAddress(regs.smaphore_address.SmaphoreAddress());
+    const u32 word = Memory::Read32(*address);
+    const auto value = regs.semaphore_acquire;
+    if (word != value) {
+        regs.acquire_active = true;
+        regs.acquire_value = value;
+        // TODO(kemathe73) figure out how to do the acquire_timeout
+        regs.acquire_mode = false;
+        regs.acquire_source = false;
+    }
+}
+
 } // namespace Tegra

src/video_core/gpu.h

@@ -156,6 +156,46 @@ public:
     /// Returns a const reference to the GPU DMA pusher.
     const Tegra::DmaPusher& DmaPusher() const;
 
+    struct Regs {
+        static constexpr size_t NUM_REGS = 0x100;
+
+        union {
+            struct {
+                INSERT_PADDING_WORDS(0x4);
+                struct {
+                    u32 address_high;
+                    u32 address_low;
+
+                    GPUVAddr SmaphoreAddress() const {
+                        return static_cast<GPUVAddr>((static_cast<GPUVAddr>(address_high) << 32) |
+                                                     address_low);
+                    }
+                } smaphore_address;
+
+                u32 semaphore_sequence;
+                u32 semaphore_trigger;
+                INSERT_PADDING_WORDS(0xC);
+
+                // The puser and the puller share the reference counter, the pusher only has read
+                // access
+                u32 reference_count;
+                INSERT_PADDING_WORDS(0x5);
+
+                u32 semaphore_acquire;
+                u32 semaphore_release;
+                INSERT_PADDING_WORDS(0xE4);
+
+                // Puller state
+                u32 acquire_mode;
+                u32 acquire_source;
+                u32 acquire_active;
+                u32 acquire_timeout;
+                u32 acquire_value;
+            };
+            std::array<u32, NUM_REGS> reg_array;
+        };
+    } regs{};
+
 private:
     std::unique_ptr<Tegra::DmaPusher> dma_pusher;
     std::unique_ptr<Tegra::MemoryManager> memory_manager;
@@ -173,6 +213,37 @@ private:
     std::unique_ptr<Engines::MaxwellDMA> maxwell_dma;
     /// Inline memory engine
     std::unique_ptr<Engines::KeplerMemory> kepler_memory;
+
+    void ProcessBindMethod(const MethodCall& method_call);
+    void ProcessSemaphoreTriggerMethod();
+    void ProcessSemaphoreRelease();
+    void ProcessSemaphoreAcquire();
+
+    // Calls a GPU puller method.
+    void CallPullerMethod(const MethodCall& method_call);
+    // Calls a GPU engine method.
+    void CallEngineMethod(const MethodCall& method_call);
+    // Determines where the method should be executed.
+    bool ExecuteMethodOnEngine(const MethodCall& method_call);
 };
 
+#define ASSERT_REG_POSITION(field_name, position)                                                  \
+    static_assert(offsetof(GPU::Regs, field_name) == position * 4,                                 \
+                  "Field " #field_name " has invalid position")
+
+ASSERT_REG_POSITION(smaphore_address, 0x4);
+ASSERT_REG_POSITION(semaphore_sequence, 0x6);
+ASSERT_REG_POSITION(semaphore_trigger, 0x7);
+ASSERT_REG_POSITION(reference_count, 0x14);
+ASSERT_REG_POSITION(semaphore_acquire, 0x1A);
+ASSERT_REG_POSITION(semaphore_release, 0x1B);
+
+ASSERT_REG_POSITION(acquire_mode, 0x100);
+ASSERT_REG_POSITION(acquire_source, 0x101);
+ASSERT_REG_POSITION(acquire_active, 0x102);
+ASSERT_REG_POSITION(acquire_timeout, 0x103);
+ASSERT_REG_POSITION(acquire_value, 0x104);
+
+#undef ASSERT_REG_POSITION
+
 } // namespace Tegra

src/video_core/rasterizer_interface.h

@@ -49,11 +49,6 @@ public:
         return false;
     }
 
-    /// Attempt to use a faster method to fill a region
-    virtual bool AccelerateFill(const void* config) {
-        return false;
-    }
-
     /// Attempt to use a faster method to display the framebuffer to screen
     virtual bool AccelerateDisplay(const Tegra::FramebufferConfig& config, VAddr framebuffer_addr,
                                    u32 pixel_stride) {

src/video_core/renderer_opengl/gl_global_cache.cpp

@@ -4,8 +4,13 @@
 
 #include <glad/glad.h>
 
+#include "common/assert.h"
+#include "common/logging/log.h"
+#include "core/core.h"
+#include "core/memory.h"
 #include "video_core/renderer_opengl/gl_global_cache.h"
 #include "video_core/renderer_opengl/gl_rasterizer.h"
+#include "video_core/renderer_opengl/gl_shader_decompiler.h"
 #include "video_core/renderer_opengl/utils.h"
 
 namespace OpenGL {
@@ -18,7 +23,72 @@ CachedGlobalRegion::CachedGlobalRegion(VAddr addr, u32 size) : addr{addr}, size{
     LabelGLObject(GL_BUFFER, buffer.handle, addr, "GlobalMemory");
 }
 
+void CachedGlobalRegion::Reload(u32 size_) {
+    constexpr auto max_size = static_cast<u32>(RasterizerOpenGL::MaxGlobalMemorySize);
+
+    size = size_;
+    if (size > max_size) {
+        size = max_size;
+        LOG_CRITICAL(HW_GPU, "Global region size {} exceeded the expected size {}!", size_,
+                     max_size);
+    }
+
+    // TODO(Rodrigo): Get rid of Memory::GetPointer with a staging buffer
+    glBindBuffer(GL_SHADER_STORAGE_BUFFER, buffer.handle);
+    glBufferData(GL_SHADER_STORAGE_BUFFER, size, Memory::GetPointer(addr), GL_DYNAMIC_DRAW);
+}
+
+GlobalRegion GlobalRegionCacheOpenGL::TryGetReservedGlobalRegion(VAddr addr, u32 size) const {
+    const auto search{reserve.find(addr)};
+    if (search == reserve.end()) {
+        return {};
+    }
+    return search->second;
+}
+
+GlobalRegion GlobalRegionCacheOpenGL::GetUncachedGlobalRegion(VAddr addr, u32 size) {
+    GlobalRegion region{TryGetReservedGlobalRegion(addr, size)};
+    if (!region) {
+        // No reserved surface available, create a new one and reserve it
+        region = std::make_shared<CachedGlobalRegion>(addr, size);
+        ReserveGlobalRegion(region);
+    }
+    region->Reload(size);
+    return region;
+}
+
+void GlobalRegionCacheOpenGL::ReserveGlobalRegion(const GlobalRegion& region) {
+    reserve[region->GetAddr()] = region;
+}
+
 GlobalRegionCacheOpenGL::GlobalRegionCacheOpenGL(RasterizerOpenGL& rasterizer)
     : RasterizerCache{rasterizer} {}
 
+GlobalRegion GlobalRegionCacheOpenGL::GetGlobalRegion(
+    const GLShader::GlobalMemoryEntry& global_region,
+    Tegra::Engines::Maxwell3D::Regs::ShaderStage stage) {
+
+    auto& gpu{Core::System::GetInstance().GPU()};
+    const auto cbufs = gpu.Maxwell3D().state.shader_stages[static_cast<u64>(stage)];
+    const auto cbuf_addr = gpu.MemoryManager().GpuToCpuAddress(
+        cbufs.const_buffers[global_region.GetCbufIndex()].address + global_region.GetCbufOffset());
+    ASSERT(cbuf_addr);
+
+    const auto actual_addr_gpu = Memory::Read64(*cbuf_addr);
+    const auto size = Memory::Read32(*cbuf_addr + 8);
+    const auto actual_addr = gpu.MemoryManager().GpuToCpuAddress(actual_addr_gpu);
+    ASSERT(actual_addr);
+
+    // Look up global region in the cache based on address
+    GlobalRegion region = TryGet(*actual_addr);
+
+    if (!region) {
+        // No global region found - create a new one
+        region = GetUncachedGlobalRegion(*actual_addr, size);
+        Register(region);
+    }
+
+    return region;
+}
+
 } // namespace OpenGL

src/video_core/renderer_opengl/gl_global_cache.h

@@ -5,9 +5,13 @@
 #pragma once
 
 #include <memory>
+#include <unordered_map>
+
 #include <glad/glad.h>
 
+#include "common/assert.h"
 #include "common/common_types.h"
+#include "video_core/engines/maxwell_3d.h"
 #include "video_core/rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
 
@@ -40,6 +44,9 @@ public:
         return buffer.handle;
     }
 
+    /// Reloads the global region from guest memory
+    void Reload(u32 size_);
+
     // TODO(Rodrigo): When global memory is written (STG), implement flushing
     void Flush() override {
         UNIMPLEMENTED();
@@ -55,6 +62,17 @@ private:
 class GlobalRegionCacheOpenGL final : public RasterizerCache<GlobalRegion> {
 public:
     explicit GlobalRegionCacheOpenGL(RasterizerOpenGL& rasterizer);
+
+    /// Gets the current specified shader stage program
+    GlobalRegion GetGlobalRegion(const GLShader::GlobalMemoryEntry& descriptor,
+                                 Tegra::Engines::Maxwell3D::Regs::ShaderStage stage);
+
+private:
+    GlobalRegion TryGetReservedGlobalRegion(VAddr addr, u32 size) const;
+    GlobalRegion GetUncachedGlobalRegion(VAddr addr, u32 size);
+    void ReserveGlobalRegion(const GlobalRegion& region);
+
+    std::unordered_map<VAddr, GlobalRegion> reserve;
 };
 
 } // namespace OpenGL

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -297,10 +297,7 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
     MICROPROFILE_SCOPE(OpenGL_Shader);
     auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
 
-    // Next available bindpoints to use when uploading the const buffers and textures to the GLSL
-    // shaders. The constbuffer bindpoint starts after the shader stage configuration bind points.
-    u32 current_constbuffer_bindpoint = Tegra::Engines::Maxwell3D::Regs::MaxShaderStage;
-    u32 current_texture_bindpoint = 0;
+    BaseBindings base_bindings;
     std::array<bool, Maxwell::NumClipDistances> clip_distances{};
 
     for (std::size_t index = 0; index < Maxwell::MaxShaderProgram; ++index) {
@@ -324,43 +321,35 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
         const GLintptr offset = buffer_cache.UploadHostMemory(
             &ubo, sizeof(ubo), static_cast<std::size_t>(uniform_buffer_alignment));
 
-        // Bind the buffer
-        glBindBufferRange(GL_UNIFORM_BUFFER, static_cast<GLuint>(stage), buffer_cache.GetHandle(),
-                          offset, static_cast<GLsizeiptr>(sizeof(ubo)));
+        // Bind the emulation info buffer
+        glBindBufferRange(GL_UNIFORM_BUFFER, base_bindings.cbuf, buffer_cache.GetHandle(), offset,
+                          static_cast<GLsizeiptr>(sizeof(ubo)));
 
         Shader shader{shader_cache.GetStageProgram(program)};
+        const auto [program_handle, next_bindings] =
+            shader->GetProgramHandle(primitive_mode, base_bindings);
 
         switch (program) {
         case Maxwell::ShaderProgram::VertexA:
-        case Maxwell::ShaderProgram::VertexB: {
-            shader_program_manager->UseProgrammableVertexShader(
-                shader->GetProgramHandle(primitive_mode));
+        case Maxwell::ShaderProgram::VertexB:
+            shader_program_manager->UseProgrammableVertexShader(program_handle);
             break;
-        }
-        case Maxwell::ShaderProgram::Geometry: {
-            shader_program_manager->UseProgrammableGeometryShader(
-                shader->GetProgramHandle(primitive_mode));
+        case Maxwell::ShaderProgram::Geometry:
+            shader_program_manager->UseProgrammableGeometryShader(program_handle);
             break;
-        }
-        case Maxwell::ShaderProgram::Fragment: {
-            shader_program_manager->UseProgrammableFragmentShader(
-                shader->GetProgramHandle(primitive_mode));
+        case Maxwell::ShaderProgram::Fragment:
+            shader_program_manager->UseProgrammableFragmentShader(program_handle);
             break;
-        }
         default:
             LOG_CRITICAL(HW_GPU, "Unimplemented shader index={}, enable={}, offset=0x{:08X}", index,
                          shader_config.enable.Value(), shader_config.offset);
             UNREACHABLE();
         }
 
-        // Configure the const buffers for this shader stage.
-        current_constbuffer_bindpoint =
-            SetupConstBuffers(static_cast<Maxwell::ShaderStage>(stage), shader, primitive_mode,
-                              current_constbuffer_bindpoint);
-
-        // Configure the textures for this shader stage.
-        current_texture_bindpoint = SetupTextures(static_cast<Maxwell::ShaderStage>(stage), shader,
-                                                  primitive_mode, current_texture_bindpoint);
+        const auto stage_enum = static_cast<Maxwell::ShaderStage>(stage);
+        SetupConstBuffers(stage_enum, shader, program_handle, base_bindings);
+        SetupGlobalRegions(stage_enum, shader, program_handle, base_bindings);
+        SetupTextures(stage_enum, shader, program_handle, base_bindings);
 
         // Workaround for Intel drivers.
         // When a clip distance is enabled but not set in the shader it crops parts of the screen
@@ -375,6 +364,8 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
             // VertexB was combined with VertexA, so we skip the VertexB iteration
             index++;
         }
+
+        base_bindings = next_bindings;
     }
 
     SyncClipEnabled(clip_distances);
@@ -486,9 +477,9 @@ void RasterizerOpenGL::UpdatePagesCachedCount(VAddr addr, u64 size, int delta) {
         cached_pages.add({pages_interval, delta});
 }
 
