common_funcs: silence sign-conversion warnings in MakeMagic()

We can trivially resolve these by casting the characters to unsigned
values and then shifting the bits.

.ci/scripts/linux/upload.sh

@@ -7,9 +7,9 @@ ARCHIVE_NAME="${REV_NAME}.tar.xz"
 COMPRESSION_FLAGS="-cJvf"
 
 if [ "${RELEASE_NAME}" = "mainline" ]; then
-    DIR_NAME="${REV_NAME}_${RELEASE_NAME}"
-else
     DIR_NAME="${REV_NAME}"
+else
+    DIR_NAME="${REV_NAME}_${RELEASE_NAME}"
 fi
 
 mkdir "$DIR_NAME"

.ci/scripts/windows/upload.ps1

@@ -4,9 +4,9 @@ $GITDATE = $(git show -s --date=short --format='%ad') -replace "-",""
 $GITREV = $(git show -s --format='%h')
 
 if ("$BUILD_NAME" -eq "mainline") {
-    $RELEASE_DIST = "yuzu-windows-msvc-$BUILD_NAME"
-} else {
     $RELEASE_DIST = "yuzu-windows-msvc"
+} else {
+    $RELEASE_DIST = "yuzu-windows-msvc-$BUILD_NAME"
 }
 
 $MSVC_BUILD_ZIP = "yuzu-windows-msvc-$GITDATE-$GITREV.zip" -replace " ", ""

.ci/scripts/windows/upload.sh

@@ -7,9 +7,9 @@ ARCHIVE_NAME="${REV_NAME}.tar.gz"
 COMPRESSION_FLAGS="-czvf"
 
 if [ "${RELEASE_NAME}" = "mainline" ]; then
-    DIR_NAME="${REV_NAME}_${RELEASE_NAME}"
-else
     DIR_NAME="${REV_NAME}"
+else
+    DIR_NAME="${REV_NAME}_${RELEASE_NAME}"
 fi
 
 mkdir "$DIR_NAME"

src/common/common_funcs.h

@@ -56,7 +56,7 @@ std::string GetLastErrorMsg();
 namespace Common {
 
 constexpr u32 MakeMagic(char a, char b, char c, char d) {
-    return a | b << 8 | c << 16 | d << 24;
+    return u32(a) | u32(b) << 8 | u32(c) << 16 | u32(d) << 24;
 }
 
 } // namespace Common

src/core/hle/kernel/hle_ipc.cpp

@@ -58,8 +58,7 @@ SharedPtr<WritableEvent> HLERequestContext::SleepClientThread(
     auto& kernel = Core::System::GetInstance().Kernel();
     if (!writable_event) {
         // Create event if not provided
-        const auto pair = WritableEvent::CreateEventPair(kernel, ResetType::Automatic,
-                                                         "HLE Pause Event: " + reason);
+        const auto pair = WritableEvent::CreateEventPair(kernel, "HLE Pause Event: " + reason);
         writable_event = pair.writable;
     }
 

src/core/hle/kernel/object.h

@@ -32,11 +32,6 @@ enum class HandleType : u32 {
     ServerSession,
 };
 
-enum class ResetType {
-    Automatic, ///< Reset automatically on object acquisition
-    Manual,    ///< Never reset automatically
-};
-
 class Object : NonCopyable {
 public:
     explicit Object(KernelCore& kernel);

src/core/hle/kernel/readable_event.cpp

@@ -20,15 +20,13 @@ bool ReadableEvent::ShouldWait(const Thread* thread) const {
 
 void ReadableEvent::Acquire(Thread* thread) {
     ASSERT_MSG(!ShouldWait(thread), "object unavailable!");
-
-    if (reset_type == ResetType::Automatic) {
-        signaled = false;
-    }
 }
 
 void ReadableEvent::Signal() {
-    signaled = true;
-    WakeupAllWaitingThreads();
+    if (!signaled) {
+        signaled = true;
+        WakeupAllWaitingThreads();
+    };
 }
 
 void ReadableEvent::Clear() {

src/core/hle/kernel/readable_event.h

@@ -27,10 +27,6 @@ public:
         return name;
     }
 
-    ResetType GetResetType() const {
-        return reset_type;
-    }
-
     static constexpr HandleType HANDLE_TYPE = HandleType::ReadableEvent;
     HandleType GetHandleType() const override {
         return HANDLE_TYPE;
@@ -55,8 +51,7 @@ private:
 
     void Signal();
 
-    ResetType reset_type;
-    bool signaled;
+    bool signaled{};
 
     std::string name; ///< Name of event (optional)
 };

src/core/hle/kernel/svc.cpp

@@ -2099,7 +2099,7 @@ static ResultCode CreateEvent(Core::System& system, Handle* write_handle, Handle
 
     auto& kernel = system.Kernel();
     const auto [readable_event, writable_event] =
-        WritableEvent::CreateEventPair(kernel, ResetType::Manual, "CreateEvent");
+        WritableEvent::CreateEventPair(kernel, "CreateEvent");
 
     HandleTable& handle_table = kernel.CurrentProcess()->GetHandleTable();
 

src/core/hle/kernel/writable_event.cpp

@@ -15,8 +15,7 @@ namespace Kernel {
 WritableEvent::WritableEvent(KernelCore& kernel) : Object{kernel} {}
 WritableEvent::~WritableEvent() = default;
 
-EventPair WritableEvent::CreateEventPair(KernelCore& kernel, ResetType reset_type,
-                                         std::string name) {
+EventPair WritableEvent::CreateEventPair(KernelCore& kernel, std::string name) {
     SharedPtr<WritableEvent> writable_event(new WritableEvent(kernel));
     SharedPtr<ReadableEvent> readable_event(new ReadableEvent(kernel));
 
@@ -24,7 +23,6 @@ EventPair WritableEvent::CreateEventPair(KernelCore& kernel, ResetType reset_typ
     writable_event->readable = readable_event;
     readable_event->name = name + ":Readable";
     readable_event->signaled = false;
-    readable_event->reset_type = reset_type;
 
     return {std::move(readable_event), std::move(writable_event)};
 }
@@ -33,10 +31,6 @@ SharedPtr<ReadableEvent> WritableEvent::GetReadableEvent() const {
     return readable;
 }
 
-ResetType WritableEvent::GetResetType() const {
-    return readable->reset_type;
-}
-
 void WritableEvent::Signal() {
     readable->Signal();
 }

src/core/hle/kernel/writable_event.h

@@ -24,11 +24,9 @@ public:
     /**
      * Creates an event
      * @param kernel The kernel instance to create this event under.
-     * @param reset_type ResetType describing how to create event
      * @param name Optional name of event
      */
-    static EventPair CreateEventPair(KernelCore& kernel, ResetType reset_type,
-                                     std::string name = "Unknown");
+    static EventPair CreateEventPair(KernelCore& kernel, std::string name = "Unknown");
 
     std::string GetTypeName() const override {
         return "WritableEvent";
@@ -44,8 +42,6 @@ public:
 
     SharedPtr<ReadableEvent> GetReadableEvent() const;
 
-    ResetType GetResetType() const;
-
     void Signal();
     void Clear();
     bool IsSignaled() const;

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

@@ -289,8 +289,8 @@ ISelfController::ISelfController(Core::System& system,
     RegisterHandlers(functions);
 
     auto& kernel = system.Kernel();
-    launchable_event = Kernel::WritableEvent::CreateEventPair(kernel, Kernel::ResetType::Manual,
-                                                              "ISelfController:LaunchableEvent");
+    launchable_event =
+        Kernel::WritableEvent::CreateEventPair(kernel, "ISelfController:LaunchableEvent");
 
     // This event is created by AM on the first time GetAccumulatedSuspendedTickChangedEvent() is
     // called. Yuzu can just create it unconditionally, since it doesn't need to support multiple
@@ -298,7 +298,7 @@ ISelfController::ISelfController(Core::System& system,
     // suspended if the event has previously been created by a call to
     // GetAccumulatedSuspendedTickChangedEvent.
     accumulated_suspended_tick_changed_event = Kernel::WritableEvent::CreateEventPair(
-        kernel, Kernel::ResetType::Manual, "ISelfController:AccumulatedSuspendedTickChangedEvent");
+        kernel, "ISelfController:AccumulatedSuspendedTickChangedEvent");
     accumulated_suspended_tick_changed_event.writable->Signal();
 }
 
@@ -523,10 +523,10 @@ void ISelfController::GetAccumulatedSuspendedTickChangedEvent(Kernel::HLERequest
 }
 
 AppletMessageQueue::AppletMessageQueue(Kernel::KernelCore& kernel) {
-    on_new_message = Kernel::WritableEvent::CreateEventPair(kernel, Kernel::ResetType::Manual,
-                                                            "AMMessageQueue:OnMessageRecieved");
-    on_operation_mode_changed = Kernel::WritableEvent::CreateEventPair(
-        kernel, Kernel::ResetType::Automatic, "AMMessageQueue:OperationModeChanged");
+    on_new_message =
+        Kernel::WritableEvent::CreateEventPair(kernel, "AMMessageQueue:OnMessageRecieved");
+    on_operation_mode_changed =
+        Kernel::WritableEvent::CreateEventPair(kernel, "AMMessageQueue:OperationModeChanged");
 }
 
 AppletMessageQueue::~AppletMessageQueue() = default;
@@ -1091,7 +1091,7 @@ IApplicationFunctions::IApplicationFunctions(Core::System& system_)
 
     auto& kernel = system.Kernel();
     gpu_error_detected_event = Kernel::WritableEvent::CreateEventPair(
-        kernel, Kernel::ResetType::Manual, "IApplicationFunctions:GpuErrorDetectedSystemEvent");
+        kernel, "IApplicationFunctions:GpuErrorDetectedSystemEvent");
 }
 
 IApplicationFunctions::~IApplicationFunctions() = default;