-void RasterizerOpenGL::ConfigureFramebuffers(OpenGLState& current_state, bool using_color_fb,
-                                             bool using_depth_fb, bool preserve_contents,
-                                             std::optional<std::size_t> single_color_target) {
+std::pair<bool, bool> RasterizerOpenGL::ConfigureFramebuffers(
+    OpenGLState& current_state, bool using_color_fb, bool using_depth_fb, bool preserve_contents,
+    std::optional<std::size_t> single_color_target) {
     MICROPROFILE_SCOPE(OpenGL_Framebuffer);
     const auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
     const auto& regs = gpu.regs;
@@ -500,7 +491,7 @@ void RasterizerOpenGL::ConfigureFramebuffers(OpenGLState& current_state, bool us
         // Only skip if the previous ConfigureFramebuffers call was from the same kind (multiple or
         // single color targets). This is done because the guest registers may not change but the
         // host framebuffer may contain different attachments
-        return;
+        return current_depth_stencil_usage;
     }
     current_framebuffer_config_state = fb_config_state;
 
@@ -570,12 +561,14 @@ void RasterizerOpenGL::ConfigureFramebuffers(OpenGLState& current_state, bool us
         depth_surface->MarkAsModified(true, res_cache);
 
         fbkey.zeta = depth_surface->Texture().handle;
-        fbkey.stencil_enable = regs.stencil_enable;
+        fbkey.stencil_enable = regs.stencil_enable &&
+                               depth_surface->GetSurfaceParams().type == SurfaceType::DepthStencil;
     }
 
     SetupCachedFramebuffer(fbkey, current_state);
-
     SyncViewport(current_state);
+
+    return current_depth_stencil_usage = {static_cast<bool>(depth_surface), fbkey.stencil_enable};
 }
 
 void RasterizerOpenGL::Clear() {
@@ -643,8 +636,8 @@ void RasterizerOpenGL::Clear() {
         return;
     }
 
-    ConfigureFramebuffers(clear_state, use_color, use_depth || use_stencil, false,
-                          regs.clear_buffers.RT.Value());
+    const auto [clear_depth, clear_stencil] = ConfigureFramebuffers(
+        clear_state, use_color, use_depth || use_stencil, false, regs.clear_buffers.RT.Value());
     if (regs.clear_flags.scissor) {
         SyncScissorTest(clear_state);
     }
@@ -659,11 +652,11 @@ void RasterizerOpenGL::Clear() {
         glClearBufferfv(GL_COLOR, regs.clear_buffers.RT, regs.clear_color);
     }
 
-    if (use_depth && use_stencil) {
+    if (clear_depth && clear_stencil) {
         glClearBufferfi(GL_DEPTH_STENCIL, 0, regs.clear_depth, regs.clear_stencil);
-    } else if (use_depth) {
+    } else if (clear_depth) {
         glClearBufferfv(GL_DEPTH, 0, &regs.clear_depth);
-    } else if (use_stencil) {
+    } else if (clear_stencil) {
         glClearBufferiv(GL_STENCIL, 0, &regs.clear_stencil);
     }
 }
@@ -790,11 +783,6 @@ bool RasterizerOpenGL::AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Regs
     return true;
 }
 
-bool RasterizerOpenGL::AccelerateFill(const void* config) {
-    UNREACHABLE();
-    return true;
-}
-
 bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& config,
                                          VAddr framebuffer_addr, u32 pixel_stride) {
     if (!framebuffer_addr) {
@@ -924,8 +912,9 @@ void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntr
     }
 }
 
-u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, Shader& shader,
-                                        GLenum primitive_mode, u32 current_bindpoint) {
+void RasterizerOpenGL::SetupConstBuffers(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage,
+                                         const Shader& shader, GLuint program_handle,
+                                         BaseBindings base_bindings) {
     MICROPROFILE_SCOPE(OpenGL_UBO);
     const auto& gpu = Core::System::GetInstance().GPU();
     const auto& maxwell3d = gpu.Maxwell3D();
@@ -965,7 +954,7 @@ u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, Shader& shad
             }
         } else {
             // Buffer is accessed directly, upload just what we use
-            size = used_buffer.GetSize() * sizeof(float);
+            size = used_buffer.GetSize();
         }
 
         // Align the actual size so it ends up being a multiple of vec4 to meet the OpenGL std140
@@ -973,75 +962,73 @@ u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, Shader& shad
         size = Common::AlignUp(size, sizeof(GLvec4));
         ASSERT_MSG(size <= MaxConstbufferSize, "Constbuffer too big");
 
-        GLintptr const_buffer_offset = buffer_cache.UploadMemory(
+        const GLintptr const_buffer_offset = buffer_cache.UploadMemory(
             buffer.address, size, static_cast<std::size_t>(uniform_buffer_alignment));
 
-        // Now configure the bindpoint of the buffer inside the shader
-        glUniformBlockBinding(shader->GetProgramHandle(primitive_mode),
-                              shader->GetProgramResourceIndex(used_buffer),
-                              current_bindpoint + bindpoint);
-
         // Prepare values for multibind
         bind_buffers[bindpoint] = buffer_cache.GetHandle();
         bind_offsets[bindpoint] = const_buffer_offset;
         bind_sizes[bindpoint] = size;
     }
 
-    glBindBuffersRange(GL_UNIFORM_BUFFER, current_bindpoint, static_cast<GLsizei>(entries.size()),
+    // The first binding is reserved for emulation values
+    const GLuint ubo_base_binding = base_bindings.cbuf + 1;
+    glBindBuffersRange(GL_UNIFORM_BUFFER, ubo_base_binding, static_cast<GLsizei>(entries.size()),
                        bind_buffers.data(), bind_offsets.data(), bind_sizes.data());
-
-    return current_bindpoint + static_cast<u32>(entries.size());
 }
 
-u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, Shader& shader,
-                                    GLenum primitive_mode, u32 current_unit) {
+void RasterizerOpenGL::SetupGlobalRegions(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage,
+                                          const Shader& shader, GLenum primitive_mode,
+                                          BaseBindings base_bindings) {
+    // TODO(Rodrigo): Use ARB_multi_bind here
+    const auto& entries = shader->GetShaderEntries().global_memory_entries;
+
+    for (u32 bindpoint = 0; bindpoint < static_cast<u32>(entries.size()); ++bindpoint) {
+        const auto& entry = entries[bindpoint];
+        const u32 current_bindpoint = base_bindings.gmem + bindpoint;
+        const auto& region = global_cache.GetGlobalRegion(entry, stage);
+
+        glBindBufferBase(GL_SHADER_STORAGE_BUFFER, current_bindpoint, region->GetBufferHandle());
+    }
+}
+
+void RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, const Shader& shader,
+                                     GLuint program_handle, BaseBindings base_bindings) {
     MICROPROFILE_SCOPE(OpenGL_Texture);
     const auto& gpu = Core::System::GetInstance().GPU();
     const auto& maxwell3d = gpu.Maxwell3D();
     const auto& entries = shader->GetShaderEntries().samplers;
 
-    ASSERT_MSG(current_unit + entries.size() <= std::size(state.texture_units),
+    ASSERT_MSG(base_bindings.sampler + 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];
-        const u32 current_bindpoint = current_unit + bindpoint;
-
-        // Bind the uniform to the sampler.
-
-        glProgramUniform1i(shader->GetProgramHandle(primitive_mode),
-                           shader->GetUniformLocation(entry), current_bindpoint);
+        const u32 current_bindpoint = base_bindings.sampler + bindpoint;
+        auto& unit = state.texture_units[current_bindpoint];
 
         const auto texture = maxwell3d.GetStageTexture(entry.GetStage(), entry.GetOffset());
-
         if (!texture.enabled) {
-            state.texture_units[current_bindpoint].texture = 0;
+            unit.texture = 0;
             continue;
         }
 
         texture_samplers[current_bindpoint].SyncWithConfig(texture.tsc);
+
         Surface surface = res_cache.GetTextureSurface(texture, entry);
         if (surface != nullptr) {
-            const GLuint handle =
+            unit.texture =
                 entry.IsArray() ? surface->TextureLayer().handle : surface->Texture().handle;
-            const GLenum target = entry.IsArray() ? surface->TargetLayer() : surface->Target();
-            state.texture_units[current_bindpoint].texture = handle;
-            state.texture_units[current_bindpoint].target = target;
-            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);
+            unit.target = entry.IsArray() ? surface->TargetLayer() : surface->Target();
+            unit.swizzle.r = MaxwellToGL::SwizzleSource(texture.tic.x_source);
+            unit.swizzle.g = MaxwellToGL::SwizzleSource(texture.tic.y_source);
+            unit.swizzle.b = MaxwellToGL::SwizzleSource(texture.tic.z_source);
+            unit.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 = 0;
+            unit.texture = 0;
         }
     }
-
-    return current_unit + static_cast<u32>(entries.size());
 }
 
 void RasterizerOpenGL::SyncViewport(OpenGLState& current_state) {

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -56,7 +56,6 @@ public:
     void FlushAndInvalidateRegion(VAddr addr, u64 size) override;
     bool AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Regs::Surface& src,
                                const Tegra::Engines::Fermi2D::Regs::Surface& dst) override;
-    bool AccelerateFill(const void* config) override;
     bool AccelerateDisplay(const Tegra::FramebufferConfig& config, VAddr framebuffer_addr,
                            u32 pixel_stride) override;
     bool AccelerateDrawBatch(bool is_indexed) override;
@@ -122,30 +121,25 @@ private:
      * @param using_depth_fb If true, configure the depth/stencil framebuffer.
      * @param preserve_contents If true, tries to preserve data from a previously used framebuffer.
      * @param single_color_target Specifies if a single color buffer target should be used.
+     * @returns If depth (first) or stencil (second) are being stored in the bound zeta texture
+     * (requires using_depth_fb to be true)
      */
-    void ConfigureFramebuffers(OpenGLState& current_state, bool use_color_fb = true,
-                               bool using_depth_fb = true, bool preserve_contents = true,
-                               std::optional<std::size_t> single_color_target = {});
+    std::pair<bool, bool> ConfigureFramebuffers(
+        OpenGLState& current_state, bool use_color_fb = true, bool using_depth_fb = true,
+        bool preserve_contents = true, std::optional<std::size_t> single_color_target = {});
 
-    /**
-     * Configures the current constbuffers to use for the draw command.
-     * @param stage The shader stage to configure buffers for.
-     * @param shader The shader object that contains the specified stage.
-     * @param current_bindpoint The offset at which to start counting new buffer bindpoints.
-     * @returns The next available bindpoint for use in the next shader stage.
-     */
-    u32 SetupConstBuffers(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, Shader& shader,
-                          GLenum primitive_mode, u32 current_bindpoint);
+    /// Configures the current constbuffers to use for the draw command.
+    void SetupConstBuffers(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, const Shader& shader,
+                           GLuint program_handle, BaseBindings base_bindings);
 
-    /**
-     * Configures the current textures to use for the draw command.
-     * @param stage The shader stage to configure textures for.
-     * @param shader The shader object that contains the specified stage.
-     * @param current_unit The offset at which to start counting unused texture units.
-     * @returns The next available bindpoint for use in the next shader stage.
-     */
-    u32 SetupTextures(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, Shader& shader,
-                      GLenum primitive_mode, u32 current_unit);
+    /// Configures the current global memory entries to use for the draw command.
+    void SetupGlobalRegions(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage,
+                            const Shader& shader, GLenum primitive_mode,
+                            BaseBindings base_bindings);
+
+    /// Configures the current textures to use for the draw command.
+    void SetupTextures(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, const Shader& shader,
+                       GLuint program_handle, BaseBindings base_bindings);
 
     /// Syncs the viewport and depth range to match the guest state
     void SyncViewport(OpenGLState& current_state);
@@ -221,6 +215,7 @@ private:
 
     std::map<FramebufferCacheKey, OGLFramebuffer> framebuffer_cache;
     FramebufferConfigState current_framebuffer_config_state;
+    std::pair<bool, bool> current_depth_stencil_usage{};
 
     std::array<SamplerInfo, Tegra::Engines::Maxwell3D::Regs::NumTextureSamplers> texture_samplers;
 

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -128,6 +128,7 @@ std::size_t SurfaceParams::InnerMemorySize(bool force_gl, bool layer_only,
     params.height = Common::AlignUp(config.tic.Height(), GetCompressionFactor(params.pixel_format));
     params.unaligned_height = config.tic.Height();
     params.target = SurfaceTargetFromTextureType(config.tic.texture_type);
+    params.identity = SurfaceClass::Uploaded;
 
     switch (params.target) {
     case SurfaceTarget::Texture1D:
@@ -167,6 +168,7 @@ std::size_t SurfaceParams::InnerMemorySize(bool force_gl, bool layer_only,
     }
 
     params.is_layered = SurfaceTargetIsLayered(params.target);
+    params.is_array = SurfaceTargetIsArray(params.target);
     params.max_mip_level = config.tic.max_mip_level + 1;
     params.rt = {};
 
@@ -194,6 +196,7 @@ std::size_t SurfaceParams::InnerMemorySize(bool force_gl, bool layer_only,
     params.height = config.height;
     params.unaligned_height = config.height;
     params.target = SurfaceTarget::Texture2D;
+    params.identity = SurfaceClass::RenderTarget;
     params.depth = 1;
     params.max_mip_level = 1;
     params.is_layered = false;
@@ -229,6 +232,7 @@ std::size_t SurfaceParams::InnerMemorySize(bool force_gl, bool layer_only,
     params.height = zeta_height;
     params.unaligned_height = zeta_height;
     params.target = SurfaceTarget::Texture2D;
+    params.identity = SurfaceClass::DepthBuffer;
     params.depth = 1;
     params.max_mip_level = 1;
     params.is_layered = false;
@@ -257,6 +261,7 @@ std::size_t SurfaceParams::InnerMemorySize(bool force_gl, bool layer_only,
     params.height = config.height;
     params.unaligned_height = config.height;
     params.target = SurfaceTarget::Texture2D;
+    params.identity = SurfaceClass::Copy;
     params.depth = 1;
     params.max_mip_level = 1;
     params.rt = {};
@@ -574,8 +579,7 @@ CachedSurface::CachedSurface(const SurfaceParams& params)
 
     ApplyTextureDefaults(SurfaceTargetToGL(params.target), params.max_mip_level);
 