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

@@ -24,12 +24,12 @@
 namespace Service::AM::Applets {
 
 AppletDataBroker::AppletDataBroker(Kernel::KernelCore& kernel) {
-    state_changed_event = Kernel::WritableEvent::CreateEventPair(
-        kernel, Kernel::ResetType::Manual, "ILibraryAppletAccessor:StateChangedEvent");
-    pop_out_data_event = Kernel::WritableEvent::CreateEventPair(
-        kernel, Kernel::ResetType::Manual, "ILibraryAppletAccessor:PopDataOutEvent");
+    state_changed_event =
+        Kernel::WritableEvent::CreateEventPair(kernel, "ILibraryAppletAccessor:StateChangedEvent");
+    pop_out_data_event =
+        Kernel::WritableEvent::CreateEventPair(kernel, "ILibraryAppletAccessor:PopDataOutEvent");
     pop_interactive_out_data_event = Kernel::WritableEvent::CreateEventPair(
-        kernel, Kernel::ResetType::Manual, "ILibraryAppletAccessor:PopInteractiveDataOutEvent");
+        kernel, "ILibraryAppletAccessor:PopInteractiveDataOutEvent");
 }
 
 AppletDataBroker::~AppletDataBroker() = default;

src/core/hle/service/aoc/aoc_u.cpp

@@ -67,8 +67,8 @@ AOC_U::AOC_U(Core::System& system)
     RegisterHandlers(functions);
 
     auto& kernel = system.Kernel();
-    aoc_change_event = Kernel::WritableEvent::CreateEventPair(kernel, Kernel::ResetType::Manual,
-                                                              "GetAddOnContentListChanged:Event");
+    aoc_change_event =
+        Kernel::WritableEvent::CreateEventPair(kernel, "GetAddOnContentListChanged:Event");
 }
 
 AOC_U::~AOC_U() = default;

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

@@ -65,8 +65,8 @@ public:
         RegisterHandlers(functions);
 
         // This is the event handle used to check if the audio buffer was released
-        buffer_event = Kernel::WritableEvent::CreateEventPair(
-            system.Kernel(), Kernel::ResetType::Manual, "IAudioOutBufferReleased");
+        buffer_event =
+            Kernel::WritableEvent::CreateEventPair(system.Kernel(), "IAudioOutBufferReleased");
 
         stream = audio_core.OpenStream(system.CoreTiming(), audio_params.sample_rate,
                                        audio_params.channel_count, std::move(unique_name),

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

@@ -47,8 +47,8 @@ public:
         // clang-format on
         RegisterHandlers(functions);
 
-        system_event = Kernel::WritableEvent::CreateEventPair(
-            system.Kernel(), Kernel::ResetType::Manual, "IAudioRenderer:SystemEvent");
+        system_event =
+            Kernel::WritableEvent::CreateEventPair(system.Kernel(), "IAudioRenderer:SystemEvent");
         renderer = std::make_unique<AudioCore::AudioRenderer>(
             system.CoreTiming(), audren_params, system_event.writable, instance_number);
     }
@@ -180,17 +180,17 @@ public:
         RegisterHandlers(functions);
 
         auto& kernel = system.Kernel();
-        buffer_event = Kernel::WritableEvent::CreateEventPair(kernel, Kernel::ResetType::Automatic,
-                                                              "IAudioOutBufferReleasedEvent");
+        buffer_event =
+            Kernel::WritableEvent::CreateEventPair(kernel, "IAudioOutBufferReleasedEvent");
 
         // Should be similar to audio_output_device_switch_event
         audio_input_device_switch_event = Kernel::WritableEvent::CreateEventPair(
-            kernel, Kernel::ResetType::Automatic, "IAudioDevice:AudioInputDeviceSwitchedEvent");
+            kernel, "IAudioDevice:AudioInputDeviceSwitchedEvent");
 
         // Should only be signalled when an audio output device has been changed, example: speaker
         // to headset
         audio_output_device_switch_event = Kernel::WritableEvent::CreateEventPair(
-            kernel, Kernel::ResetType::Automatic, "IAudioDevice:AudioOutputDeviceSwitchedEvent");
+            kernel, "IAudioDevice:AudioOutputDeviceSwitchedEvent");
     }
 
 private:

src/core/hle/service/bcat/backend/backend.cpp

@@ -13,8 +13,7 @@ namespace Service::BCAT {
 ProgressServiceBackend::ProgressServiceBackend(Kernel::KernelCore& kernel,
                                                std::string_view event_name) {
     event = Kernel::WritableEvent::CreateEventPair(
-        kernel, Kernel::ResetType::Automatic,
-        std::string("ProgressServiceBackend:UpdateEvent:").append(event_name));
+        kernel, std::string("ProgressServiceBackend:UpdateEvent:").append(event_name));
 }
 
 Kernel::SharedPtr<Kernel::ReadableEvent> ProgressServiceBackend::GetEvent() const {

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

@@ -34,8 +34,7 @@ public:
         RegisterHandlers(functions);
 
         auto& kernel = system.Kernel();
-        register_event = Kernel::WritableEvent::CreateEventPair(
-            kernel, Kernel::ResetType::Automatic, "BT:RegisterEvent");
+        register_event = Kernel::WritableEvent::CreateEventPair(kernel, "BT:RegisterEvent");
     }
 
 private:

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

@@ -57,14 +57,12 @@ public:
         RegisterHandlers(functions);
 
         auto& kernel = system.Kernel();
-        scan_event = Kernel::WritableEvent::CreateEventPair(kernel, Kernel::ResetType::Automatic,
-                                                            "IBtmUserCore:ScanEvent");
-        connection_event = Kernel::WritableEvent::CreateEventPair(
-            kernel, Kernel::ResetType::Automatic, "IBtmUserCore:ConnectionEvent");
-        service_discovery = Kernel::WritableEvent::CreateEventPair(
-            kernel, Kernel::ResetType::Automatic, "IBtmUserCore:Discovery");
-        config_event = Kernel::WritableEvent::CreateEventPair(kernel, Kernel::ResetType::Automatic,
-                                                              "IBtmUserCore:ConfigEvent");
+        scan_event = Kernel::WritableEvent::CreateEventPair(kernel, "IBtmUserCore:ScanEvent");
+        connection_event =
+            Kernel::WritableEvent::CreateEventPair(kernel, "IBtmUserCore:ConnectionEvent");
+        service_discovery =
+            Kernel::WritableEvent::CreateEventPair(kernel, "IBtmUserCore:Discovery");
+        config_event = Kernel::WritableEvent::CreateEventPair(kernel, "IBtmUserCore:ConfigEvent");
     }
 
 private:

src/core/hle/service/friend/friend.cpp

@@ -162,7 +162,7 @@ public:
         RegisterHandlers(functions);
 
         notification_event = Kernel::WritableEvent::CreateEventPair(
-            system.Kernel(), Kernel::ResetType::Manual, "INotificationService:NotifyEvent");
+            system.Kernel(), "INotificationService:NotifyEvent");
     }
 
 private:

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

@@ -174,7 +174,7 @@ void Controller_NPad::OnInit() {
     auto& kernel = system.Kernel();
     for (std::size_t i = 0; i < styleset_changed_events.size(); i++) {
         styleset_changed_events[i] = Kernel::WritableEvent::CreateEventPair(
-            kernel, Kernel::ResetType::Manual, fmt::format("npad:NpadStyleSetChanged_{}", i));
+            kernel, fmt::format("npad:NpadStyleSetChanged_{}", i));
     }
 
     if (!IsControllerActivated()) {

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

@@ -26,8 +26,7 @@ constexpr ResultCode ERR_NO_APPLICATION_AREA(ErrorModule::NFP, 152);
 Module::Interface::Interface(std::shared_ptr<Module> module, Core::System& system, const char* name)
     : ServiceFramework(name), module(std::move(module)), system(system) {
     auto& kernel = system.Kernel();
-    nfc_tag_load = Kernel::WritableEvent::CreateEventPair(kernel, Kernel::ResetType::Automatic,
-                                                          "IUser:NFCTagDetected");
+    nfc_tag_load = Kernel::WritableEvent::CreateEventPair(kernel, "IUser:NFCTagDetected");
 }
 
 Module::Interface::~Interface() = default;
@@ -66,10 +65,9 @@ public:
         RegisterHandlers(functions);
 
         auto& kernel = system.Kernel();
-        deactivate_event = Kernel::WritableEvent::CreateEventPair(
-            kernel, Kernel::ResetType::Automatic, "IUser:DeactivateEvent");
-        availability_change_event = Kernel::WritableEvent::CreateEventPair(
-            kernel, Kernel::ResetType::Automatic, "IUser:AvailabilityChangeEvent");
+        deactivate_event = Kernel::WritableEvent::CreateEventPair(kernel, "IUser:DeactivateEvent");
+        availability_change_event =
+            Kernel::WritableEvent::CreateEventPair(kernel, "IUser:AvailabilityChangeEvent");
     }
 
 private:

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

@@ -69,10 +69,8 @@ public:
         RegisterHandlers(functions);
 
         auto& kernel = system.Kernel();
-        event1 = Kernel::WritableEvent::CreateEventPair(kernel, Kernel::ResetType::Automatic,
-                                                        "IRequest:Event1");
-        event2 = Kernel::WritableEvent::CreateEventPair(kernel, Kernel::ResetType::Automatic,
-                                                        "IRequest:Event2");
+        event1 = Kernel::WritableEvent::CreateEventPair(kernel, "IRequest:Event1");
+        event2 = Kernel::WritableEvent::CreateEventPair(kernel, "IRequest:Event2");
     }
 
 private:

src/core/hle/service/nim/nim.cpp

@@ -141,8 +141,7 @@ public:
 
         auto& kernel = system.Kernel();
         finished_event = Kernel::WritableEvent::CreateEventPair(
-            kernel, Kernel::ResetType::Automatic,
-            "IEnsureNetworkClockAvailabilityService:FinishEvent");
+            kernel, "IEnsureNetworkClockAvailabilityService:FinishEvent");
     }
 
 private:

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

@@ -40,8 +40,7 @@ Module::Module(Core::System& system) {
     auto& kernel = system.Kernel();
     for (u32 i = 0; i < MaxNvEvents; i++) {
         std::string event_label = fmt::format("NVDRV::NvEvent_{}", i);
-        events_interface.events[i] =
-            Kernel::WritableEvent::CreateEventPair(kernel, Kernel::ResetType::Manual, event_label);
+        events_interface.events[i] = Kernel::WritableEvent::CreateEventPair(kernel, event_label);
         events_interface.status[i] = EventState::Free;
         events_interface.registered[i] = false;
     }

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

@@ -16,8 +16,7 @@ namespace Service::NVFlinger {
 
 BufferQueue::BufferQueue(Kernel::KernelCore& kernel, u32 id, u64 layer_id)
     : id(id), layer_id(layer_id) {
-    buffer_wait_event = Kernel::WritableEvent::CreateEventPair(kernel, Kernel::ResetType::Manual,
-                                                               "BufferQueue NativeHandle");
+    buffer_wait_event = Kernel::WritableEvent::CreateEventPair(kernel, "BufferQueue NativeHandle");
 }
 
 BufferQueue::~BufferQueue() = default;

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

@@ -17,8 +17,8 @@ namespace Service::VI {
 
 Display::Display(u64 id, std::string name, Core::System& system) : id{id}, name{std::move(name)} {
     auto& kernel = system.Kernel();
-    vsync_event = Kernel::WritableEvent::CreateEventPair(kernel, Kernel::ResetType::Manual,
-                                                         fmt::format("Display VSync Event {}", id));
+    vsync_event =
+        Kernel::WritableEvent::CreateEventPair(kernel, fmt::format("Display VSync Event {}", id));
 }
 
 Display::~Display() = default;

src/video_core/buffer_cache/buffer_cache.h

@@ -12,6 +12,10 @@
 #include <utility>
 #include <vector>
 
+#include <boost/icl/interval_map.hpp>
+#include <boost/icl/interval_set.hpp>
+#include <boost/range/iterator_range.hpp>
+
 #include "common/alignment.h"
 #include "common/common_types.h"
 #include "core/core.h"
@@ -30,7 +34,7 @@ public:
     using BufferInfo = std::pair<const TBufferType*, u64>;
 
     BufferInfo UploadMemory(GPUVAddr gpu_addr, std::size_t size, std::size_t alignment = 4,
-                            bool is_written = false) {
+                            bool is_written = false, bool use_fast_cbuf = false) {
         std::lock_guard lock{mutex};
 
         auto& memory_manager = system.GPU().MemoryManager();
@@ -43,9 +47,13 @@ public:
         // Cache management is a big overhead, so only cache entries with a given size.
         // TODO: Figure out which size is the best for given games.
         constexpr std::size_t max_stream_size = 0x800;
-        if (size < max_stream_size) {
+        if (use_fast_cbuf || size < max_stream_size) {
             if (!is_written && !IsRegionWritten(cache_addr, cache_addr + size - 1)) {
-                return StreamBufferUpload(host_ptr, size, alignment);
+                if (use_fast_cbuf) {
+                    return ConstBufferUpload(host_ptr, size);
+                } else {
+                    return StreamBufferUpload(host_ptr, size, alignment);
+                }
             }
         }
 
@@ -152,6 +160,10 @@ protected:
     virtual void CopyBlock(const TBuffer& src, const TBuffer& dst, std::size_t src_offset,
                            std::size_t dst_offset, std::size_t size) = 0;
 
+    virtual BufferInfo ConstBufferUpload(const void* raw_pointer, std::size_t size) {
+        return {};
+    }
+
     /// Register an object into the cache
     void Register(const MapInterval& new_map, bool inherit_written = false) {
         const CacheAddr cache_ptr = new_map->GetStart();

src/video_core/renderer_opengl/gl_buffer_cache.cpp

@@ -8,13 +8,17 @@
 
 #include "common/assert.h"
 #include "common/microprofile.h"
+#include "video_core/engines/maxwell_3d.h"
 #include "video_core/rasterizer_interface.h"
 #include "video_core/renderer_opengl/gl_buffer_cache.h"
+#include "video_core/renderer_opengl/gl_device.h"
 #include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
 
 namespace OpenGL {
 
+using Maxwell = Tegra::Engines::Maxwell3D::Regs;
+
 MICROPROFILE_DEFINE(OpenGL_Buffer_Download, "OpenGL", "Buffer Download", MP_RGB(192, 192, 128));
 
 CachedBufferBlock::CachedBufferBlock(CacheAddr cache_addr, const std::size_t size)
@@ -26,11 +30,22 @@ CachedBufferBlock::CachedBufferBlock(CacheAddr cache_addr, const std::size_t siz
 CachedBufferBlock::~CachedBufferBlock() = default;
 
 OGLBufferCache::OGLBufferCache(RasterizerOpenGL& rasterizer, Core::System& system,
-                               std::size_t stream_size)
-    : VideoCommon::BufferCache<Buffer, GLuint, OGLStreamBuffer>{
-          rasterizer, system, std::make_unique<OGLStreamBuffer>(stream_size, true)} {}
+                               const Device& device, std::size_t stream_size)
+    : GenericBufferCache{rasterizer, system, std::make_unique<OGLStreamBuffer>(stream_size, true)} {
+    if (!device.HasFastBufferSubData()) {
+        return;
+    }
 
-OGLBufferCache::~OGLBufferCache() = default;
+    static constexpr auto size = static_cast<GLsizeiptr>(Maxwell::MaxConstBufferSize);
+    glCreateBuffers(static_cast<GLsizei>(std::size(cbufs)), std::data(cbufs));
+    for (const GLuint cbuf : cbufs) {
+        glNamedBufferData(cbuf, size, nullptr, GL_STREAM_DRAW);
+    }
+}
+
+OGLBufferCache::~OGLBufferCache() {
+    glDeleteBuffers(static_cast<GLsizei>(std::size(cbufs)), std::data(cbufs));
+}
 
 Buffer OGLBufferCache::CreateBlock(CacheAddr cache_addr, std::size_t size) {
     return std::make_shared<CachedBufferBlock>(cache_addr, size);
@@ -69,4 +84,12 @@ void OGLBufferCache::CopyBlock(const Buffer& src, const Buffer& dst, std::size_t
                              static_cast<GLsizeiptr>(size));
 }
 
+OGLBufferCache::BufferInfo OGLBufferCache::ConstBufferUpload(const void* raw_pointer,
+                                                             std::size_t size) {
+    DEBUG_ASSERT(cbuf_cursor < std::size(cbufs));
+    const GLuint& cbuf = cbufs[cbuf_cursor++];
+    glNamedBufferSubData(cbuf, 0, static_cast<GLsizeiptr>(size), raw_pointer);
+    return {&cbuf, 0};
+}
+
 } // namespace OpenGL

src/video_core/renderer_opengl/gl_buffer_cache.h

@@ -4,10 +4,12 @@
 
 #pragma once
 
+#include <array>
 #include <memory>
 
 #include "common/common_types.h"
 #include "video_core/buffer_cache/buffer_cache.h"
+#include "video_core/engines/maxwell_3d.h"
 #include "video_core/rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
 #include "video_core/renderer_opengl/gl_stream_buffer.h"
@@ -18,12 +20,14 @@ class System;
 
 namespace OpenGL {
 
+class Device;
 class OGLStreamBuffer;
 class RasterizerOpenGL;
 
 class CachedBufferBlock;
 
 using Buffer = std::shared_ptr<CachedBufferBlock>;
+using GenericBufferCache = VideoCommon::BufferCache<Buffer, GLuint, OGLStreamBuffer>;
 
 class CachedBufferBlock : public VideoCommon::BufferBlock {
 public:
@@ -38,14 +42,18 @@ private:
     OGLBuffer gl_buffer{};
 };
 
-class OGLBufferCache final : public VideoCommon::BufferCache<Buffer, GLuint, OGLStreamBuffer> {
+class OGLBufferCache final : public GenericBufferCache {
 public:
     explicit OGLBufferCache(RasterizerOpenGL& rasterizer, Core::System& system,
-                            std::size_t stream_size);
+                            const Device& device, std::size_t stream_size);
     ~OGLBufferCache();
 
     const GLuint* GetEmptyBuffer(std::size_t) override;
 
+    void Acquire() noexcept {
+        cbuf_cursor = 0;
+    }
+
 protected:
     Buffer CreateBlock(CacheAddr cache_addr, std::size_t size) override;
 
@@ -61,6 +69,14 @@ protected:
 
     void CopyBlock(const Buffer& src, const Buffer& dst, std::size_t src_offset,
                    std::size_t dst_offset, std::size_t size) override;
+
+    BufferInfo ConstBufferUpload(const void* raw_pointer, std::size_t size) override;
+
+private:
+    std::size_t cbuf_cursor = 0;
+    std::array<GLuint, Tegra::Engines::Maxwell3D::Regs::MaxConstBuffers *
+                           Tegra::Engines::Maxwell3D::Regs::MaxShaderProgram>
+        cbufs;
 };
 
 } // namespace OpenGL

src/video_core/renderer_opengl/gl_device.cpp

@@ -51,8 +51,11 @@ bool HasExtension(const std::vector<std::string_view>& images, std::string_view
 } // Anonymous namespace
 
 Device::Device() {
+    const std::string_view vendor = reinterpret_cast<const char*>(glGetString(GL_VENDOR));
     const std::vector extensions = GetExtensions();
 
+    const bool is_nvidia = vendor == "NVIDIA Corporation";
+
     uniform_buffer_alignment = GetInteger<std::size_t>(GL_UNIFORM_BUFFER_OFFSET_ALIGNMENT);
     shader_storage_alignment = GetInteger<std::size_t>(GL_SHADER_STORAGE_BUFFER_OFFSET_ALIGNMENT);
     max_vertex_attributes = GetInteger<u32>(GL_MAX_VERTEX_ATTRIBS);
@@ -64,6 +67,7 @@ Device::Device() {
     has_variable_aoffi = TestVariableAoffi();
     has_component_indexing_bug = TestComponentIndexingBug();
     has_precise_bug = TestPreciseBug();
+    has_fast_buffer_sub_data = is_nvidia;
 