-    LabelGLObject(GL_TEXTURE, texture.handle, params.addr,
-                  SurfaceParams::SurfaceTargetName(params.target));
+    OpenGL::LabelGLObject(GL_TEXTURE, texture.handle, params.addr, params.IdentityString());
 
     // Clamp size to mapped GPU memory region
     // TODO(bunnei): Super Mario Odyssey maps a 0x40000 byte region and then uses it for a 0x80000
@@ -730,7 +734,6 @@ void CachedSurface::FlushGLBuffer() {
     glPixelStorei(GL_PACK_ROW_LENGTH, 0);
     ConvertFormatAsNeeded_FlushGLBuffer(gl_buffer[0], params.pixel_format, params.width,
                                         params.height);
-    ASSERT(params.type != SurfaceType::Fill);
     const u8* const texture_src_data = Memory::GetPointer(params.addr);
     ASSERT(texture_src_data);
     if (params.is_tiled) {
@@ -877,10 +880,13 @@ void CachedSurface::EnsureTextureView() {
     UNIMPLEMENTED_IF(gl_is_compressed);
 
     const GLenum target{TargetLayer()};
+    const GLuint num_layers{target == GL_TEXTURE_CUBE_MAP_ARRAY ? 6u : 1u};
+    constexpr GLuint min_layer = 0;
+    constexpr GLuint min_level = 0;
 
     texture_view.Create();
-    glTextureView(texture_view.handle, target, texture.handle, gl_internal_format, 0,
-                  params.max_mip_level, 0, 1);
+    glTextureView(texture_view.handle, target, texture.handle, gl_internal_format, min_level,
+                  params.max_mip_level, min_layer, num_layers);
 
     OpenGLState cur_state = OpenGLState::GetCurState();
     const auto& old_tex = cur_state.texture_units[0];
@@ -897,9 +903,6 @@ void CachedSurface::EnsureTextureView() {
 
 MICROPROFILE_DEFINE(OpenGL_TextureUL, "OpenGL", "Texture Upload", MP_RGB(128, 192, 64));
 void CachedSurface::UploadGLTexture(GLuint read_fb_handle, GLuint draw_fb_handle) {
-    if (params.type == SurfaceType::Fill)
-        return;
-
     MICROPROFILE_SCOPE(OpenGL_TextureUL);
 
     for (u32 i = 0; i < params.max_mip_level; i++)

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -35,6 +35,14 @@ using PixelFormat = VideoCore::Surface::PixelFormat;
 using ComponentType = VideoCore::Surface::ComponentType;
 
 struct SurfaceParams {
+
+    enum class SurfaceClass {
+        Uploaded,
+        RenderTarget,
+        DepthBuffer,
+        Copy,
+    };
+
     static std::string SurfaceTargetName(SurfaceTarget target) {
         switch (target) {
         case SurfaceTarget::Texture1D:
@@ -210,6 +218,48 @@ struct SurfaceParams {
     /// Initializes parameters for caching, should be called after everything has been initialized
     void InitCacheParameters(Tegra::GPUVAddr gpu_addr);
 
+    std::string TargetName() const {
+        switch (target) {
+        case SurfaceTarget::Texture1D:
+            return "1D";
+        case SurfaceTarget::Texture2D:
+            return "2D";
+        case SurfaceTarget::Texture3D:
+            return "3D";
+        case SurfaceTarget::Texture1DArray:
+            return "1DArray";
+        case SurfaceTarget::Texture2DArray:
+            return "2DArray";
+        case SurfaceTarget::TextureCubemap:
+            return "Cube";
+        default:
+            LOG_CRITICAL(HW_GPU, "Unimplemented surface_target={}", static_cast<u32>(target));
+            UNREACHABLE();
+            return fmt::format("TUK({})", static_cast<u32>(target));
+        }
+    }
+
+    std::string ClassName() const {
+        switch (identity) {
+        case SurfaceClass::Uploaded:
+            return "UP";
+        case SurfaceClass::RenderTarget:
+            return "RT";
+        case SurfaceClass::DepthBuffer:
+            return "DB";
+        case SurfaceClass::Copy:
+            return "CP";
+        default:
+            LOG_CRITICAL(HW_GPU, "Unimplemented surface_class={}", static_cast<u32>(identity));
+            UNREACHABLE();
+            return fmt::format("CUK({})", static_cast<u32>(identity));
+        }
+    }
+
+    std::string IdentityString() const {
+        return ClassName() + '_' + TargetName() + '_' + (is_tiled ? 'T' : 'L');
+    }
+
     bool is_tiled;
     u32 block_width;
     u32 block_height;
@@ -223,8 +273,10 @@ struct SurfaceParams {
     u32 depth;
     u32 unaligned_height;
     SurfaceTarget target;
+    SurfaceClass identity;
     u32 max_mip_level;
     bool is_layered;
+    bool is_array;
     bool srgb_conversion;
     // Parameters used for caching
     VAddr addr;
@@ -255,6 +307,7 @@ struct SurfaceReserveKey : Common::HashableStruct<OpenGL::SurfaceParams> {
     static SurfaceReserveKey Create(const OpenGL::SurfaceParams& params) {
         SurfaceReserveKey res;
         res.state = params;
+        res.state.identity = {}; // Ignore the origin of the texture
         res.state.gpu_addr = {}; // Ignore GPU vaddr in caching
         res.state.rt = {};       // Ignore rt config in caching
         return res;
@@ -294,7 +347,7 @@ public:
     }
 
     const OGLTexture& TextureLayer() {
-        if (params.is_layered) {
+        if (params.is_array) {
             return Texture();
         }
         EnsureTextureView();

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -34,36 +34,25 @@ static ProgramCode GetShaderCode(VAddr addr) {
     return program_code;
 }
 
-/// Helper function to set shader uniform block bindings for a single shader stage
-static void SetShaderUniformBlockBinding(GLuint shader, const char* name,
-                                         Maxwell::ShaderStage binding, std::size_t expected_size) {
-    const GLuint ub_index = glGetUniformBlockIndex(shader, name);
-    if (ub_index == GL_INVALID_INDEX) {
-        return;
+/// Gets the shader type from a Maxwell program type
+constexpr GLenum GetShaderType(Maxwell::ShaderProgram program_type) {
+    switch (program_type) {
+    case Maxwell::ShaderProgram::VertexA:
+    case Maxwell::ShaderProgram::VertexB:
+        return GL_VERTEX_SHADER;
+    case Maxwell::ShaderProgram::Geometry:
+        return GL_GEOMETRY_SHADER;
+    case Maxwell::ShaderProgram::Fragment:
+        return GL_FRAGMENT_SHADER;
+    default:
+        return GL_NONE;
     }
-
-    GLint ub_size = 0;
-    glGetActiveUniformBlockiv(shader, ub_index, GL_UNIFORM_BLOCK_DATA_SIZE, &ub_size);
-    ASSERT_MSG(static_cast<std::size_t>(ub_size) == expected_size,
-               "Uniform block size did not match! Got {}, expected {}", ub_size, expected_size);
-    glUniformBlockBinding(shader, ub_index, static_cast<GLuint>(binding));
-}
-
-/// Sets shader uniform block bindings for an entire shader program
-static void SetShaderUniformBlockBindings(GLuint shader) {
-    SetShaderUniformBlockBinding(shader, "vs_config", Maxwell::ShaderStage::Vertex,
-                                 sizeof(GLShader::MaxwellUniformData));
-    SetShaderUniformBlockBinding(shader, "gs_config", Maxwell::ShaderStage::Geometry,
-                                 sizeof(GLShader::MaxwellUniformData));
-    SetShaderUniformBlockBinding(shader, "fs_config", Maxwell::ShaderStage::Fragment,
-                                 sizeof(GLShader::MaxwellUniformData));
 }
 
 CachedShader::CachedShader(VAddr addr, Maxwell::ShaderProgram program_type)
     : addr{addr}, program_type{program_type}, setup{GetShaderCode(addr)} {
 
     GLShader::ProgramResult program_result;
-    GLenum gl_type{};
 
     switch (program_type) {
     case Maxwell::ShaderProgram::VertexA:
@@ -74,17 +63,14 @@ CachedShader::CachedShader(VAddr addr, Maxwell::ShaderProgram program_type)
     case Maxwell::ShaderProgram::VertexB:
         CalculateProperties();
         program_result = GLShader::GenerateVertexShader(setup);
-        gl_type = GL_VERTEX_SHADER;
         break;
     case Maxwell::ShaderProgram::Geometry:
         CalculateProperties();
         program_result = GLShader::GenerateGeometryShader(setup);
-        gl_type = GL_GEOMETRY_SHADER;
         break;
     case Maxwell::ShaderProgram::Fragment:
         CalculateProperties();
         program_result = GLShader::GenerateFragmentShader(setup);
-        gl_type = GL_FRAGMENT_SHADER;
         break;
     default:
         LOG_CRITICAL(HW_GPU, "Unimplemented program_type={}", static_cast<u32>(program_type));
@@ -92,59 +78,105 @@ CachedShader::CachedShader(VAddr addr, Maxwell::ShaderProgram program_type)
         return;
     }
 
+    code = program_result.first;
     entries = program_result.second;
     shader_length = entries.shader_length;
+}
 
-    if (program_type != Maxwell::ShaderProgram::Geometry) {
-        OGLShader shader;
-        shader.Create(program_result.first.c_str(), gl_type);
-        program.Create(true, shader.handle);
-        SetShaderUniformBlockBindings(program.handle);
-        LabelGLObject(GL_PROGRAM, program.handle, addr);
+std::tuple<GLuint, BaseBindings> CachedShader::GetProgramHandle(GLenum primitive_mode,
+                                                                BaseBindings base_bindings) {
+    GLuint handle{};
+    if (program_type == Maxwell::ShaderProgram::Geometry) {
+        handle = GetGeometryShader(primitive_mode, base_bindings);
     } else {
-        // Store shader's code to lazily build it on draw
-        geometry_programs.code = program_result.first;
+        const auto [entry, is_cache_miss] = programs.try_emplace(base_bindings);
+        auto& program = entry->second;
+        if (is_cache_miss) {
+            std::string source = AllocateBindings(base_bindings);
+            source += code;
+
+            OGLShader shader;
+            shader.Create(source.c_str(), GetShaderType(program_type));
+            program.Create(true, shader.handle);
+            LabelGLObject(GL_PROGRAM, program.handle, addr);
+        }
+
+        handle = program.handle;
+    }
+
+    // Add const buffer and samplers offset reserved by this shader. One UBO binding is reserved for
+    // emulation values
+    base_bindings.cbuf += static_cast<u32>(entries.const_buffers.size()) + 1;
+    base_bindings.gmem += static_cast<u32>(entries.global_memory_entries.size());
+    base_bindings.sampler += static_cast<u32>(entries.samplers.size());
+
+    return {handle, base_bindings};
+}
+
+std::string CachedShader::AllocateBindings(BaseBindings base_bindings) {
+    std::string code = "#version 430 core\n";
+    code += fmt::format("#define EMULATION_UBO_BINDING {}\n", base_bindings.cbuf++);
+
+    for (const auto& cbuf : entries.const_buffers) {
+        code += fmt::format("#define CBUF_BINDING_{} {}\n", cbuf.GetIndex(), base_bindings.cbuf++);
+    }
+
+    for (const auto& gmem : entries.global_memory_entries) {
+        code += fmt::format("#define GMEM_BINDING_{}_{} {}\n", gmem.GetCbufIndex(),
+                            gmem.GetCbufOffset(), base_bindings.gmem++);
+    }
+
+    for (const auto& sampler : entries.samplers) {
+        code += fmt::format("#define SAMPLER_BINDING_{} {}\n", sampler.GetIndex(),
+                            base_bindings.sampler++);
+    }
+
+    return code;
+}
+
+GLuint CachedShader::GetGeometryShader(GLenum primitive_mode, BaseBindings base_bindings) {
+    const auto [entry, is_cache_miss] = geometry_programs.try_emplace(base_bindings);
+    auto& programs = entry->second;
+
+    switch (primitive_mode) {
+    case GL_POINTS:
+        return LazyGeometryProgram(programs.points, base_bindings, "points", 1, "ShaderPoints");
+    case GL_LINES:
+    case GL_LINE_STRIP:
+        return LazyGeometryProgram(programs.lines, base_bindings, "lines", 2, "ShaderLines");
+    case GL_LINES_ADJACENCY:
+    case GL_LINE_STRIP_ADJACENCY:
+        return LazyGeometryProgram(programs.lines_adjacency, base_bindings, "lines_adjacency", 4,
+                                   "ShaderLinesAdjacency");
+    case GL_TRIANGLES:
+    case GL_TRIANGLE_STRIP:
+    case GL_TRIANGLE_FAN:
+        return LazyGeometryProgram(programs.triangles, base_bindings, "triangles", 3,
+                                   "ShaderTriangles");
+    case GL_TRIANGLES_ADJACENCY:
+    case GL_TRIANGLE_STRIP_ADJACENCY:
+        return LazyGeometryProgram(programs.triangles_adjacency, base_bindings,
+                                   "triangles_adjacency", 6, "ShaderTrianglesAdjacency");
+    default:
+        UNREACHABLE_MSG("Unknown primitive mode.");
+        return LazyGeometryProgram(programs.points, base_bindings, "points", 1, "ShaderPoints");
     }
 }
 
-GLuint CachedShader::GetProgramResourceIndex(const GLShader::ConstBufferEntry& buffer) {
-    const auto search{resource_cache.find(buffer.GetHash())};
-    if (search == resource_cache.end()) {
-        const GLuint index{
-            glGetProgramResourceIndex(program.handle, GL_UNIFORM_BLOCK, buffer.GetName().c_str())};
-        resource_cache[buffer.GetHash()] = index;
-        return index;
-    }
-
-    return search->second;
-}
-
-GLint CachedShader::GetUniformLocation(const GLShader::SamplerEntry& sampler) {
-    const auto search{uniform_cache.find(sampler.GetHash())};
-    if (search == uniform_cache.end()) {
-        const GLint index{glGetUniformLocation(program.handle, sampler.GetName().c_str())};
-        uniform_cache[sampler.GetHash()] = index;
-        return index;
-    }
-
-    return search->second;
-}
-
-GLuint CachedShader::LazyGeometryProgram(OGLProgram& target_program,
+GLuint CachedShader::LazyGeometryProgram(OGLProgram& target_program, BaseBindings base_bindings,
                                          const std::string& glsl_topology, u32 max_vertices,
                                          const std::string& debug_name) {
     if (target_program.handle != 0) {
         return target_program.handle;
     }
-    std::string source = "#version 430 core\n";
+    std::string source = AllocateBindings(base_bindings);
     source += "layout (" + glsl_topology + ") in;\n";
     source += "#define MAX_VERTEX_INPUT " + std::to_string(max_vertices) + '\n';
-    source += geometry_programs.code;
+    source += code;
 