     LOG_INFO(Render_OpenGL, "Renderer_VariableAOFFI: {}", has_variable_aoffi);
     LOG_INFO(Render_OpenGL, "Renderer_ComponentIndexingBug: {}", has_component_indexing_bug);

src/video_core/renderer_opengl/gl_device.h

@@ -54,6 +54,10 @@ public:
         return has_precise_bug;
     }
 
+    bool HasFastBufferSubData() const {
+        return has_fast_buffer_sub_data;
+    }
+
 private:
     static bool TestVariableAoffi();
     static bool TestComponentIndexingBug();
@@ -69,6 +73,7 @@ private:
     bool has_variable_aoffi{};
     bool has_component_indexing_bug{};
     bool has_precise_bug{};
+    bool has_fast_buffer_sub_data{};
 };
 
 } // namespace OpenGL

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -67,7 +67,7 @@ static std::size_t GetConstBufferSize(const Tegra::Engines::ConstBufferInfo& buf
 RasterizerOpenGL::RasterizerOpenGL(Core::System& system, Core::Frontend::EmuWindow& emu_window,
                                    ScreenInfo& info)
     : texture_cache{system, *this, device}, shader_cache{*this, system, emu_window, device},
-      system{system}, screen_info{info}, buffer_cache{*this, system, STREAM_BUFFER_SIZE} {
+      system{system}, screen_info{info}, buffer_cache{*this, system, device, STREAM_BUFFER_SIZE} {
     shader_program_manager = std::make_unique<GLShader::ProgramManager>();
     state.draw.shader_program = 0;
     state.Apply();
@@ -558,6 +558,8 @@ void RasterizerOpenGL::DrawPrelude() {
     SyncPolygonOffset();
     SyncAlphaTest();
 
+    buffer_cache.Acquire();
+
     // Draw the vertex batch
     const bool is_indexed = accelerate_draw == AccelDraw::Indexed;
 
@@ -879,7 +881,8 @@ void RasterizerOpenGL::SetupConstBuffer(const Tegra::Engines::ConstBufferInfo& b
     const std::size_t size = Common::AlignUp(GetConstBufferSize(buffer, entry), sizeof(GLvec4));
 
     const auto alignment = device.GetUniformBufferAlignment();
-    const auto [cbuf, offset] = buffer_cache.UploadMemory(buffer.address, size, alignment);
+    const auto [cbuf, offset] = buffer_cache.UploadMemory(buffer.address, size, alignment, false,
+                                                          device.HasFastBufferSubData());
     bind_ubo_pushbuffer.Push(cbuf, offset, size);
 }
 
@@ -935,10 +938,9 @@ TextureBufferUsage RasterizerOpenGL::SetupDrawTextures(Maxwell::ShaderStage stag
             if (!entry.IsBindless()) {
                 return maxwell3d.GetStageTexture(stage, entry.GetOffset());
             }
-            const auto cbuf = entry.GetBindlessCBuf();
-            Tegra::Texture::TextureHandle tex_handle;
-            Tegra::Engines::ShaderType shader_type = static_cast<Tegra::Engines::ShaderType>(stage);
-            tex_handle.raw = maxwell3d.AccessConstBuffer32(shader_type, cbuf.first, cbuf.second);
+            const auto shader_type = static_cast<Tegra::Engines::ShaderType>(stage);
+            const Tegra::Texture::TextureHandle tex_handle =
+                maxwell3d.AccessConstBuffer32(shader_type, entry.GetBuffer(), entry.GetOffset());
             return maxwell3d.GetTextureInfo(tex_handle);
         }();
 
@@ -966,10 +968,8 @@ TextureBufferUsage RasterizerOpenGL::SetupComputeTextures(const Shader& kernel)
             if (!entry.IsBindless()) {
                 return compute.GetTexture(entry.GetOffset());
             }
-            const auto cbuf = entry.GetBindlessCBuf();
-            Tegra::Texture::TextureHandle tex_handle;
-            tex_handle.raw = compute.AccessConstBuffer32(Tegra::Engines::ShaderType::Compute,
-                                                         cbuf.first, cbuf.second);
+            const Tegra::Texture::TextureHandle tex_handle = compute.AccessConstBuffer32(
+                Tegra::Engines::ShaderType::Compute, entry.GetBuffer(), entry.GetOffset());
             return compute.GetTextureInfo(tex_handle);
         }();
 
@@ -1012,10 +1012,8 @@ void RasterizerOpenGL::SetupComputeImages(const Shader& shader) {
             if (!entry.IsBindless()) {
                 return compute.GetTexture(entry.GetOffset()).tic;
             }
-            const auto cbuf = entry.GetBindlessCBuf();
-            Tegra::Texture::TextureHandle tex_handle;
-            tex_handle.raw = compute.AccessConstBuffer32(Tegra::Engines::ShaderType::Compute,
-                                                         cbuf.first, cbuf.second);
+            const Tegra::Texture::TextureHandle tex_handle = compute.AccessConstBuffer32(
+                Tegra::Engines::ShaderType::Compute, entry.GetBuffer(), entry.GetOffset());
             return compute.GetTextureInfo(tex_handle).tic;
         }();
         SetupImage(bindpoint, tic, entry);

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -394,7 +394,8 @@ Shader CachedShader::CreateStageFromMemory(const ShaderParameters& params,
     params.disk_cache.SaveRaw(ShaderDiskCacheRaw(
         params.unique_identifier, GetProgramType(program_type), program_code, program_code_b));
 
-    ConstBufferLocker locker(GetEnginesShaderType(GetProgramType(program_type)));
+    ConstBufferLocker locker(GetEnginesShaderType(GetProgramType(program_type)),
+                             params.system.GPU().Maxwell3D());
     const ShaderIR ir(program_code, STAGE_MAIN_OFFSET, COMPILER_SETTINGS, locker);
     // TODO(Rodrigo): Handle VertexA shaders
     // std::optional<ShaderIR> ir_b;
@@ -410,7 +411,8 @@ Shader CachedShader::CreateKernelFromMemory(const ShaderParameters& params, Prog
     params.disk_cache.SaveRaw(
         ShaderDiskCacheRaw(params.unique_identifier, ProgramType::Compute, code));
 
-    ConstBufferLocker locker(Tegra::Engines::ShaderType::Compute);
+    ConstBufferLocker locker(Tegra::Engines::ShaderType::Compute,
+                             params.system.GPU().KeplerCompute());
     const ShaderIR ir(code, KERNEL_MAIN_OFFSET, COMPILER_SETTINGS, locker);
     return std::shared_ptr<CachedShader>(new CachedShader(
         params, ProgramType::Compute, GLShader::GetEntries(ir), std::move(code), {}));

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -735,7 +735,7 @@ private:
 
     void DeclareImages() {
         const auto& images{ir.GetImages()};
-        for (const auto& [offset, image] : images) {
+        for (const auto& image : images) {
             std::string qualifier = "coherent volatile";
             if (image.IsRead() && !image.IsWritten()) {
                 qualifier += " readonly";
@@ -1670,7 +1670,7 @@ private:
 
         const auto type = meta->sampler.IsShadow() ? Type::Float : Type::Int;
         return {GenerateTexture(operation, "Gather",
-                                {TextureArgument{type, meta->component}, TextureAoffi{}}) +
+                                {TextureAoffi{}, TextureArgument{type, meta->component}}) +
                     GetSwizzle(meta->element),
                 Type::Float};
     }
@@ -2466,16 +2466,16 @@ ShaderEntries GetEntries(const VideoCommon::Shader::ShaderIR& ir) {
         entries.const_buffers.emplace_back(cbuf.second.GetMaxOffset(), cbuf.second.IsIndirect(),
                                            cbuf.first);
     }
-    for (const auto& sampler : ir.GetSamplers()) {
-        entries.samplers.emplace_back(sampler);
-    }
-    for (const auto& [offset, image] : ir.GetImages()) {
-        entries.images.emplace_back(image);
-    }
     for (const auto& [base, usage] : ir.GetGlobalMemory()) {
         entries.global_memory_entries.emplace_back(base.cbuf_index, base.cbuf_offset, usage.is_read,
                                                    usage.is_written);
     }
+    for (const auto& sampler : ir.GetSamplers()) {
+        entries.samplers.emplace_back(sampler);
+    }
+    for (const auto& image : ir.GetImages()) {
+        entries.images.emplace_back(image);
+    }
     entries.clip_distances = ir.GetClipDistances();
     entries.shader_length = ir.GetLength();
     return entries;

src/video_core/renderer_opengl/gl_shader_decompiler.h

@@ -82,10 +82,9 @@ private:
 
 struct ShaderEntries {
     std::vector<ConstBufferEntry> const_buffers;
-    std::vector<SamplerEntry> samplers;
-    std::vector<SamplerEntry> bindless_samplers;
-    std::vector<ImageEntry> images;
     std::vector<GlobalMemoryEntry> global_memory_entries;
+    std::vector<SamplerEntry> samplers;
+    std::vector<ImageEntry> images;
     std::array<bool, Maxwell::NumClipDistances> clip_distances{};
     std::size_t shader_length{};
 };

src/video_core/shader/control_flow.cpp

@@ -16,7 +16,9 @@
 #include "video_core/shader/shader_ir.h"
 
 namespace VideoCommon::Shader {
+
 namespace {
+
 using Tegra::Shader::Instruction;
 using Tegra::Shader::OpCode;
 
@@ -68,15 +70,15 @@ struct CFGRebuildState {
     const ProgramCode& program_code;
     ConstBufferLocker& locker;
     u32 start{};
-    std::vector<BlockInfo> block_info{};
-    std::list<u32> inspect_queries{};
-    std::list<Query> queries{};
-    std::unordered_map<u32, u32> registered{};
-    std::set<u32> labels{};
-    std::map<u32, u32> ssy_labels{};
-    std::map<u32, u32> pbk_labels{};
-    std::unordered_map<u32, BlockStack> stacks{};
-    ASTManager* manager;
+    std::vector<BlockInfo> block_info;
+    std::list<u32> inspect_queries;
+    std::list<Query> queries;
+    std::unordered_map<u32, u32> registered;
+    std::set<u32> labels;
+    std::map<u32, u32> ssy_labels;
+    std::map<u32, u32> pbk_labels;
+    std::unordered_map<u32, BlockStack> stacks;
+    ASTManager* manager{};
 };
 
 enum class BlockCollision : u32 { None, Found, Inside };
@@ -109,7 +111,7 @@ BlockInfo& CreateBlockInfo(CFGRebuildState& state, u32 start, u32 end) {
 }
 
 Pred GetPredicate(u32 index, bool negated) {
-    return static_cast<Pred>(index + (negated ? 8 : 0));
+    return static_cast<Pred>(static_cast<u64>(index) + (negated ? 8ULL : 0ULL));
 }
 
 /**
@@ -136,15 +138,13 @@ struct BranchIndirectInfo {
     s32 relative_position{};
 };
 
-std::optional<BranchIndirectInfo> TrackBranchIndirectInfo(const CFGRebuildState& state,
-                                                          u32 start_address, u32 current_position) {
-    const u32 shader_start = state.start;
-    u32 pos = current_position;
-    BranchIndirectInfo result{};
-    u64 track_register = 0;
+struct BufferInfo {
+    u32 index;
+    u32 offset;
+};
 
-    // Step 0 Get BRX Info
-    const Instruction instr = {state.program_code[pos]};
+std::optional<std::pair<s32, u64>> GetBRXInfo(const CFGRebuildState& state, u32& pos) {
+    const Instruction instr = state.program_code[pos];
     const auto opcode = OpCode::Decode(instr);
     if (opcode->get().GetId() != OpCode::Id::BRX) {
         return std::nullopt;
@@ -152,86 +152,94 @@ std::optional<BranchIndirectInfo> TrackBranchIndirectInfo(const CFGRebuildState&
     if (instr.brx.constant_buffer != 0) {
         return std::nullopt;
     }
-    track_register = instr.gpr8.Value();
-    result.relative_position = instr.brx.GetBranchExtend();
-    pos--;
-    bool found_track = false;
+    --pos;
+    return std::make_pair(instr.brx.GetBranchExtend(), instr.gpr8.Value());
+}
 
-    // Step 1 Track LDC
-    while (pos >= shader_start) {
-        if (IsSchedInstruction(pos, shader_start)) {
-            pos--;
-            continue;
-        }
-        const Instruction instr = {state.program_code[pos]};
-        const auto opcode = OpCode::Decode(instr);
-        if (opcode->get().GetId() == OpCode::Id::LD_C) {
-            if (instr.gpr0.Value() == track_register &&
-                instr.ld_c.type.Value() == Tegra::Shader::UniformType::Single) {
-                result.buffer = instr.cbuf36.index.Value();
-                result.offset = static_cast<u32>(instr.cbuf36.GetOffset());
-                track_register = instr.gpr8.Value();
-                pos--;
-                found_track = true;
-                break;
-            }
-        }
-        pos--;
-    }
-
-    if (!found_track) {
-        return std::nullopt;
-    }
-    found_track = false;
-
-    // Step 2 Track SHL
-    while (pos >= shader_start) {
-        if (IsSchedInstruction(pos, shader_start)) {
-            pos--;
+template <typename Result, typename TestCallable, typename PackCallable>
+// requires std::predicate<TestCallable, Instruction, const OpCode::Matcher&>
+// requires std::invocable<PackCallable, Instruction, const OpCode::Matcher&>
+std::optional<Result> TrackInstruction(const CFGRebuildState& state, u32& pos, TestCallable test,
+                                       PackCallable pack) {
+    for (; pos >= state.start; --pos) {
+        if (IsSchedInstruction(pos, state.start)) {
             continue;
         }
         const Instruction instr = state.program_code[pos];
         const auto opcode = OpCode::Decode(instr);
-        if (opcode->get().GetId() == OpCode::Id::SHL_IMM) {
-            if (instr.gpr0.Value() == track_register) {
-                track_register = instr.gpr8.Value();
-                pos--;
-                found_track = true;
-                break;
-            }
-        }
-        pos--;
-    }
-
-    if (!found_track) {
-        return std::nullopt;
-    }
-    found_track = false;
-
-    // Step 3 Track IMNMX
-    while (pos >= shader_start) {
-        if (IsSchedInstruction(pos, shader_start)) {
-            pos--;
+        if (!opcode) {
             continue;
         }
-        const Instruction instr = state.program_code[pos];
-        const auto opcode = OpCode::Decode(instr);
-        if (opcode->get().GetId() == OpCode::Id::IMNMX_IMM) {
-            if (instr.gpr0.Value() == track_register) {
-                track_register = instr.gpr8.Value();
-                result.entries = instr.alu.GetSignedImm20_20() + 1;
-                pos--;
-                found_track = true;
-                break;
-            }
+        if (test(instr, opcode->get())) {
+            --pos;
+            return std::make_optional(pack(instr, opcode->get()));
         }
-        pos--;
     }
+    return std::nullopt;
+}
 
-    if (!found_track) {
+std::optional<std::pair<BufferInfo, u64>> TrackLDC(const CFGRebuildState& state, u32& pos,
+                                                   u64 brx_tracked_register) {
+    return TrackInstruction<std::pair<BufferInfo, u64>>(
+        state, pos,
+        [brx_tracked_register](auto instr, const auto& opcode) {
+            return opcode.GetId() == OpCode::Id::LD_C &&
+                   instr.gpr0.Value() == brx_tracked_register &&
+                   instr.ld_c.type.Value() == Tegra::Shader::UniformType::Single;
+        },
+        [](auto instr, const auto& opcode) {
+            const BufferInfo info = {static_cast<u32>(instr.cbuf36.index.Value()),
+                                     static_cast<u32>(instr.cbuf36.GetOffset())};
+            return std::make_pair(info, instr.gpr8.Value());
+        });
+}
+
+std::optional<u64> TrackSHLRegister(const CFGRebuildState& state, u32& pos,
+                                    u64 ldc_tracked_register) {
+    return TrackInstruction<u64>(state, pos,
+                                 [ldc_tracked_register](auto instr, const auto& opcode) {
+                                     return opcode.GetId() == OpCode::Id::SHL_IMM &&
+                                            instr.gpr0.Value() == ldc_tracked_register;
+                                 },
+                                 [](auto instr, const auto&) { return instr.gpr8.Value(); });
+}
+
+std::optional<u32> TrackIMNMXValue(const CFGRebuildState& state, u32& pos,
+                                   u64 shl_tracked_register) {
+    return TrackInstruction<u32>(state, pos,
+                                 [shl_tracked_register](auto instr, const auto& opcode) {
+                                     return opcode.GetId() == OpCode::Id::IMNMX_IMM &&
+                                            instr.gpr0.Value() == shl_tracked_register;
+                                 },
+                                 [](auto instr, const auto&) {
+                                     return static_cast<u32>(instr.alu.GetSignedImm20_20() + 1);
+                                 });
+}
+
+std::optional<BranchIndirectInfo> TrackBranchIndirectInfo(const CFGRebuildState& state, u32 pos) {
+    const auto brx_info = GetBRXInfo(state, pos);
+    if (!brx_info) {
         return std::nullopt;
     }
-    return result;
+    const auto [relative_position, brx_tracked_register] = *brx_info;
+
+    const auto ldc_info = TrackLDC(state, pos, brx_tracked_register);
+    if (!ldc_info) {
+        return std::nullopt;
+    }
+    const auto [buffer_info, ldc_tracked_register] = *ldc_info;
+
+    const auto shl_tracked_register = TrackSHLRegister(state, pos, ldc_tracked_register);
+    if (!shl_tracked_register) {
+        return std::nullopt;
+    }
+
+    const auto entries = TrackIMNMXValue(state, pos, *shl_tracked_register);
+    if (!entries) {
+        return std::nullopt;
+    }
+
+    return BranchIndirectInfo{buffer_info.index, buffer_info.offset, *entries, relative_position};
 }
 
 std::pair<ParseResult, ParseInfo> ParseCode(CFGRebuildState& state, u32 address) {
@@ -420,30 +428,30 @@ std::pair<ParseResult, ParseInfo> ParseCode(CFGRebuildState& state, u32 address)
             break;
         }
         case OpCode::Id::BRX: {
-            auto tmp = TrackBranchIndirectInfo(state, address, offset);
-            if (tmp) {
-                auto result = *tmp;
-                std::vector<CaseBranch> branches{};
-                s32 pc_target = offset + result.relative_position;
-                for (u32 i = 0; i < result.entries; i++) {
-                    auto k = state.locker.ObtainKey(result.buffer, result.offset + i * 4);
-                    if (!k) {
-                        return {ParseResult::AbnormalFlow, parse_info};
-                    }
-                    u32 value = *k;
-                    u32 target = static_cast<u32>((value >> 3) + pc_target);
-                    insert_label(state, target);
-                    branches.emplace_back(value, target);
-                }
-                parse_info.end_address = offset;
-                parse_info.branch_info = MakeBranchInfo<MultiBranch>(
-                    static_cast<u32>(instr.gpr8.Value()), std::move(branches));
-
-                return {ParseResult::ControlCaught, parse_info};
-            } else {
+            const auto tmp = TrackBranchIndirectInfo(state, offset);
+            if (!tmp) {
                 LOG_WARNING(HW_GPU, "BRX Track Unsuccesful");
+                return {ParseResult::AbnormalFlow, parse_info};
             }
-            return {ParseResult::AbnormalFlow, parse_info};
+
+            const auto result = *tmp;
+            const s32 pc_target = offset + result.relative_position;
+            std::vector<CaseBranch> branches;
+            for (u32 i = 0; i < result.entries; i++) {
+                auto key = state.locker.ObtainKey(result.buffer, result.offset + i * 4);
+                if (!key) {
+                    return {ParseResult::AbnormalFlow, parse_info};
+                }
+                u32 value = *key;
+                u32 target = static_cast<u32>((value >> 3) + pc_target);
+                insert_label(state, target);
+                branches.emplace_back(value, target);
+            }
+            parse_info.end_address = offset;
+            parse_info.branch_info = MakeBranchInfo<MultiBranch>(
+                static_cast<u32>(instr.gpr8.Value()), std::move(branches));
+
+            return {ParseResult::ControlCaught, parse_info};
         }
         default:
             break;