     OGLShader shader;
     shader.Create(source.c_str(), GL_GEOMETRY_SHADER);
     target_program.Create(true, shader.handle);
-    SetShaderUniformBlockBindings(target_program.handle);
     LabelGLObject(GL_PROGRAM, target_program.handle, addr, debug_name);
     return target_program.handle;
 };

src/video_core/renderer_opengl/gl_shader_cache.h

@@ -7,6 +7,9 @@
 #include <array>
 #include <map>
 #include <memory>
+#include <tuple>
+
+#include <glad/glad.h>
 
 #include "common/assert.h"
 #include "common/common_types.h"
@@ -23,6 +26,16 @@ class RasterizerOpenGL;
 using Shader = std::shared_ptr<CachedShader>;
 using Maxwell = Tegra::Engines::Maxwell3D::Regs;
 
+struct BaseBindings {
+    u32 cbuf{};
+    u32 gmem{};
+    u32 sampler{};
+
+    bool operator<(const BaseBindings& rhs) const {
+        return std::tie(cbuf, gmem, sampler) < std::tie(rhs.cbuf, rhs.gmem, rhs.sampler);
+    }
+};
+
 class CachedShader final : public RasterizerCacheObject {
 public:
     CachedShader(VAddr addr, Maxwell::ShaderProgram program_type);
@@ -44,70 +57,45 @@ public:
     }
 
     /// Gets the GL program handle for the shader
-    GLuint GetProgramHandle(GLenum primitive_mode) {
-        if (program_type != Maxwell::ShaderProgram::Geometry) {
-            return program.handle;
-        }
-        switch (primitive_mode) {
-        case GL_POINTS:
-            return LazyGeometryProgram(geometry_programs.points, "points", 1, "ShaderPoints");
-        case GL_LINES:
-        case GL_LINE_STRIP:
-            return LazyGeometryProgram(geometry_programs.lines, "lines", 2, "ShaderLines");
-        case GL_LINES_ADJACENCY:
-        case GL_LINE_STRIP_ADJACENCY:
-            return LazyGeometryProgram(geometry_programs.lines_adjacency, "lines_adjacency", 4,
-                                       "ShaderLinesAdjacency");
-        case GL_TRIANGLES:
-        case GL_TRIANGLE_STRIP:
-        case GL_TRIANGLE_FAN:
-            return LazyGeometryProgram(geometry_programs.triangles, "triangles", 3,
-                                       "ShaderTriangles");
-        case GL_TRIANGLES_ADJACENCY:
-        case GL_TRIANGLE_STRIP_ADJACENCY:
-            return LazyGeometryProgram(geometry_programs.triangles_adjacency, "triangles_adjacency",
-                                       6, "ShaderTrianglesAdjacency");
-        default:
-            UNREACHABLE_MSG("Unknown primitive mode.");
-            return LazyGeometryProgram(geometry_programs.points, "points", 1, "ShaderPoints");
-        }
-    }
-
-    /// Gets the GL program resource location for the specified resource, caching as needed
-    GLuint GetProgramResourceIndex(const GLShader::ConstBufferEntry& buffer);
-
-    /// Gets the GL uniform location for the specified resource, caching as needed
-    GLint GetUniformLocation(const GLShader::SamplerEntry& sampler);
+    std::tuple<GLuint, BaseBindings> GetProgramHandle(GLenum primitive_mode,
+                                                      BaseBindings base_bindings);
 
 private:
-    /// Generates a geometry shader or returns one that already exists.
-    GLuint LazyGeometryProgram(OGLProgram& target_program, const std::string& glsl_topology,
-                               u32 max_vertices, const std::string& debug_name);
-
-    void CalculateProperties();
-
-    VAddr addr;
-    std::size_t shader_length;
-    Maxwell::ShaderProgram program_type;
-    GLShader::ShaderSetup setup;
-    GLShader::ShaderEntries entries;
-
-    // Non-geometry program.
-    OGLProgram program;
-
     // Geometry programs. These are needed because GLSL needs an input topology but it's not
     // declared by the hardware. Workaround this issue by generating a different shader per input
     // topology class.
-    struct {
-        std::string code;
+    struct GeometryPrograms {
         OGLProgram points;
         OGLProgram lines;
         OGLProgram lines_adjacency;
         OGLProgram triangles;
         OGLProgram triangles_adjacency;
-    } geometry_programs;
+    };
 
-    std::map<u32, GLuint> resource_cache;
+    std::string AllocateBindings(BaseBindings base_bindings);
+
+    GLuint GetGeometryShader(GLenum primitive_mode, BaseBindings base_bindings);
+
+    /// Generates a geometry shader or returns one that already exists.
+    GLuint LazyGeometryProgram(OGLProgram& target_program, BaseBindings base_bindings,
+                               const std::string& glsl_topology, u32 max_vertices,
+                               const std::string& debug_name);
+
+    void CalculateProperties();
+
+    VAddr addr{};
+    std::size_t shader_length{};
+    Maxwell::ShaderProgram program_type{};
+    GLShader::ShaderSetup setup;
+    GLShader::ShaderEntries entries;
+
+    std::string code;
+
+    std::map<BaseBindings, OGLProgram> programs;
+    std::map<BaseBindings, GeometryPrograms> geometry_programs;
+
+    std::map<u32, GLuint> cbuf_resource_cache;
+    std::map<u32, GLuint> gmem_resource_cache;
     std::map<u32, GLint> uniform_cache;
 };
 

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -34,6 +34,8 @@ using Operation = const OperationNode&;
 enum : u32 { POSITION_VARYING_LOCATION = 0, GENERIC_VARYING_START_LOCATION = 1 };
 constexpr u32 MAX_CONSTBUFFER_ELEMENTS =
     static_cast<u32>(RasterizerOpenGL::MaxConstbufferSize) / (4 * sizeof(float));
+constexpr u32 MAX_GLOBALMEMORY_ELEMENTS =
+    static_cast<u32>(RasterizerOpenGL::MaxGlobalMemorySize) / sizeof(float);
 
 enum class Type { Bool, Bool2, Float, Int, Uint, HalfFloat };
 
@@ -143,6 +145,7 @@ public:
         DeclareInputAttributes();
         DeclareOutputAttributes();
         DeclareConstantBuffers();
+        DeclareGlobalMemory();
         DeclareSamplers();
 
         code.AddLine("void execute_" + suffix + "() {");
@@ -168,7 +171,7 @@ public:
             code.AddLine(fmt::format("case 0x{:x}u: {{", address));
             ++code.scope;
 
-            VisitBasicBlock(bb);
+            VisitBlock(bb);
 
             --code.scope;
             code.AddLine('}');
@@ -190,12 +193,15 @@ public:
     ShaderEntries GetShaderEntries() const {
         ShaderEntries entries;
         for (const auto& cbuf : ir.GetConstantBuffers()) {
-            ConstBufferEntry desc(cbuf.second, stage, GetConstBufferBlock(cbuf.first), cbuf.first);
-            entries.const_buffers.push_back(desc);
+            entries.const_buffers.emplace_back(cbuf.second, stage, GetConstBufferBlock(cbuf.first),
+                                               cbuf.first);
         }
         for (const auto& sampler : ir.GetSamplers()) {
-            SamplerEntry desc(sampler, stage, GetSampler(sampler));
-            entries.samplers.push_back(desc);
+            entries.samplers.emplace_back(sampler, stage, GetSampler(sampler));
+        }
+        for (const auto& gmem : ir.GetGlobalMemoryBases()) {
+            entries.global_memory_entries.emplace_back(gmem.cbuf_index, gmem.cbuf_offset, stage,
+                                                       GetGlobalMemoryBlock(gmem));
         }
         entries.clip_distances = ir.GetClipDistances();
         entries.shader_length = ir.GetLength();
@@ -368,13 +374,26 @@ private:
     void DeclareConstantBuffers() {
         for (const auto& entry : ir.GetConstantBuffers()) {
             const auto [index, size] = entry;
-            code.AddLine("layout (std140) uniform " + GetConstBufferBlock(index) + " {");
+            code.AddLine("layout (std140, binding = CBUF_BINDING_" + std::to_string(index) +
+                         ") uniform " + GetConstBufferBlock(index) + " {");
             code.AddLine("    vec4 " + GetConstBuffer(index) + "[MAX_CONSTBUFFER_ELEMENTS];");
             code.AddLine("};");
             code.AddNewLine();
         }
     }
 
+    void DeclareGlobalMemory() {
+        for (const auto& entry : ir.GetGlobalMemoryBases()) {
+            const std::string binding =
+                fmt::format("GMEM_BINDING_{}_{}", entry.cbuf_index, entry.cbuf_offset);
+            code.AddLine("layout (std430, binding = " + binding + ") buffer " +
+                         GetGlobalMemoryBlock(entry) + " {");
+            code.AddLine("    float " + GetGlobalMemory(entry) + "[MAX_GLOBALMEMORY_ELEMENTS];");
+            code.AddLine("};");
+            code.AddNewLine();
+        }
+    }
+
     void DeclareSamplers() {
         const auto& samplers = ir.GetSamplers();
         for (const auto& sampler : samplers) {
@@ -398,13 +417,14 @@ private:
             if (sampler.IsShadow())
                 sampler_type += "Shadow";
 
-            code.AddLine("uniform " + sampler_type + ' ' + GetSampler(sampler) + ';');
+            code.AddLine("layout (binding = SAMPLER_BINDING_" + std::to_string(sampler.GetIndex()) +
+                         ") uniform " + sampler_type + ' ' + GetSampler(sampler) + ';');
         }
         if (!samplers.empty())
             code.AddNewLine();
     }
 
-    void VisitBasicBlock(const BasicBlock& bb) {
+    void VisitBlock(const NodeBlock& bb) {
         for (const Node node : bb) {
             if (const std::string expr = Visit(node); !expr.empty()) {
                 code.AddLine(expr);
@@ -523,8 +543,9 @@ private:
             if (const auto immediate = std::get_if<ImmediateNode>(offset)) {
                 // Direct access
                 const u32 offset_imm = immediate->GetValue();
-                return fmt::format("{}[{}][{}]", GetConstBuffer(cbuf->GetIndex()), offset_imm / 4,
-                                   offset_imm % 4);
+                ASSERT_MSG(offset_imm % 4 == 0, "Unaligned cbuf direct access");
+                return fmt::format("{}[{}][{}]", GetConstBuffer(cbuf->GetIndex()),
+                                   offset_imm / (4 * 4), (offset_imm / 4) % 4);
 
             } else if (std::holds_alternative<OperationNode>(*offset)) {
                 // Indirect access
@@ -538,6 +559,12 @@ private:
                 UNREACHABLE_MSG("Unmanaged offset node type");
             }
 
+        } else if (const auto gmem = std::get_if<GmemNode>(node)) {
+            const std::string real = Visit(gmem->GetRealAddress());
+            const std::string base = Visit(gmem->GetBaseAddress());
+            const std::string final_offset = "(ftou(" + real + ") - ftou(" + base + ")) / 4";
+            return fmt::format("{}[{}]", GetGlobalMemory(gmem->GetDescriptor()), final_offset);
+
         } else if (const auto lmem = std::get_if<LmemNode>(node)) {
             return fmt::format("{}[ftou({}) / 4]", GetLocalMemory(), Visit(lmem->GetAddress()));
 
@@ -549,7 +576,7 @@ private:
             code.AddLine("if (" + Visit(conditional->GetCondition()) + ") {");
             ++code.scope;
 
-            VisitBasicBlock(conditional->GetCode());
+            VisitBlock(conditional->GetCode());
 
             --code.scope;
             code.AddLine('}');
@@ -1471,6 +1498,15 @@ private:
         return GetDeclarationWithSuffix(index, "cbuf");
     }
 
+    std::string GetGlobalMemory(const GlobalMemoryBase& descriptor) const {
+        return fmt::format("gmem_{}_{}_{}", descriptor.cbuf_index, descriptor.cbuf_offset, suffix);
+    }
+
+    std::string GetGlobalMemoryBlock(const GlobalMemoryBase& descriptor) const {
+        return fmt::format("gmem_block_{}_{}_{}", descriptor.cbuf_index, descriptor.cbuf_offset,
+                           suffix);
+    }
+
     std::string GetConstBufferBlock(u32 index) const {
         return GetDeclarationWithSuffix(index, "cbuf_block");
     }
@@ -1505,8 +1541,10 @@ private:
 };
 
 std::string GetCommonDeclarations() {
-    return "#define MAX_CONSTBUFFER_ELEMENTS " + std::to_string(MAX_CONSTBUFFER_ELEMENTS) +
-           "\n"
+    const auto cbuf = std::to_string(MAX_CONSTBUFFER_ELEMENTS);
+    const auto gmem = std::to_string(MAX_GLOBALMEMORY_ELEMENTS);
+    return "#define MAX_CONSTBUFFER_ELEMENTS " + cbuf + "\n" +
+           "#define MAX_GLOBALMEMORY_ELEMENTS " + gmem + "\n" +
            "#define ftoi floatBitsToInt\n"
            "#define ftou floatBitsToUint\n"
            "#define itof intBitsToFloat\n"

src/video_core/renderer_opengl/gl_shader_decompiler.h

@@ -38,10 +38,6 @@ public:
         return index;
     }
 
-    u32 GetHash() const {
-        return (static_cast<u32>(stage) << 16) | index;
-    }
-
 private:
     std::string name;
     Maxwell::ShaderStage stage{};
@@ -62,18 +58,44 @@ public:
         return stage;
     }
 
-    u32 GetHash() const {
-        return (static_cast<u32>(stage) << 16) | static_cast<u32>(GetIndex());
-    }
-
 private:
     std::string name;
     Maxwell::ShaderStage stage{};
 };
 
+class GlobalMemoryEntry {
+public:
+    explicit GlobalMemoryEntry(u32 cbuf_index, u32 cbuf_offset, Maxwell::ShaderStage stage,
+                               std::string name)
+        : cbuf_index{cbuf_index}, cbuf_offset{cbuf_offset}, stage{stage}, name{std::move(name)} {}
+
+    u32 GetCbufIndex() const {
+        return cbuf_index;
+    }
+
+    u32 GetCbufOffset() const {
+        return cbuf_offset;
+    }
+
+    const std::string& GetName() const {
+        return name;
+    }
+
+    Maxwell::ShaderStage GetStage() const {
+        return stage;
+    }
+
+private:
+    u32 cbuf_index{};
+    u32 cbuf_offset{};
+    Maxwell::ShaderStage stage{};
+    std::string name;
+};
+
 struct ShaderEntries {
     std::vector<ConstBufferEntry> const_buffers;
     std::vector<SamplerEntry> samplers;
+    std::vector<GlobalMemoryEntry> global_memory_entries;
     std::array<bool, Maxwell::NumClipDistances> clip_distances{};
     std::size_t shader_length{};
 };

src/video_core/renderer_opengl/gl_shader_gen.cpp

@@ -20,15 +20,14 @@ static constexpr u32 PROGRAM_OFFSET{10};
 ProgramResult GenerateVertexShader(const ShaderSetup& setup) {
     const std::string id = fmt::format("{:016x}", setup.program.unique_identifier);
 
-    std::string out = "#version 430 core\n";
-    out += "#extension GL_ARB_separate_shader_objects : enable\n\n";
+    std::string out = "#extension GL_ARB_separate_shader_objects : enable\n\n";
     out += "// Shader Unique Id: VS" + id + "\n\n";
     out += GetCommonDeclarations();
 
     out += R"(
 layout (location = 0) out vec4 position;
 
-layout(std140) uniform vs_config {
+layout (std140, binding = EMULATION_UBO_BINDING) uniform vs_config {
     vec4 viewport_flip;
     uvec4 config_pack; // instance_id, flip_stage, y_direction, padding
     uvec4 alpha_test;
@@ -78,7 +77,6 @@ void main() {
 }
 
 ProgramResult GenerateGeometryShader(const ShaderSetup& setup) {
-    // Version is intentionally skipped in shader generation, it's added by the lazy compilation.
     const std::string id = fmt::format("{:016x}", setup.program.unique_identifier);
 
     std::string out = "#extension GL_ARB_separate_shader_objects : enable\n\n";
@@ -89,7 +87,7 @@ ProgramResult GenerateGeometryShader(const ShaderSetup& setup) {
 layout (location = 0) in vec4 gs_position[];
 layout (location = 0) out vec4 position;
 
-layout (std140) uniform gs_config {
+layout (std140, binding = EMULATION_UBO_BINDING) uniform gs_config {
     vec4 viewport_flip;
     uvec4 config_pack; // instance_id, flip_stage, y_direction, padding
     uvec4 alpha_test;
@@ -112,8 +110,7 @@ void main() {
 ProgramResult GenerateFragmentShader(const ShaderSetup& setup) {
     const std::string id = fmt::format("{:016x}", setup.program.unique_identifier);
 
-    std::string out = "#version 430 core\n";
-    out += "#extension GL_ARB_separate_shader_objects : enable\n\n";
+    std::string out = "#extension GL_ARB_separate_shader_objects : enable\n\n";
     out += "// Shader Unique Id: FS" + id + "\n\n";
     out += GetCommonDeclarations();
 
@@ -129,7 +126,7 @@ layout (location = 7) out vec4 FragColor7;
 
 layout (location = 0) in vec4 position;
 
-layout (std140) uniform fs_config {
+layout (std140, binding = EMULATION_UBO_BINDING) uniform fs_config {
     vec4 viewport_flip;
     uvec4 config_pack; // instance_id, flip_stage, y_direction, padding
     uvec4 alpha_test;

src/video_core/shader/decode.cpp

@@ -121,15 +121,15 @@ ExitMethod ShaderIR::Scan(u32 begin, u32 end, std::set<u32>& labels) {
     return exit_method = ExitMethod::AlwaysReturn;
 }
 
-BasicBlock ShaderIR::DecodeRange(u32 begin, u32 end) {
-    BasicBlock basic_block;
+NodeBlock ShaderIR::DecodeRange(u32 begin, u32 end) {
+    NodeBlock basic_block;
     for (u32 pc = begin; pc < (begin > end ? MAX_PROGRAM_LENGTH : end);) {
         pc = DecodeInstr(basic_block, pc);
     }
-    return std::move(basic_block);
+    return basic_block;
 }
 
-u32 ShaderIR::DecodeInstr(BasicBlock& bb, u32 pc) {
+u32 ShaderIR::DecodeInstr(NodeBlock& bb, u32 pc) {
     // Ignore sched instructions when generating code.
     if (IsSchedInstruction(pc, main_offset)) {
         return pc + 1;
@@ -151,39 +151,38 @@ u32 ShaderIR::DecodeInstr(BasicBlock& bb, u32 pc) {
     UNIMPLEMENTED_IF_MSG(instr.pred.full_pred == Pred::NeverExecute,
                          "NeverExecute predicate not implemented");
 
-    static const std::map<OpCode::Type, u32 (ShaderIR::*)(BasicBlock&, const BasicBlock&, u32)>
-        decoders = {
-            {OpCode::Type::Arithmetic, &ShaderIR::DecodeArithmetic},
-            {OpCode::Type::ArithmeticImmediate, &ShaderIR::DecodeArithmeticImmediate},
-            {OpCode::Type::Bfe, &ShaderIR::DecodeBfe},
-            {OpCode::Type::Bfi, &ShaderIR::DecodeBfi},
-            {OpCode::Type::Shift, &ShaderIR::DecodeShift},
-            {OpCode::Type::ArithmeticInteger, &ShaderIR::DecodeArithmeticInteger},
-            {OpCode::Type::ArithmeticIntegerImmediate, &ShaderIR::DecodeArithmeticIntegerImmediate},
-            {OpCode::Type::ArithmeticHalf, &ShaderIR::DecodeArithmeticHalf},
-            {OpCode::Type::ArithmeticHalfImmediate, &ShaderIR::DecodeArithmeticHalfImmediate},
-            {OpCode::Type::Ffma, &ShaderIR::DecodeFfma},
-            {OpCode::Type::Hfma2, &ShaderIR::DecodeHfma2},
-            {OpCode::Type::Conversion, &ShaderIR::DecodeConversion},
-            {OpCode::Type::Memory, &ShaderIR::DecodeMemory},
-            {OpCode::Type::FloatSetPredicate, &ShaderIR::DecodeFloatSetPredicate},
-            {OpCode::Type::IntegerSetPredicate, &ShaderIR::DecodeIntegerSetPredicate},
-            {OpCode::Type::HalfSetPredicate, &ShaderIR::DecodeHalfSetPredicate},
-            {OpCode::Type::PredicateSetRegister, &ShaderIR::DecodePredicateSetRegister},
-            {OpCode::Type::PredicateSetPredicate, &ShaderIR::DecodePredicateSetPredicate},
-            {OpCode::Type::RegisterSetPredicate, &ShaderIR::DecodeRegisterSetPredicate},
-            {OpCode::Type::FloatSet, &ShaderIR::DecodeFloatSet},
-            {OpCode::Type::IntegerSet, &ShaderIR::DecodeIntegerSet},
-            {OpCode::Type::HalfSet, &ShaderIR::DecodeHalfSet},
-            {OpCode::Type::Video, &ShaderIR::DecodeVideo},
-            {OpCode::Type::Xmad, &ShaderIR::DecodeXmad},
-        };
+    static const std::map<OpCode::Type, u32 (ShaderIR::*)(NodeBlock&, u32)> decoders = {
+        {OpCode::Type::Arithmetic, &ShaderIR::DecodeArithmetic},
+        {OpCode::Type::ArithmeticImmediate, &ShaderIR::DecodeArithmeticImmediate},
+        {OpCode::Type::Bfe, &ShaderIR::DecodeBfe},
+        {OpCode::Type::Bfi, &ShaderIR::DecodeBfi},
+        {OpCode::Type::Shift, &ShaderIR::DecodeShift},
+        {OpCode::Type::ArithmeticInteger, &ShaderIR::DecodeArithmeticInteger},
+        {OpCode::Type::ArithmeticIntegerImmediate, &ShaderIR::DecodeArithmeticIntegerImmediate},
+        {OpCode::Type::ArithmeticHalf, &ShaderIR::DecodeArithmeticHalf},
+        {OpCode::Type::ArithmeticHalfImmediate, &ShaderIR::DecodeArithmeticHalfImmediate},
+        {OpCode::Type::Ffma, &ShaderIR::DecodeFfma},
+        {OpCode::Type::Hfma2, &ShaderIR::DecodeHfma2},
+        {OpCode::Type::Conversion, &ShaderIR::DecodeConversion},
+        {OpCode::Type::Memory, &ShaderIR::DecodeMemory},
+        {OpCode::Type::FloatSetPredicate, &ShaderIR::DecodeFloatSetPredicate},
+        {OpCode::Type::IntegerSetPredicate, &ShaderIR::DecodeIntegerSetPredicate},
+        {OpCode::Type::HalfSetPredicate, &ShaderIR::DecodeHalfSetPredicate},
+        {OpCode::Type::PredicateSetRegister, &ShaderIR::DecodePredicateSetRegister},
+        {OpCode::Type::PredicateSetPredicate, &ShaderIR::DecodePredicateSetPredicate},
+        {OpCode::Type::RegisterSetPredicate, &ShaderIR::DecodeRegisterSetPredicate},
+        {OpCode::Type::FloatSet, &ShaderIR::DecodeFloatSet},
+        {OpCode::Type::IntegerSet, &ShaderIR::DecodeIntegerSet},
+        {OpCode::Type::HalfSet, &ShaderIR::DecodeHalfSet},
+        {OpCode::Type::Video, &ShaderIR::DecodeVideo},
+        {OpCode::Type::Xmad, &ShaderIR::DecodeXmad},
+    };
 
     std::vector<Node> tmp_block;
     if (const auto decoder = decoders.find(opcode->get().GetType()); decoder != decoders.end()) {
-        pc = (this->*decoder->second)(tmp_block, bb, pc);
+        pc = (this->*decoder->second)(tmp_block, pc);
     } else {
-        pc = DecodeOther(tmp_block, bb, pc);
+        pc = DecodeOther(tmp_block, pc);
     }
 
     // Some instructions (like SSY) don't have a predicate field, they are always unconditionally
@@ -192,11 +191,14 @@ u32 ShaderIR::DecodeInstr(BasicBlock& bb, u32 pc) {
     const auto pred_index = static_cast<u32>(instr.pred.pred_index);
 
     if (can_be_predicated && pred_index != static_cast<u32>(Pred::UnusedIndex)) {
-        bb.push_back(
-            Conditional(GetPredicate(pred_index, instr.negate_pred != 0), std::move(tmp_block)));
+        const Node conditional =
+            Conditional(GetPredicate(pred_index, instr.negate_pred != 0), std::move(tmp_block));
+        global_code.push_back(conditional);
+        bb.push_back(conditional);
     } else {
         for (auto& node : tmp_block) {
-            bb.push_back(std::move(node));
+            global_code.push_back(node);
+            bb.push_back(node);
         }
     }
 

src/video_core/shader/decode/arithmetic.cpp

@@ -13,7 +13,7 @@ using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 using Tegra::Shader::SubOp;
 
-u32 ShaderIR::DecodeArithmetic(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeArithmetic(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 
@@ -25,7 +25,7 @@ u32 ShaderIR::DecodeArithmetic(BasicBlock& bb, const BasicBlock& code, u32 pc) {
         } else if (instr.is_b_gpr) {
             return GetRegister(instr.gpr20);
         } else {
-            return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.offset);
+            return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.GetOffset());
         }
     }();
 

src/video_core/shader/decode/arithmetic_half.cpp

@@ -12,7 +12,7 @@ namespace VideoCommon::Shader {
 using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 
-u32 ShaderIR::DecodeArithmeticHalf(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeArithmeticHalf(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 
@@ -35,7 +35,7 @@ u32 ShaderIR::DecodeArithmeticHalf(BasicBlock& bb, const BasicBlock& code, u32 p
         switch (opcode->get().GetId()) {
         case OpCode::Id::HADD2_C:
         case OpCode::Id::HMUL2_C:
-            return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.offset);
+            return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.GetOffset());
         case OpCode::Id::HADD2_R:
         case OpCode::Id::HMUL2_R:
             return GetRegister(instr.gpr20);

src/video_core/shader/decode/arithmetic_half_immediate.cpp

@@ -12,7 +12,7 @@ namespace VideoCommon::Shader {
 using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 
-u32 ShaderIR::DecodeArithmeticHalfImmediate(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeArithmeticHalfImmediate(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 

src/video_core/shader/decode/arithmetic_immediate.cpp

@@ -12,7 +12,7 @@ namespace VideoCommon::Shader {
 using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 
-u32 ShaderIR::DecodeArithmeticImmediate(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeArithmeticImmediate(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 

src/video_core/shader/decode/arithmetic_integer.cpp

@@ -15,7 +15,7 @@ using Tegra::Shader::OpCode;
 using Tegra::Shader::Pred;
 using Tegra::Shader::Register;
 
-u32 ShaderIR::DecodeArithmeticInteger(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeArithmeticInteger(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 
@@ -26,7 +26,7 @@ u32 ShaderIR::DecodeArithmeticInteger(BasicBlock& bb, const BasicBlock& code, u3
         } else if (instr.is_b_gpr) {
             return GetRegister(instr.gpr20);
         } else {
-            return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.offset);
+            return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.GetOffset());
         }
     }();
 