src/video_core/shader/decode/arithmetic.cpp

@@ -43,12 +43,12 @@ u32 ShaderIR::DecodeArithmetic(NodeBlock& bb, u32 pc) {
     case OpCode::Id::FMUL_IMM: {
         // FMUL does not have 'abs' bits and only the second operand has a 'neg' bit.
         if (instr.fmul.tab5cb8_2 != 0) {
-            LOG_WARNING(HW_GPU, "FMUL tab5cb8_2({}) is not implemented",
-                        instr.fmul.tab5cb8_2.Value());
+            LOG_DEBUG(HW_GPU, "FMUL tab5cb8_2({}) is not implemented",
+                      instr.fmul.tab5cb8_2.Value());
         }
         if (instr.fmul.tab5c68_0 != 1) {
-            LOG_WARNING(HW_GPU, "FMUL tab5cb8_0({}) is not implemented",
-                        instr.fmul.tab5c68_0.Value());
+            LOG_DEBUG(HW_GPU, "FMUL tab5cb8_0({}) is not implemented",
+                      instr.fmul.tab5c68_0.Value());
         }
 
         op_b = GetOperandAbsNegFloat(op_b, false, instr.fmul.negate_b);
@@ -144,10 +144,11 @@ u32 ShaderIR::DecodeArithmetic(NodeBlock& bb, u32 pc) {
     case OpCode::Id::RRO_C:
     case OpCode::Id::RRO_R:
     case OpCode::Id::RRO_IMM: {
+        LOG_DEBUG(HW_GPU, "(STUBBED) RRO used");
+
         // Currently RRO is only implemented as a register move.
         op_b = GetOperandAbsNegFloat(op_b, instr.alu.abs_b, instr.alu.negate_b);
         SetRegister(bb, instr.gpr0, op_b);
-        LOG_WARNING(HW_GPU, "RRO instruction is incomplete");
         break;
     }
     default:

src/video_core/shader/decode/arithmetic_half.cpp

@@ -21,8 +21,8 @@ u32 ShaderIR::DecodeArithmeticHalf(NodeBlock& bb, u32 pc) {
 
     if (opcode->get().GetId() == OpCode::Id::HADD2_C ||
         opcode->get().GetId() == OpCode::Id::HADD2_R) {
-        if (instr.alu_half.ftz != 0) {
-            LOG_WARNING(HW_GPU, "{} FTZ not implemented", opcode->get().GetName());
+        if (instr.alu_half.ftz == 0) {
+            LOG_DEBUG(HW_GPU, "{} without FTZ is not implemented", opcode->get().GetName());
         }
     }
 

src/video_core/shader/decode/arithmetic_half_immediate.cpp

@@ -19,12 +19,12 @@ u32 ShaderIR::DecodeArithmeticHalfImmediate(NodeBlock& bb, u32 pc) {
     const auto opcode = OpCode::Decode(instr);
 
     if (opcode->get().GetId() == OpCode::Id::HADD2_IMM) {
-        if (instr.alu_half_imm.ftz != 0) {
-            LOG_WARNING(HW_GPU, "{} FTZ not implemented", opcode->get().GetName());
+        if (instr.alu_half_imm.ftz == 0) {
+            LOG_DEBUG(HW_GPU, "{} without FTZ is not implemented", opcode->get().GetName());
         }
     } else {
-        if (instr.alu_half_imm.precision != Tegra::Shader::HalfPrecision::None) {
-            LOG_WARNING(HW_GPU, "{} FTZ not implemented", opcode->get().GetName());
+        if (instr.alu_half_imm.precision != Tegra::Shader::HalfPrecision::FTZ) {
+            LOG_DEBUG(HW_GPU, "{} without FTZ is not implemented", opcode->get().GetName());
         }
     }
 

src/video_core/shader/decode/ffma.cpp

@@ -19,10 +19,10 @@ u32 ShaderIR::DecodeFfma(NodeBlock& bb, u32 pc) {
 
     UNIMPLEMENTED_IF_MSG(instr.ffma.cc != 0, "FFMA cc not implemented");
     if (instr.ffma.tab5980_0 != 1) {
-        LOG_WARNING(HW_GPU, "FFMA tab5980_0({}) not implemented", instr.ffma.tab5980_0.Value());
+        LOG_DEBUG(HW_GPU, "FFMA tab5980_0({}) not implemented", instr.ffma.tab5980_0.Value());
     }
     if (instr.ffma.tab5980_1 != 0) {
-        LOG_WARNING(HW_GPU, "FFMA tab5980_1({}) not implemented", instr.ffma.tab5980_1.Value());
+        LOG_DEBUG(HW_GPU, "FFMA tab5980_1({}) not implemented", instr.ffma.tab5980_1.Value());
     }
 
     const Node op_a = GetRegister(instr.gpr8);

src/video_core/shader/decode/half_set.cpp

@@ -20,8 +20,8 @@ u32 ShaderIR::DecodeHalfSet(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 
-    if (instr.hset2.ftz != 0) {
-        LOG_WARNING(HW_GPU, "{} FTZ not implemented", opcode->get().GetName());
+    if (instr.hset2.ftz == 0) {
+        LOG_DEBUG(HW_GPU, "{} without FTZ is not implemented", opcode->get().GetName());
     }
 
     Node op_a = UnpackHalfFloat(GetRegister(instr.gpr8), instr.hset2.type_a);

src/video_core/shader/decode/half_set_predicate.cpp

@@ -19,7 +19,9 @@ u32 ShaderIR::DecodeHalfSetPredicate(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
 
-    LOG_DEBUG(HW_GPU, "ftz={}", static_cast<u32>(instr.hsetp2.ftz));
+    if (instr.hsetp2.ftz != 0) {
+        LOG_DEBUG(HW_GPU, "{} without FTZ is not implemented", opcode->get().GetName());
+    }
 
     Node op_a = UnpackHalfFloat(GetRegister(instr.gpr8), instr.hsetp2.type_a);
     op_a = GetOperandAbsNegHalf(op_a, instr.hsetp2.abs_a, instr.hsetp2.negate_a);

src/video_core/shader/decode/image.cpp

@@ -143,39 +143,37 @@ u32 ShaderIR::DecodeImage(NodeBlock& bb, u32 pc) {
 }
 
 Image& ShaderIR::GetImage(Tegra::Shader::Image image, Tegra::Shader::ImageType type) {
-    const auto offset{static_cast<std::size_t>(image.index.Value())};
-    if (const auto existing_image = TryUseExistingImage(offset, type)) {
-        return *existing_image;
+    const auto offset = static_cast<u32>(image.index.Value());
+
+    const auto it =
+        std::find_if(std::begin(used_images), std::end(used_images),
+                     [offset](const Image& entry) { return entry.GetOffset() == offset; });
+    if (it != std::end(used_images)) {
+        ASSERT(!it->IsBindless() && it->GetType() == it->GetType());
+        return *it;
     }
 
-    const std::size_t next_index{used_images.size()};
-    return used_images.emplace(offset, Image{offset, next_index, type}).first->second;
+    const auto next_index = static_cast<u32>(used_images.size());
+    return used_images.emplace_back(next_index, offset, type);
 }
 
 Image& ShaderIR::GetBindlessImage(Tegra::Shader::Register reg, Tegra::Shader::ImageType type) {
-    const Node image_register{GetRegister(reg)};
-    const auto [base_image, cbuf_index, cbuf_offset]{
-        TrackCbuf(image_register, global_code, static_cast<s64>(global_code.size()))};
-    const auto cbuf_key{(static_cast<u64>(cbuf_index) << 32) | static_cast<u64>(cbuf_offset)};
+    const Node image_register = GetRegister(reg);
+    const auto [base_image, buffer, offset] =
+        TrackCbuf(image_register, global_code, static_cast<s64>(global_code.size()));
 
-    if (const auto image = TryUseExistingImage(cbuf_key, type)) {
-        return *image;
+    const auto it =
+        std::find_if(std::begin(used_images), std::end(used_images),
+                     [buffer = buffer, offset = offset](const Image& entry) {
+                         return entry.GetBuffer() == buffer && entry.GetOffset() == offset;
+                     });
+    if (it != std::end(used_images)) {
+        ASSERT(it->IsBindless() && it->GetType() == it->GetType());
+        return *it;
     }
 
-    const std::size_t next_index{used_images.size()};
-    return used_images.emplace(cbuf_key, Image{cbuf_index, cbuf_offset, next_index, type})
-        .first->second;
-}
-
-Image* ShaderIR::TryUseExistingImage(u64 offset, Tegra::Shader::ImageType type) {
-    auto it = used_images.find(offset);
-    if (it == used_images.end()) {
-        return nullptr;
-    }
-    auto& image = it->second;
-    ASSERT(image.GetType() == type);
-
-    return &image;
+    const auto next_index = static_cast<u32>(used_images.size());
+    return used_images.emplace_back(next_index, offset, buffer, type);
 }
 
 } // namespace VideoCommon::Shader

src/video_core/shader/decode/texture.cpp

@@ -44,10 +44,6 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
     bool is_bindless = false;
     switch (opcode->get().GetId()) {
     case OpCode::Id::TEX: {
-        if (instr.tex.UsesMiscMode(TextureMiscMode::NODEP)) {
-            LOG_WARNING(HW_GPU, "TEX.NODEP implementation is incomplete");
-        }
-
         const TextureType texture_type{instr.tex.texture_type};
         const bool is_array = instr.tex.array != 0;
         const bool is_aoffi = instr.tex.UsesMiscMode(TextureMiscMode::AOFFI);
@@ -62,10 +58,6 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
         UNIMPLEMENTED_IF_MSG(instr.tex.UsesMiscMode(TextureMiscMode::AOFFI),
                              "AOFFI is not implemented");
 