@@ -242,7 +242,7 @@ u32 ShaderIR::DecodeArithmeticInteger(BasicBlock& bb, const BasicBlock& code, u3
     return pc;
 }
 
-void ShaderIR::WriteLop3Instruction(BasicBlock& bb, Register dest, Node op_a, Node op_b, Node op_c,
+void ShaderIR::WriteLop3Instruction(NodeBlock& bb, Register dest, Node op_a, Node op_b, Node op_c,
                                     Node imm_lut, bool sets_cc) {
     constexpr u32 lop_iterations = 32;
     const Node one = Immediate(1);

src/video_core/shader/decode/arithmetic_integer_immediate.cpp

@@ -16,7 +16,7 @@ using Tegra::Shader::Pred;
 using Tegra::Shader::PredicateResultMode;
 using Tegra::Shader::Register;
 
-u32 ShaderIR::DecodeArithmeticIntegerImmediate(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeArithmeticIntegerImmediate(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 
@@ -54,9 +54,9 @@ u32 ShaderIR::DecodeArithmeticIntegerImmediate(BasicBlock& bb, const BasicBlock&
     return pc;
 }
 
-void ShaderIR::WriteLogicOperation(BasicBlock& bb, Register dest, LogicOperation logic_op,
-                                   Node op_a, Node op_b, PredicateResultMode predicate_mode,
-                                   Pred predicate, bool sets_cc) {
+void ShaderIR::WriteLogicOperation(NodeBlock& bb, Register dest, LogicOperation logic_op, Node op_a,
+                                   Node op_b, PredicateResultMode predicate_mode, Pred predicate,
+                                   bool sets_cc) {
     const Node result = [&]() {
         switch (logic_op) {
         case LogicOperation::And:

src/video_core/shader/decode/bfe.cpp

@@ -12,7 +12,7 @@ namespace VideoCommon::Shader {
 using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 
-u32 ShaderIR::DecodeBfe(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeBfe(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 

src/video_core/shader/decode/bfi.cpp

@@ -12,7 +12,7 @@ namespace VideoCommon::Shader {
 using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 
-u32 ShaderIR::DecodeBfi(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeBfi(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 

src/video_core/shader/decode/conversion.cpp

@@ -13,7 +13,7 @@ using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 using Tegra::Shader::Register;
 
-u32 ShaderIR::DecodeConversion(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeConversion(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 
@@ -48,7 +48,7 @@ u32 ShaderIR::DecodeConversion(BasicBlock& bb, const BasicBlock& code, u32 pc) {
             if (instr.is_b_gpr) {
                 return GetRegister(instr.gpr20);
             } else {
-                return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.offset);
+                return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.GetOffset());
             }
         }();
         const bool input_signed = instr.conversion.is_input_signed;
@@ -72,7 +72,7 @@ u32 ShaderIR::DecodeConversion(BasicBlock& bb, const BasicBlock& code, u32 pc) {
             if (instr.is_b_gpr) {
                 return GetRegister(instr.gpr20);
             } else {
-                return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.offset);
+                return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.GetOffset());
             }
         }();
 
@@ -110,7 +110,7 @@ u32 ShaderIR::DecodeConversion(BasicBlock& bb, const BasicBlock& code, u32 pc) {
             if (instr.is_b_gpr) {
                 return GetRegister(instr.gpr20);
             } else {
-                return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.offset);
+                return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.GetOffset());
             }
         }();
 

src/video_core/shader/decode/ffma.cpp

@@ -12,7 +12,7 @@ namespace VideoCommon::Shader {
 using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 
-u32 ShaderIR::DecodeFfma(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeFfma(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 
@@ -27,14 +27,14 @@ u32 ShaderIR::DecodeFfma(BasicBlock& bb, const BasicBlock& code, u32 pc) {
     auto [op_b, op_c] = [&]() -> std::tuple<Node, Node> {
         switch (opcode->get().GetId()) {
         case OpCode::Id::FFMA_CR: {
-            return {GetConstBuffer(instr.cbuf34.index, instr.cbuf34.offset),
+            return {GetConstBuffer(instr.cbuf34.index, instr.cbuf34.GetOffset()),
                     GetRegister(instr.gpr39)};
         }
         case OpCode::Id::FFMA_RR:
             return {GetRegister(instr.gpr20), GetRegister(instr.gpr39)};
         case OpCode::Id::FFMA_RC: {
             return {GetRegister(instr.gpr39),
-                    GetConstBuffer(instr.cbuf34.index, instr.cbuf34.offset)};
+                    GetConstBuffer(instr.cbuf34.index, instr.cbuf34.GetOffset())};
         }
         case OpCode::Id::FFMA_IMM:
             return {GetImmediate19(instr), GetRegister(instr.gpr39)};

src/video_core/shader/decode/float_set.cpp

@@ -12,7 +12,7 @@ namespace VideoCommon::Shader {
 using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 
-u32 ShaderIR::DecodeFloatSet(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeFloatSet(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 
@@ -25,7 +25,7 @@ u32 ShaderIR::DecodeFloatSet(BasicBlock& bb, const BasicBlock& code, u32 pc) {
         } else if (instr.is_b_gpr) {
             return GetRegister(instr.gpr20);
         } else {
-            return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.offset);
+            return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.GetOffset());
         }
     }();
 

src/video_core/shader/decode/float_set_predicate.cpp

@@ -13,7 +13,7 @@ using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 using Tegra::Shader::Pred;
 
-u32 ShaderIR::DecodeFloatSetPredicate(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeFloatSetPredicate(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 
@@ -25,7 +25,7 @@ u32 ShaderIR::DecodeFloatSetPredicate(BasicBlock& bb, const BasicBlock& code, u3
         } else if (instr.is_b_gpr) {
             return GetRegister(instr.gpr20);
         } else {
-            return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.offset);
+            return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.GetOffset());
         }
     }();
     op_b = GetOperandAbsNegFloat(op_b, instr.fsetp.abs_b, false);

src/video_core/shader/decode/half_set.cpp

@@ -14,7 +14,7 @@ namespace VideoCommon::Shader {
 using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 
-u32 ShaderIR::DecodeHalfSet(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeHalfSet(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 

src/video_core/shader/decode/half_set_predicate.cpp

@@ -13,7 +13,7 @@ using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 using Tegra::Shader::Pred;
 
-u32 ShaderIR::DecodeHalfSetPredicate(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeHalfSetPredicate(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 

src/video_core/shader/decode/hfma2.cpp

@@ -16,7 +16,7 @@ using Tegra::Shader::HalfType;
 using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 
-u32 ShaderIR::DecodeHfma2(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeHfma2(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 
@@ -39,13 +39,14 @@ u32 ShaderIR::DecodeHfma2(BasicBlock& bb, const BasicBlock& code, u32 pc) {
             neg_b = instr.hfma2.negate_b;
             neg_c = instr.hfma2.negate_c;
             return {instr.hfma2.saturate, instr.hfma2.type_b,
-                    GetConstBuffer(instr.cbuf34.index, instr.cbuf34.offset), instr.hfma2.type_reg39,
-                    GetRegister(instr.gpr39)};
+                    GetConstBuffer(instr.cbuf34.index, instr.cbuf34.GetOffset()),
+                    instr.hfma2.type_reg39, GetRegister(instr.gpr39)};
         case OpCode::Id::HFMA2_RC:
             neg_b = instr.hfma2.negate_b;
             neg_c = instr.hfma2.negate_c;
             return {instr.hfma2.saturate, instr.hfma2.type_reg39, GetRegister(instr.gpr39),
-                    instr.hfma2.type_b, GetConstBuffer(instr.cbuf34.index, instr.cbuf34.offset)};
+                    instr.hfma2.type_b,
+                    GetConstBuffer(instr.cbuf34.index, instr.cbuf34.GetOffset())};
         case OpCode::Id::HFMA2_RR:
             neg_b = instr.hfma2.rr.negate_b;
             neg_c = instr.hfma2.rr.negate_c;

src/video_core/shader/decode/integer_set.cpp

@@ -12,7 +12,7 @@ namespace VideoCommon::Shader {
 using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 
-u32 ShaderIR::DecodeIntegerSet(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeIntegerSet(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 
@@ -23,7 +23,7 @@ u32 ShaderIR::DecodeIntegerSet(BasicBlock& bb, const BasicBlock& code, u32 pc) {
         } else if (instr.is_b_gpr) {
             return GetRegister(instr.gpr20);
         } else {
-            return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.offset);
+            return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.GetOffset());
         }
     }();
 

src/video_core/shader/decode/integer_set_predicate.cpp

@@ -13,7 +13,7 @@ using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 using Tegra::Shader::Pred;
 
-u32 ShaderIR::DecodeIntegerSetPredicate(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeIntegerSetPredicate(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 
@@ -25,7 +25,7 @@ u32 ShaderIR::DecodeIntegerSetPredicate(BasicBlock& bb, const BasicBlock& code,
         } else if (instr.is_b_gpr) {
             return GetRegister(instr.gpr20);
         } else {
-            return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.offset);
+            return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.GetOffset());
         }
     }();
 

src/video_core/shader/decode/memory.cpp

@@ -4,6 +4,7 @@
 
 #include <algorithm>
 #include <vector>
+#include <fmt/format.h>
 
 #include "common/assert.h"
 #include "common/common_types.h"
@@ -35,7 +36,7 @@ static std::size_t GetCoordCount(TextureType texture_type) {
     }
 }
 
-u32 ShaderIR::DecodeMemory(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 
@@ -79,7 +80,7 @@ u32 ShaderIR::DecodeMemory(BasicBlock& bb, const BasicBlock& code, u32 pc) {
         Node index = GetRegister(instr.gpr8);
 
         const Node op_a =
-            GetConstBufferIndirect(instr.cbuf36.index, instr.cbuf36.offset + 0, index);
+            GetConstBufferIndirect(instr.cbuf36.index, instr.cbuf36.GetOffset() + 0, index);
 
         switch (instr.ld_c.type.Value()) {
         case Tegra::Shader::UniformType::Single:
@@ -88,7 +89,7 @@ u32 ShaderIR::DecodeMemory(BasicBlock& bb, const BasicBlock& code, u32 pc) {
 
         case Tegra::Shader::UniformType::Double: {
             const Node op_b =
-                GetConstBufferIndirect(instr.cbuf36.index, instr.cbuf36.offset + 4, index);
+                GetConstBufferIndirect(instr.cbuf36.index, instr.cbuf36.GetOffset() + 4, index);
 
             SetTemporal(bb, 0, op_a);
             SetTemporal(bb, 1, op_b);
@@ -119,6 +120,55 @@ u32 ShaderIR::DecodeMemory(BasicBlock& bb, const BasicBlock& code, u32 pc) {
         }
         break;
     }
+    case OpCode::Id::LDG: {
+        const u32 count = [&]() {
+            switch (instr.ldg.type) {
+            case Tegra::Shader::UniformType::Single:
+                return 1;
+            case Tegra::Shader::UniformType::Double:
+                return 2;
+            case Tegra::Shader::UniformType::Quad:
+            case Tegra::Shader::UniformType::UnsignedQuad:
+                return 4;
+            default:
+                UNIMPLEMENTED_MSG("Unimplemented LDG size!");
+                return 1;
+            }
+        }();
+
+        const Node addr_register = GetRegister(instr.gpr8);
+        const Node base_address =
+            TrackCbuf(addr_register, global_code, static_cast<s64>(global_code.size()));
+        const auto cbuf = std::get_if<CbufNode>(base_address);
+        ASSERT(cbuf != nullptr);
+        const auto cbuf_offset_imm = std::get_if<ImmediateNode>(cbuf->GetOffset());
+        ASSERT(cbuf_offset_imm != nullptr);
+        const auto cbuf_offset = cbuf_offset_imm->GetValue();
+
+        bb.push_back(Comment(
+            fmt::format("Base address is c[0x{:x}][0x{:x}]", cbuf->GetIndex(), cbuf_offset)));
+
+        const GlobalMemoryBase descriptor{cbuf->GetIndex(), cbuf_offset};
+        used_global_memory_bases.insert(descriptor);
+
+        const Node immediate_offset =
+            Immediate(static_cast<u32>(instr.ldg.immediate_offset.Value()));
+        const Node base_real_address =
+            Operation(OperationCode::UAdd, NO_PRECISE, immediate_offset, addr_register);
+
+        for (u32 i = 0; i < count; ++i) {
+            const Node it_offset = Immediate(i * 4);
+            const Node real_address =
+                Operation(OperationCode::UAdd, NO_PRECISE, base_real_address, it_offset);
+            const Node gmem = StoreNode(GmemNode(real_address, base_address, descriptor));
+
+            SetTemporal(bb, i, gmem);
+        }
+        for (u32 i = 0; i < count; ++i) {
+            SetRegister(bb, instr.gpr0.Value() + i, GetTemporal(i));
+        }
+        break;
+    }
     case OpCode::Id::ST_A: {
         UNIMPLEMENTED_IF_MSG(instr.gpr8.Value() != Register::ZeroIndex,
                              "Indirect attribute loads are not supported");
@@ -381,8 +431,7 @@ const Sampler& ShaderIR::GetSampler(const Tegra::Shader::Sampler& sampler, Textu
     return *used_samplers.emplace(entry).first;
 }
 
-void ShaderIR::WriteTexInstructionFloat(BasicBlock& bb, Instruction instr,
-                                        const Node4& components) {
+void ShaderIR::WriteTexInstructionFloat(NodeBlock& bb, Instruction instr, const Node4& components) {
     u32 dest_elem = 0;
     for (u32 elem = 0; elem < 4; ++elem) {
         if (!instr.tex.IsComponentEnabled(elem)) {
@@ -397,7 +446,7 @@ void ShaderIR::WriteTexInstructionFloat(BasicBlock& bb, Instruction instr,
     }
 }
 