-        if (instr.tex.UsesMiscMode(TextureMiscMode::NODEP)) {
-            LOG_WARNING(HW_GPU, "TEX.NODEP implementation is incomplete");
-        }
-
         const TextureType texture_type{instr.tex_b.texture_type};
         const bool is_array = instr.tex_b.array != 0;
         const bool is_aoffi = instr.tex.UsesMiscMode(TextureMiscMode::AOFFI);
@@ -82,10 +74,6 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
         const bool depth_compare = instr.texs.UsesMiscMode(TextureMiscMode::DC);
         const auto process_mode = instr.texs.GetTextureProcessMode();
 
-        if (instr.texs.UsesMiscMode(TextureMiscMode::NODEP)) {
-            LOG_WARNING(HW_GPU, "TEXS.NODEP implementation is incomplete");
-        }
-
         const Node4 components =
             GetTexsCode(instr, texture_type, process_mode, depth_compare, is_array);
 
@@ -107,10 +95,6 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
         UNIMPLEMENTED_IF_MSG(instr.tld4.UsesMiscMode(TextureMiscMode::PTP),
                              "PTP is not implemented");
 
-        if (instr.tld4.UsesMiscMode(TextureMiscMode::NODEP)) {
-            LOG_WARNING(HW_GPU, "TLD4.NODEP implementation is incomplete");
-        }
-
         const auto texture_type = instr.tld4.texture_type.Value();
         const bool depth_compare = is_bindless ? instr.tld4_b.UsesMiscMode(TextureMiscMode::DC)
                                                : instr.tld4.UsesMiscMode(TextureMiscMode::DC);
@@ -125,9 +109,6 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
     case OpCode::Id::TLD4S: {
         UNIMPLEMENTED_IF_MSG(instr.tld4s.UsesMiscMode(TextureMiscMode::AOFFI),
                              "AOFFI is not implemented");
-        if (instr.tld4s.UsesMiscMode(TextureMiscMode::NODEP)) {
-            LOG_WARNING(HW_GPU, "TLD4S.NODEP implementation is incomplete");
-        }
 
         const bool depth_compare = instr.tld4s.UsesMiscMode(TextureMiscMode::DC);
         const Node op_a = GetRegister(instr.gpr8);
@@ -164,10 +145,6 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
         is_bindless = true;
         [[fallthrough]];
     case OpCode::Id::TXQ: {
-        if (instr.txq.UsesMiscMode(TextureMiscMode::NODEP)) {
-            LOG_WARNING(HW_GPU, "TXQ.NODEP implementation is incomplete");
-        }
-
         // TODO: The new commits on the texture refactor, change the way samplers work.
         // Sadly, not all texture instructions specify the type of texture their sampler
         // uses. This must be fixed at a later instance.
@@ -205,10 +182,6 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
         UNIMPLEMENTED_IF_MSG(instr.tmml.UsesMiscMode(Tegra::Shader::TextureMiscMode::NDV),
                              "NDV is not implemented");
 
-        if (instr.tmml.UsesMiscMode(TextureMiscMode::NODEP)) {
-            LOG_WARNING(HW_GPU, "TMML.NODEP implementation is incomplete");
-        }
-
         auto texture_type = instr.tmml.texture_type.Value();
         const bool is_array = instr.tmml.array != 0;
         const auto& sampler =
@@ -254,25 +227,17 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
         UNIMPLEMENTED_IF_MSG(instr.tld.ms, "MS is not implemented");
         UNIMPLEMENTED_IF_MSG(instr.tld.cl, "CL is not implemented");
 
-        if (instr.tld.nodep_flag) {
-            LOG_WARNING(HW_GPU, "TLD.NODEP implementation is incomplete");
-        }
-
         WriteTexInstructionFloat(bb, instr, GetTldCode(instr));
         break;
     }
     case OpCode::Id::TLDS: {
-        const Tegra::Shader::TextureType texture_type{instr.tlds.GetTextureType()};
+        const TextureType texture_type{instr.tlds.GetTextureType()};
         const bool is_array{instr.tlds.IsArrayTexture()};
 
         UNIMPLEMENTED_IF_MSG(instr.tlds.UsesMiscMode(TextureMiscMode::AOFFI),
                              "AOFFI is not implemented");
         UNIMPLEMENTED_IF_MSG(instr.tlds.UsesMiscMode(TextureMiscMode::MZ), "MZ is not implemented");
 
-        if (instr.tlds.UsesMiscMode(TextureMiscMode::NODEP)) {
-            LOG_WARNING(HW_GPU, "TLDS.NODEP implementation is incomplete");
-        }
-
         const Node4 components = GetTldsCode(instr, texture_type, is_array);
 
         if (instr.tlds.fp32_flag) {
@@ -293,77 +258,80 @@ const Sampler& ShaderIR::GetSampler(const Tegra::Shader::Sampler& sampler,
                                     std::optional<SamplerInfo> sampler_info) {
     const auto offset = static_cast<u32>(sampler.index.Value());
 
-    Tegra::Shader::TextureType type;
+    TextureType type;
     bool is_array;
     bool is_shadow;
     if (sampler_info) {
         type = sampler_info->type;
         is_array = sampler_info->is_array;
         is_shadow = sampler_info->is_shadow;
-    } else if (auto sampler = locker.ObtainBoundSampler(offset); sampler) {
+    } else if (const auto sampler = locker.ObtainBoundSampler(offset)) {
         type = sampler->texture_type.Value();
         is_array = sampler->is_array.Value() != 0;
         is_shadow = sampler->is_shadow.Value() != 0;
     } else {
-        type = Tegra::Shader::TextureType::Texture2D;
+        LOG_WARNING(HW_GPU, "Unknown sampler info");
+        type = TextureType::Texture2D;
         is_array = false;
         is_shadow = false;
     }
 
     // If this sampler has already been used, return the existing mapping.
-    const auto itr =
+    const auto it =
         std::find_if(used_samplers.begin(), used_samplers.end(),
-                     [&](const Sampler& entry) { return entry.GetOffset() == offset; });
-    if (itr != used_samplers.end()) {
-        ASSERT(itr->GetType() == type && itr->IsArray() == is_array &&
-               itr->IsShadow() == is_shadow);
-        return *itr;
+                     [offset](const Sampler& entry) { return entry.GetOffset() == offset; });
+    if (it != used_samplers.end()) {
+        ASSERT(!it->IsBindless() && it->GetType() == type && it->IsArray() == is_array &&
+               it->IsShadow() == is_shadow);
+        return *it;
     }
 
     // Otherwise create a new mapping for this sampler
-    const std::size_t next_index = used_samplers.size();
-    const Sampler entry{offset, next_index, type, is_array, is_shadow};
-    return *used_samplers.emplace(entry).first;
-} // namespace VideoCommon::Shader
+    const auto next_index = static_cast<u32>(used_samplers.size());
+    return used_samplers.emplace_back(Sampler(next_index, offset, type, is_array, is_shadow));
+}
 
 const Sampler& ShaderIR::GetBindlessSampler(const Tegra::Shader::Register& reg,
                                             std::optional<SamplerInfo> sampler_info) {
     const Node sampler_register = GetRegister(reg);
-    const auto [base_sampler, cbuf_index, cbuf_offset] =
+    const auto [base_sampler, buffer, offset] =
         TrackCbuf(sampler_register, global_code, static_cast<s64>(global_code.size()));
     ASSERT(base_sampler != nullptr);
-    const auto cbuf_key = (static_cast<u64>(cbuf_index) << 32) | static_cast<u64>(cbuf_offset);
-    Tegra::Shader::TextureType type;
+
+    TextureType type;
     bool is_array;
     bool is_shadow;
     if (sampler_info) {
         type = sampler_info->type;
         is_array = sampler_info->is_array;
         is_shadow = sampler_info->is_shadow;
-    } else if (auto sampler = locker.ObtainBindlessSampler(cbuf_index, cbuf_offset); sampler) {
+    } else if (const auto sampler = locker.ObtainBindlessSampler(buffer, offset)) {
         type = sampler->texture_type.Value();
         is_array = sampler->is_array.Value() != 0;
         is_shadow = sampler->is_shadow.Value() != 0;
     } else {
-        type = Tegra::Shader::TextureType::Texture2D;
+        LOG_WARNING(HW_GPU, "Unknown sampler info");
+        type = TextureType::Texture2D;
         is_array = false;
         is_shadow = false;
     }
 
     // If this sampler has already been used, return the existing mapping.
-    const auto itr =
+    const auto it =
         std::find_if(used_samplers.begin(), used_samplers.end(),
-                     [&](const Sampler& entry) { return entry.GetOffset() == cbuf_key; });
-    if (itr != used_samplers.end()) {
-        ASSERT(itr->GetType() == type && itr->IsArray() == is_array &&
-               itr->IsShadow() == is_shadow);
-        return *itr;
+                     [buffer = buffer, offset = offset](const Sampler& entry) {
+                         return entry.GetBuffer() == buffer && entry.GetOffset() == offset;
+                     });
+    if (it != used_samplers.end()) {
+        ASSERT(it->IsBindless() && it->GetType() == type && it->IsArray() == is_array &&
+               it->IsShadow() == is_shadow);
+        return *it;
     }
 
     // Otherwise create a new mapping for this sampler
-    const std::size_t next_index = used_samplers.size();
-    const Sampler entry{cbuf_index, cbuf_offset, next_index, type, is_array, is_shadow};
-    return *used_samplers.emplace(entry).first;
+    const auto next_index = static_cast<u32>(used_samplers.size());
+    return used_samplers.emplace_back(
+        Sampler(next_index, offset, buffer, type, is_array, is_shadow));
 }
 
 void ShaderIR::WriteTexInstructionFloat(NodeBlock& bb, Instruction instr, const Node4& components) {

src/video_core/shader/node.h

@@ -230,62 +230,49 @@ using NodeBlock = std::vector<Node>;
 class Sampler {
 public:
     /// This constructor is for bound samplers
-    explicit Sampler(std::size_t offset, std::size_t index, Tegra::Shader::TextureType type,
-                     bool is_array, bool is_shadow)
-        : offset{offset}, index{index}, type{type}, is_array{is_array}, is_shadow{is_shadow},
-          is_bindless{false} {}
+    constexpr explicit Sampler(u32 index, u32 offset, Tegra::Shader::TextureType type,
+                               bool is_array, bool is_shadow)
+        : index{index}, offset{offset}, type{type}, is_array{is_array}, is_shadow{is_shadow} {}
 
     /// This constructor is for bindless samplers
-    explicit Sampler(u32 cbuf_index, u32 cbuf_offset, std::size_t index,
-                     Tegra::Shader::TextureType type, bool is_array, bool is_shadow)
-        : offset{(static_cast<u64>(cbuf_index) << 32) | cbuf_offset}, index{index}, type{type},
-          is_array{is_array}, is_shadow{is_shadow}, is_bindless{true} {}
+    constexpr explicit Sampler(u32 index, u32 offset, u32 buffer, Tegra::Shader::TextureType type,
+                               bool is_array, bool is_shadow)
+        : index{index}, offset{offset}, buffer{buffer}, type{type}, is_array{is_array},
+          is_shadow{is_shadow}, is_bindless{true} {}
 
-    /// This constructor is for serialization/deserialization
-    explicit Sampler(std::size_t offset, std::size_t index, Tegra::Shader::TextureType type,
-                     bool is_array, bool is_shadow, bool is_bindless)
-        : offset{offset}, index{index}, type{type}, is_array{is_array}, is_shadow{is_shadow},
-          is_bindless{is_bindless} {}
-
-    std::size_t GetOffset() const {
-        return offset;
-    }
-
-    std::size_t GetIndex() const {
+    constexpr u32 GetIndex() const {
         return index;
     }
 
-    Tegra::Shader::TextureType GetType() const {
+    constexpr u32 GetOffset() const {
+        return offset;
+    }
+
+    constexpr u32 GetBuffer() const {
+        return buffer;
+    }
+
+    constexpr Tegra::Shader::TextureType GetType() const {
         return type;
     }
 
-    bool IsArray() const {
+    constexpr bool IsArray() const {
         return is_array;
     }
 
-    bool IsShadow() const {
+    constexpr bool IsShadow() const {
         return is_shadow;
     }
 
-    bool IsBindless() const {
+    constexpr bool IsBindless() const {
         return is_bindless;
     }
 
-    std::pair<u32, u32> GetBindlessCBuf() const {
-        return {static_cast<u32>(offset >> 32), static_cast<u32>(offset)};
-    }
-
-    bool operator<(const Sampler& rhs) const {
-        return std::tie(index, offset, type, is_array, is_shadow, is_bindless) <
-               std::tie(rhs.index, rhs.offset, rhs.type, rhs.is_array, rhs.is_shadow,
-                        rhs.is_bindless);
-    }
-
 private:
-    /// Offset in TSC memory from which to read the sampler object, as specified by the sampling
-    /// instruction.
-    std::size_t offset{};
-    std::size_t index{}; ///< Value used to index into the generated GLSL sampler array.
+    u32 index{};  ///< Emulated index given for the this sampler.
+    u32 offset{}; ///< Offset in the const buffer from where the sampler is being read.
+    u32 buffer{}; ///< Buffer where the bindless sampler is being read (unused on bound samplers).
+
     Tegra::Shader::TextureType type{}; ///< The type used to sample this texture (Texture2D, etc)
     bool is_array{};    ///< Whether the texture is being sampled as an array texture or not.
     bool is_shadow{};   ///< Whether the texture is being sampled as a depth texture or not.
@@ -294,18 +281,13 @@ private:
 
 class Image final {
 public:
-    constexpr explicit Image(std::size_t offset, std::size_t index, Tegra::Shader::ImageType type)
-        : offset{offset}, index{index}, type{type}, is_bindless{false} {}
+    /// This constructor is for bound images
+    constexpr explicit Image(u32 index, u32 offset, Tegra::Shader::ImageType type)
+        : index{index}, offset{offset}, type{type} {}
 
-    constexpr explicit Image(u32 cbuf_index, u32 cbuf_offset, std::size_t index,
-                             Tegra::Shader::ImageType type)
-        : offset{(static_cast<u64>(cbuf_index) << 32) | cbuf_offset}, index{index}, type{type},
-          is_bindless{true} {}
-
-    constexpr explicit Image(std::size_t offset, std::size_t index, Tegra::Shader::ImageType type,
-                             bool is_bindless, bool is_written, bool is_read, bool is_atomic)
-        : offset{offset}, index{index}, type{type}, is_bindless{is_bindless},
-          is_written{is_written}, is_read{is_read}, is_atomic{is_atomic} {}
+    /// This constructor is for bindless samplers
+    constexpr explicit Image(u32 index, u32 offset, u32 buffer, Tegra::Shader::ImageType type)
+        : index{index}, offset{offset}, buffer{buffer}, type{type}, is_bindless{true} {}
 
     void MarkWrite() {
         is_written = true;
@@ -321,12 +303,16 @@ public:
         is_atomic = true;
     }
 
-    constexpr std::size_t GetOffset() const {
+    constexpr u32 GetIndex() const {
+        return index;
+    }
+
+    constexpr u32 GetOffset() const {
         return offset;
     }
 
-    constexpr std::size_t GetIndex() const {
-        return index;
+    constexpr u32 GetBuffer() const {
+        return buffer;
     }
 
     constexpr Tegra::Shader::ImageType GetType() const {
@@ -349,18 +335,11 @@ public:
         return is_atomic;
     }
 
-    constexpr std::pair<u32, u32> GetBindlessCBuf() const {
-        return {static_cast<u32>(offset >> 32), static_cast<u32>(offset)};
-    }
-
-    constexpr bool operator<(const Image& rhs) const {
-        return std::tie(offset, index, type, is_bindless) <
-               std::tie(rhs.offset, rhs.index, rhs.type, rhs.is_bindless);
-    }
-
 private:
-    u64 offset{};
-    std::size_t index{};
+    u32 index{};
+    u32 offset{};
+    u32 buffer{};
+
     Tegra::Shader::ImageType type{};
     bool is_bindless{};
     bool is_written{};

src/video_core/shader/shader_ir.h

@@ -5,6 +5,7 @@
 #pragma once
 
 #include <array>
+#include <list>
 #include <map>
 #include <optional>
 #include <set>
@@ -95,11 +96,11 @@ public:
         return used_cbufs;
     }
 
-    const std::set<Sampler>& GetSamplers() const {
+    const std::list<Sampler>& GetSamplers() const {
         return used_samplers;
     }
 
-    const std::map<u64, Image>& GetImages() const {
+    const std::list<Image>& GetImages() const {
         return used_images;
     }
 
@@ -316,9 +317,6 @@ private:
     /// Access a bindless image sampler.
     Image& GetBindlessImage(Tegra::Shader::Register reg, Tegra::Shader::ImageType type);
 
-    /// Tries to access an existing image, updating it's state as needed
-    Image* TryUseExistingImage(u64 offset, Tegra::Shader::ImageType type);
-
     /// Extracts a sequence of bits from a node
     Node BitfieldExtract(Node value, u32 offset, u32 bits);
 
@@ -402,8 +400,8 @@ private:
     std::set<Tegra::Shader::Attribute::Index> used_input_attributes;
     std::set<Tegra::Shader::Attribute::Index> used_output_attributes;
     std::map<u32, ConstBuffer> used_cbufs;
-    std::set<Sampler> used_samplers;
-    std::map<u64, Image> used_images;
+    std::list<Sampler> used_samplers;
+    std::list<Image> used_images;
     std::array<bool, Tegra::Engines::Maxwell3D::Regs::NumClipDistances> used_clip_distances{};
     std::map<GlobalMemoryBase, GlobalMemoryUsage> used_global_memory;
     bool uses_layer{};

src/video_core/textures/texture.h

@@ -132,6 +132,8 @@ enum class SwizzleSource : u32 {
 };
 
 union TextureHandle {
+    TextureHandle(u32 raw) : raw{raw} {}
+
     u32 raw;
     BitField<0, 20, u32> tic_id;
     BitField<20, 12, u32> tsc_id;

src/web_service/web_backend.cpp

@@ -2,10 +2,12 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <array>
 #include <cstdlib>
 #include <mutex>
 #include <string>
 #include <LUrlParser.h>
+#include <fmt/format.h>
 #include <httplib.h>
 #include "common/common_types.h"
 #include "common/logging/log.h"
@@ -16,10 +18,10 @@ namespace WebService {
 
 constexpr std::array<const char, 1> API_VERSION{'1'};
 
-constexpr u32 HTTP_PORT = 80;
-constexpr u32 HTTPS_PORT = 443;
+constexpr int HTTP_PORT = 80;
+constexpr int HTTPS_PORT = 443;
 
-constexpr u32 TIMEOUT_SECONDS = 30;
+constexpr std::size_t TIMEOUT_SECONDS = 30;
 
 struct Client::Impl {
     Impl(std::string host, std::string username, std::string token)
@@ -31,8 +33,9 @@ struct Client::Impl {
     }
 
     /// A generic function handles POST, GET and DELETE request together
-    Common::WebResult GenericJson(const std::string& method, const std::string& path,
-                                  const std::string& data, bool allow_anonymous) {
+    Common::WebResult GenericRequest(const std::string& method, const std::string& path,
+                                     const std::string& data, bool allow_anonymous,
+                                     const std::string& accept) {
         if (jwt.empty()) {
             UpdateJWT();
         }
@@ -43,11 +46,11 @@ struct Client::Impl {
                                      "Credentials needed"};
         }
 
-        auto result = GenericJson(method, path