-void ShaderIR::WriteTexsInstructionFloat(BasicBlock& bb, Instruction instr,
+void ShaderIR::WriteTexsInstructionFloat(NodeBlock& bb, Instruction instr,
                                          const Node4& components) {
     // TEXS has two destination registers and a swizzle. The first two elements in the swizzle
     // go into gpr0+0 and gpr0+1, and the rest goes into gpr28+0 and gpr28+1
@@ -421,7 +470,7 @@ void ShaderIR::WriteTexsInstructionFloat(BasicBlock& bb, Instruction instr,
     }
 }
 
-void ShaderIR::WriteTexsInstructionHalfFloat(BasicBlock& bb, Instruction instr,
+void ShaderIR::WriteTexsInstructionHalfFloat(NodeBlock& bb, Instruction instr,
                                              const Node4& components) {
     // TEXS.F16 destionation registers are packed in two registers in pairs (just like any half
     // float instruction).

src/video_core/shader/decode/other.cpp

@@ -14,7 +14,7 @@ using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 using Tegra::Shader::Register;
 
-u32 ShaderIR::DecodeOther(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeOther(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 

src/video_core/shader/decode/predicate_set_predicate.cpp

@@ -13,7 +13,7 @@ using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 using Tegra::Shader::Pred;
 
-u32 ShaderIR::DecodePredicateSetPredicate(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodePredicateSetPredicate(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 

src/video_core/shader/decode/predicate_set_register.cpp

@@ -12,7 +12,7 @@ namespace VideoCommon::Shader {
 using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 
-u32 ShaderIR::DecodePredicateSetRegister(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodePredicateSetRegister(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 

src/video_core/shader/decode/register_set_predicate.cpp

@@ -12,7 +12,7 @@ namespace VideoCommon::Shader {
 using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 
-u32 ShaderIR::DecodeRegisterSetPredicate(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeRegisterSetPredicate(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 

src/video_core/shader/decode/shift.cpp

@@ -12,7 +12,7 @@ namespace VideoCommon::Shader {
 using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 
-u32 ShaderIR::DecodeShift(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeShift(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 
@@ -23,7 +23,7 @@ u32 ShaderIR::DecodeShift(BasicBlock& bb, const BasicBlock& code, u32 pc) {
         } else if (instr.is_b_gpr) {
             return GetRegister(instr.gpr20);
         } else {
-            return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.offset);
+            return GetConstBuffer(instr.cbuf34.index, instr.cbuf34.GetOffset());
         }
     }();
 

src/video_core/shader/decode/video.cpp

@@ -15,7 +15,7 @@ using Tegra::Shader::Pred;
 using Tegra::Shader::VideoType;
 using Tegra::Shader::VmadShr;
 
-u32 ShaderIR::DecodeVideo(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeVideo(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 

src/video_core/shader/decode/xmad.cpp

@@ -12,7 +12,7 @@ namespace VideoCommon::Shader {
 using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 
-u32 ShaderIR::DecodeXmad(BasicBlock& bb, const BasicBlock& code, u32 pc) {
+u32 ShaderIR::DecodeXmad(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 
@@ -32,13 +32,14 @@ u32 ShaderIR::DecodeXmad(BasicBlock& bb, const BasicBlock& code, u32 pc) {
     auto [is_merge, op_b, op_c] = [&]() -> std::tuple<bool, Node, Node> {
         switch (opcode->get().GetId()) {
         case OpCode::Id::XMAD_CR:
-            return {instr.xmad.merge_56, GetConstBuffer(instr.cbuf34.index, instr.cbuf34.offset),
+            return {instr.xmad.merge_56,
+                    GetConstBuffer(instr.cbuf34.index, instr.cbuf34.GetOffset()),
                     GetRegister(instr.gpr39)};
         case OpCode::Id::XMAD_RR:
             return {instr.xmad.merge_37, GetRegister(instr.gpr20), GetRegister(instr.gpr39)};
         case OpCode::Id::XMAD_RC:
             return {false, GetRegister(instr.gpr39),
-                    GetConstBuffer(instr.cbuf34.index, instr.cbuf34.offset)};
+                    GetConstBuffer(instr.cbuf34.index, instr.cbuf34.GetOffset())};
         case OpCode::Id::XMAD_IMM:
             return {instr.xmad.merge_37, Immediate(static_cast<u32>(instr.xmad.imm20_16)),
                     GetRegister(instr.gpr39)};

src/video_core/shader/shader_ir.cpp

@@ -337,27 +337,27 @@ Node ShaderIR::GetConditionCode(Tegra::Shader::ConditionCode cc) {
     }
 }
 
-void ShaderIR::SetRegister(BasicBlock& bb, Register dest, Node src) {
+void ShaderIR::SetRegister(NodeBlock& bb, Register dest, Node src) {
     bb.push_back(Operation(OperationCode::Assign, GetRegister(dest), src));
 }
 
-void ShaderIR::SetPredicate(BasicBlock& bb, u64 dest, Node src) {
+void ShaderIR::SetPredicate(NodeBlock& bb, u64 dest, Node src) {
     bb.push_back(Operation(OperationCode::LogicalAssign, GetPredicate(dest), src));
 }
 
-void ShaderIR::SetInternalFlag(BasicBlock& bb, InternalFlag flag, Node value) {
+void ShaderIR::SetInternalFlag(NodeBlock& bb, InternalFlag flag, Node value) {
     bb.push_back(Operation(OperationCode::LogicalAssign, GetInternalFlag(flag), value));
 }
 
-void ShaderIR::SetLocalMemory(BasicBlock& bb, Node address, Node value) {
+void ShaderIR::SetLocalMemory(NodeBlock& bb, Node address, Node value) {
     bb.push_back(Operation(OperationCode::Assign, GetLocalMemory(address), value));
 }
 
-void ShaderIR::SetTemporal(BasicBlock& bb, u32 id, Node value) {
+void ShaderIR::SetTemporal(NodeBlock& bb, u32 id, Node value) {
     SetRegister(bb, Register::ZeroIndex + 1 + id, value);
 }
 
-void ShaderIR::SetInternalFlagsFromFloat(BasicBlock& bb, Node value, bool sets_cc) {
+void ShaderIR::SetInternalFlagsFromFloat(NodeBlock& bb, Node value, bool sets_cc) {
     if (!sets_cc) {
         return;
     }
@@ -366,7 +366,7 @@ void ShaderIR::SetInternalFlagsFromFloat(BasicBlock& bb, Node value, bool sets_c
     LOG_WARNING(HW_GPU, "Condition codes implementation is incomplete");
 }
 
-void ShaderIR::SetInternalFlagsFromInteger(BasicBlock& bb, Node value, bool sets_cc) {
+void ShaderIR::SetInternalFlagsFromInteger(NodeBlock& bb, Node value, bool sets_cc) {
     if (!sets_cc) {
         return;
     }

src/video_core/shader/shader_ir.h

@@ -39,7 +39,7 @@ using NodeData =
                  PredicateNode, AbufNode, CbufNode, LmemNode, GmemNode, CommentNode>;
 using Node = const NodeData*;
 using Node4 = std::array<Node, 4>;
-using BasicBlock = std::vector<Node>;
+using NodeBlock = std::vector<Node>;
 
 constexpr u32 MAX_PROGRAM_LENGTH = 0x1000;
 
@@ -249,7 +249,7 @@ public:
     }
 
     u32 GetSize() const {
-        return max_offset + 1;
+        return max_offset + sizeof(float);
     }
 
 private:
@@ -257,6 +257,15 @@ private:
     bool is_indirect{};
 };
 
+struct GlobalMemoryBase {
+    u32 cbuf_index{};
+    u32 cbuf_offset{};
+
+    bool operator<(const GlobalMemoryBase& rhs) const {
+        return std::tie(cbuf_index, cbuf_offset) < std::tie(rhs.cbuf_index, rhs.cbuf_offset);
+    }
+};
+
 struct MetaArithmetic {
     bool precise{};
 };
@@ -391,7 +400,7 @@ private:
     const InternalFlag flag;
 };
 
-/// A predicate register, it can be negated without aditional nodes
+/// A predicate register, it can be negated without additional nodes
 class PredicateNode final {
 public:
     explicit constexpr PredicateNode(Tegra::Shader::Pred index, bool negated)
@@ -415,11 +424,11 @@ class AbufNode final {
 public:
     explicit constexpr AbufNode(Tegra::Shader::Attribute::Index index, u32 element,
                                 const Tegra::Shader::IpaMode& input_mode, Node buffer = {})
-        : input_mode{input_mode}, index{index}, element{element}, buffer{buffer} {}
+        : input_mode{input_mode}, buffer{buffer}, index{index}, element{element} {}
 
     explicit constexpr AbufNode(Tegra::Shader::Attribute::Index index, u32 element,
                                 Node buffer = {})
-        : input_mode{}, index{index}, element{element}, buffer{buffer} {}
+        : input_mode{}, buffer{buffer}, index{index}, element{element} {}
 
     Tegra::Shader::IpaMode GetInputMode() const {
         return input_mode;
@@ -478,14 +487,26 @@ private:
 /// Global memory node
 class GmemNode final {
 public:
-    explicit constexpr GmemNode(Node address) : address{address} {}
+    explicit constexpr GmemNode(Node real_address, Node base_address,
+                                const GlobalMemoryBase& descriptor)
+        : real_address{real_address}, base_address{base_address}, descriptor{descriptor} {}
 
-    Node GetAddress() const {
-        return address;
+    Node GetRealAddress() const {
+        return real_address;
+    }
+
+    Node GetBaseAddress() const {
+        return base_address;
+    }
+
+    const GlobalMemoryBase& GetDescriptor() const {
+        return descriptor;
     }
 
 private:
-    const Node address;
+    const Node real_address;
+    const Node base_address;
+    const GlobalMemoryBase descriptor;
 };
 
 /// Commentary, can be dropped
@@ -509,7 +530,7 @@ public:
         Decode();
     }
 
-    const std::map<u32, BasicBlock>& GetBasicBlocks() const {
+    const std::map<u32, NodeBlock>& GetBasicBlocks() const {
         return basic_blocks;
     }
 
@@ -543,6 +564,10 @@ public:
         return used_clip_distances;
     }
 
+    const std::set<GlobalMemoryBase>& GetGlobalMemoryBases() const {
+        return used_global_memory_bases;
+    }
+
     std::size_t GetLength() const {
         return static_cast<std::size_t>(coverage_end * sizeof(u64));
     }
@@ -556,7 +581,7 @@ private:
 
     ExitMethod Scan(u32 begin, u32 end, std::set<u32>& labels);
 
-    BasicBlock DecodeRange(u32 begin, u32 end);
+    NodeBlock DecodeRange(u32 begin, u32 end);
 
     /**
      * Decodes a single instruction from Tegra to IR.
@@ -564,33 +589,33 @@ private:
      * @param pc Program counter. Offset to decode.
      * @return Next address to decode.
      */
-    u32 DecodeInstr(BasicBlock& bb, u32 pc);
+    u32 DecodeInstr(NodeBlock& bb, u32 pc);
 
-    u32 DecodeArithmetic(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeArithmeticImmediate(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeBfe(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeBfi(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeShift(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeArithmeticInteger(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeArithmeticIntegerImmediate(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeArithmeticHalf(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeArithmeticHalfImmediate(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeFfma(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeHfma2(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeConversion(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeMemory(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeFloatSetPredicate(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeIntegerSetPredicate(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeHalfSetPredicate(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodePredicateSetRegister(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodePredicateSetPredicate(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeRegisterSetPredicate(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeFloatSet(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeIntegerSet(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeHalfSet(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeVideo(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeXmad(BasicBlock& bb, const BasicBlock& code, u32 pc);
-    u32 DecodeOther(BasicBlock& bb, const BasicBlock& code, u32 pc);
+    u32 DecodeArithmetic(NodeBlock& bb, u32 pc);
+    u32 DecodeArithmeticImmediate(NodeBlock& bb, u32 pc);
+    u32 DecodeBfe(NodeBlock& bb, u32 pc);
+    u32 DecodeBfi(NodeBlock& bb, u32 pc);
+    u32 DecodeShift(NodeBlock& bb, u32 pc);
+    u32 DecodeArithmeticInteger(NodeBlock& bb, u32 pc);
+    u32 DecodeArithmeticIntegerImmediate(NodeBlock& bb, u32 pc);
+    u32 DecodeArithmeticHalf(NodeBlock& bb, u32 pc);
+    u32 DecodeArithmeticHalfImmediate(NodeBlock& bb, u32 pc);
+    u32 DecodeFfma(NodeBlock& bb, u32 pc);
+    u32 DecodeHfma2(NodeBlock& bb, u32 pc);
+    u32 DecodeConversion(NodeBlock& bb, u32 pc);
+    u32 DecodeMemory(NodeBlock& bb, u32 pc);
+    u32 DecodeFloatSetPredicate(NodeBlock& bb, u32 pc);
+    u32 DecodeIntegerSetPredicate(NodeBlock& bb, u32 pc);
+    u32 DecodeHalfSetPredicate(NodeBlock& bb, u32 pc);
+    u32 DecodePredicateSetRegister(NodeBlock& bb, u32 pc);
+    u32 DecodePredicateSetPredicate(NodeBlock& bb, u32 pc);
+    u32 DecodeRegisterSetPredicate(NodeBlock& bb, u32 pc);
+    u32 DecodeFloatSet(NodeBlock& bb, u32 pc);
+    u32 DecodeIntegerSet(NodeBlock& bb, u32 pc);
+    u32 DecodeHalfSet(NodeBlock& bb, u32 pc);
+    u32 DecodeVideo(NodeBlock& bb, u32 pc);
+    u32 DecodeXmad(NodeBlock& bb, u32 pc);
+    u32 DecodeOther(NodeBlock& bb, u32 pc);
 
     /// Internalizes node's data and returns a managed pointer to a clone of that node
     Node StoreNode(NodeData&& node_data);
@@ -626,10 +651,10 @@ private:
     Node GetPredicate(u64 pred, bool negated = false);
     /// Generates a predicate node for an immediate true or false value
     Node GetPredicate(bool immediate);
-    /// Generates a node representing an input atttribute. Keeps track of used attributes.
+    /// Generates a node representing an input attribute. Keeps track of used attributes.
     Node GetInputAttribute(Tegra::Shader::Attribute::Index index, u64 element,
                            const Tegra::Shader::IpaMode& input_mode, Node buffer = {});
-    /// Generates a node representing an output atttribute. Keeps track of used attributes.
+    /// Generates a node representing an output attribute. Keeps track of used attributes.
     Node GetOutputAttribute(Tegra::Shader::Attribute::Index index, u64 element, Node buffer);
     /// Generates a node representing an internal flag
     Node GetInternalFlag(InternalFlag flag, bool negated = false);
@@ -639,20 +664,20 @@ private:
     Node GetTemporal(u32 id);
 
     /// Sets a register. src value must be a number-evaluated node.
-    void SetRegister(BasicBlock& bb, Tegra::Shader::Register dest, Node src);
+    void SetRegister(NodeBlock& bb, Tegra::Shader::Register dest, Node src);
     /// Sets a predicate. src value must be a bool-evaluated node
-    void SetPredicate(BasicBlock& bb, u64 dest, Node src);
+    void SetPredicate(NodeBlock& bb, u64 dest, Node src);
     /// Sets an internal flag. src value must be a bool-evaluated node
-    void SetInternalFlag(BasicBlock& bb, InternalFlag flag, Node value);
+    void SetInternalFlag(NodeBlock& bb, InternalFlag flag, Node value);
     /// Sets a local memory address. address and value must be a number-evaluated node
-    void SetLocalMemory(BasicBlock& bb, Node address, Node value);
+    void SetLocalMemory(NodeBlock& bb, Node address, Node value);
     /// Sets a temporal. Internally it uses a post-RZ register
-    void SetTemporal(BasicBlock& bb, u32 id, Node value);
+    void SetTemporal(NodeBlock& bb, u32 id, Node value);
 
     /// Sets internal flags from a float
-    void SetInternalFlagsFromFloat(BasicBlock& bb, Node value, bool sets_cc = true);
+    void SetInternalFlagsFromFloat(NodeBlock& bb, Node value, bool sets_cc = true);
     /// Sets internal flags from an integer
-    void SetInternalFlagsFromInteger(BasicBlock& bb, Node value, bool sets_cc = true);
+    void SetInternalFlagsFromInteger(NodeBlock& bb, Node value, bool sets_cc = true);
 
     /// Conditionally absolute/negated float. Absolute is applied first
     Node GetOperandAbsNegFloat(Node value, bool absolute, bool negate);
@@ -693,12 +718,12 @@ private:
     /// Extracts a sequence of bits from a node
     Node BitfieldExtract(Node value, u32 offset, u32 bits);
 
-    void WriteTexInstructionFloat(BasicBlock& bb, Tegra::Shader::Instruction instr,
+    void WriteTexInstructionFloat(NodeBlock& bb, Tegra::Shader::Instruction instr,
                                   const Node4& components);
 
-    void WriteTexsInstructionFloat(BasicBlock& bb, Tegra::Shader::Instruction instr,
+    void WriteTexsInstructionFloat(NodeBlock& bb, Tegra::Shader::Instruction instr,
                                    const Node4& components);
-    void WriteTexsInstructionHalfFloat(BasicBlock& bb, Tegra::Shader::Instruction instr,
+    void WriteTexsInstructionHalfFloat(NodeBlock& bb, Tegra::Shader::Instruction instr,
                                        const Node4& components);
 
     Node4 GetTexCode(Tegra::Shader::Instruction instr, Tegra::Shader::TextureType texture_type,
@@ -727,13 +752,17 @@ private:
     Node GetVideoOperand(Node op, bool is_chunk, bool is_signed, Tegra::Shader::VideoType type,
                          u64 byte_height);
 
-    void WriteLogicOperation(BasicBlock& bb, Tegra::Shader::Register dest,
+    void WriteLogicOperation(NodeBlock& bb, Tegra::Shader::Register dest,
                              Tegra::Shader::LogicOperation logic_op, Node op_a, Node op_b,
                              Tegra::Shader::PredicateResultMode predicate_mode,
                              Tegra::Shader::Pred predicate, bool sets_cc);
-    void WriteLop3Instruction(BasicBlock& bb, Tegra::Shader::Register dest, Node op_a, Node op_b,
+    void WriteLop3Instruction(NodeBlock& bb, Tegra::Shader::Register dest, Node op_a, Node op_b,
                               Node op_c, Node imm_lut, bool sets_cc);
 
+    Node TrackCbuf(Node tracked, const NodeBlock& code, s64 cursor);
+
+    std::pair<Node, s64> TrackRegister(const GprNode* tracked, const NodeBlock& code, s64 cursor);
+
     template <typename... T>
     Node Operation(OperationCode code, const T*... operands) {
         return StoreNode(OperationNode(code, operands...));
@@ -774,7 +803,8 @@ private:
     u32 coverage_end{};
     std::map<std::pair<u32, u32>, ExitMethod> exit_method_map;
 
-    std::map<u32, BasicBlock> basic_blocks;
+    std::map<u32, NodeBlock> basic_blocks;
+    NodeBlock global_code;
 
     std::vector<std::unique_ptr<NodeData>> stored_nodes;
 
@@ -786,6 +816,7 @@ private:
     std::map<u32, ConstBuffer> used_cbufs;
     std::set<Sampler> used_samplers;
     std::array<bool, Tegra::Engines::Maxwell3D::Regs::NumClipDistances> used_clip_distances{};
+    std::set<GlobalMemoryBase> used_global_memory_bases;
 
     Tegra::Shader::Header header;
 };

src/video_core/shader/track.cpp

@@ -0,0 +1,87 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <algorithm>
+#include <utility>
+#include <variant>
+
+#include "video_core/shader/shader_ir.h"
+
+namespace VideoCommon::Shader {
+
+namespace {
+std::pair<Node, s64> FindOperation(const NodeBlock& code, s64 cursor,
+                                   OperationCode operation_code) {
+    for (; cursor >= 0; --cursor) {
+        const Node node = code[cursor];
+        if (const auto operation = std::get_if<OperationNode>(node)) {
+            if (operation->GetCode() == operation_code)
+                return {node, cursor};
+        }
+        if (const auto conditional = std::get_if<ConditionalNode>(node)) {
+            const auto& code = conditional->GetCode();
+            const auto [found, internal_cursor] =
+                FindOperation(code, static_cast<s64>(code.size() - 1), operation_code);
+            if (found)
+                return {found, cursor};
+        }
+    }
+    return {};
+}
+} // namespace
+
+Node ShaderIR::TrackCbuf(Node tracked, const NodeBlock& code, s64 cursor) {
+    if (const auto cbuf = std::get_if<CbufNode>(tracked)) {
+        // Cbuf found, but it has to be immediate
+        return std::holds_alternative<ImmediateNode>(*cbuf->GetOffset()) ? tracked : nullptr;
+    }
+    if (const auto gpr = std::get_if<GprNode>(tracked)) {
+        if (gpr->GetIndex() == Tegra::Shader::Register::ZeroIndex) {
+            return nullptr;
+        }
+        // Reduce the cursor in one to avoid infinite loops when the instruction sets the same
+        // register that it uses as operand
+        const auto [source, new_cursor] = TrackRegister(gpr, code, cursor - 1);
+        if (!source) {
+            return nullptr;
+        }
+        return TrackCbuf(source, code, new_cursor);
+    }
+    if (const auto operation = std::get_if<OperationNode>(tracked)) {
+        for (std::size_t i = 0; i < operation->GetOperandsCount(); ++i) {
+            if (const auto found = TrackCbuf((*operation)[i], code, cursor)) {
+                // Cbuf found in operand
+                return found;
+            }
+        }
+        return nullptr;
+    }
+    if (const auto conditional = std::get_if<ConditionalNode>(tracked)) {
+        const auto& code = conditional->GetCode();
+        return TrackCbuf(tracked, code, static_cast<s64>(code.size()));
+    }
+    return nullptr;
+}
+
+std::pair<Node, s64> ShaderIR::TrackRegister(const GprNode* tracked, const NodeBlock& code,
+                                             s64 cursor) {
+    for (; cursor >= 0; --cursor) {
+        const auto [found_node, new_cursor] = FindOperation(code, cursor, OperationCode::Assign);
+        if (!found_node) {
+            return {};
+        }
+        const auto operation = std::get_if<OperationNode>(found_node);
+        ASSERT(operation);
+
+        const auto& target = (*operation)[0];
+        if (const auto gpr_target = std::get_if<GprNode>(target)) {
+            if (gpr_target->GetIndex() == tracked->GetIndex()) {
+                return {(*operation)[1], new_cursor};
+            }
+        }
+    }
+    return {};
+}
+
+} // namespace VideoCommon::Shader

src/video_core/surface.cpp

@@ -50,6 +50,24 @@ bool SurfaceTargetIsLayered(SurfaceTarget target) {
     }
 }
 
+bool SurfaceTargetIsArray(SurfaceTarget target) {
+    switch (target) {
+    case SurfaceTarget::Texture1D:
+    case SurfaceTarget::Texture2D:
+    case SurfaceTarget::Texture3D:
+    case SurfaceTarget::TextureCubemap:
+        return false;
+    case SurfaceTarget::Texture1DArray:
+    case SurfaceTarget::Texture2DArray:
+    case SurfaceTarget::TextureCubeArray:
+        return true;
+    default:
+        LOG_CRITICAL(HW_GPU, "Unimplemented surface_target={}", static_cast<u32>(target));
+        UNREACHABLE();
+        return false;
+    }
+}
+
 PixelFormat PixelFormatFromDepthFormat(Tegra::DepthFormat format) {
     switch (format) {
     case Tegra::DepthFormat::S8_Z24_UNORM:

src/video_core/surface.h

@@ -109,8 +109,7 @@ enum class SurfaceType {
     ColorTexture = 0,
     Depth = 1,
     DepthStencil = 2,
-    Fill = 3,
-    Invalid = 4,
+    Invalid = 3,
 };
 
 enum class SurfaceTarget {
@@ -441,6 +440,8 @@ SurfaceTarget SurfaceTargetFromTextureType(Tegra::Texture::TextureType texture_t
 
 bool SurfaceTargetIsLayered(SurfaceTarget target);
 
+bool SurfaceTargetIsArray(SurfaceTarget target);
+
 PixelFormat PixelFormatFromDepthFormat(Tegra::DepthFormat format);
 
 PixelFormat PixelFormatFromRenderTargetFormat(Tegra::RenderTargetFormat format);

src/video_core/video_core.cpp

@@ -16,9 +16,10 @@ std::unique_ptr<RendererBase> CreateRenderer(Core::Frontend::EmuWindow& emu_wind
 }
 
 u16 GetResolutionScaleFactor(const RendererBase& renderer) {
-    return !Settings::values.resolution_factor
-               ? renderer.GetRenderWindow().GetFramebufferLayout().GetScalingRatio()
-               : Settings::values.resolution_factor;
+    return static_cast<u16>(
+        Settings::values.resolution_factor
+            ? Settings::values.resolution_factor
+            : renderer.GetRenderWindow().GetFramebufferLayout().GetScalingRatio());
 }
 
 } // namespace VideoCore

src/yuzu/debugger/wait_tree.cpp

@@ -13,7 +13,6 @@
 #include "core/hle/kernel/readable_event.h"
 #include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/thread.h"
-#include "core/hle/kernel/timer.h"
 #include "core/hle/kernel/wait_object.h"
 #include "core/memory.h"
 
@@ -155,8 +154,6 @@ std::unique_ptr<WaitTreeWaitObject> WaitTreeWaitObject::make(const Kernel::WaitO
     switch (object.GetHandleType()) {
     case Kernel::HandleType::ReadableEvent:
         return std::make_unique<WaitTreeEvent>(static_cast<const Kernel::ReadableEvent&>(object));
-    case Kernel::HandleType::Timer:
-        return std::make_unique<WaitTreeTimer>(static_cast<const Kernel::Timer&>(object));
     case Kernel::HandleType::Thread:
         return std::make_unique<WaitTreeThread>(static_cast<const Kernel::Thread&>(object));
     default:
@@ -348,23 +345,6 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeEvent::GetChildren() const {
     return list;
 }
 
-WaitTreeTimer::WaitTreeTimer(const Kernel::Timer& object) : WaitTreeWaitObject(object) {}
-WaitTreeTimer::~WaitTreeTimer() = default;
-
-std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeTimer::GetChildren() const {
-    std::vector<std::unique_ptr<WaitTreeItem>> list(WaitTreeWaitObject::GetChildren());
-
-    const auto& timer = static_cast<const Kernel::Timer&>(object);
-
-    list.push_back(std::make_unique<WaitTreeText>(
-        tr("reset type = %1").arg(GetResetTypeQString(timer.GetResetType()))));
-    list.push_back(
-        std::make_unique<WaitTreeText>(tr("initial delay = %1").arg(timer.GetInitialDelay())));
-    list.push_back(
-        std::make_unique<WaitTreeText>(tr("interval delay = %1").arg(timer.GetIntervalDelay())));
-    return list;
-}
-
 WaitTreeThreadList::WaitTreeThreadList(const std::vector<Kernel::SharedPtr<Kernel::Thread>>& list)
     : thread_list(list) {}
 WaitTreeThreadList::~WaitTreeThreadList() = default;

src/yuzu/debugger/wait_tree.h

@@ -20,7 +20,6 @@ namespace Kernel {
 class ReadableEvent;
 class WaitObject;
 class Thread;
-class Timer;
 } // namespace Kernel
 
 class WaitTreeThread;
@@ -150,15 +149,6 @@ public:
     std::vector<std::unique_ptr<WaitTreeItem>> GetChildren() const override;
 };
 
-class WaitTreeTimer : public WaitTreeWaitObject {
-    Q_OBJECT
-public:
-    explicit WaitTreeTimer(const Kernel::Timer& object);
-    ~WaitTreeTimer() override;
-
-    std::vector<std::unique_ptr<WaitTreeItem>> GetChildren() const override;
-};
-
 class WaitTreeThreadList : public WaitTreeExpandableItem {
     Q_OBJECT
 public: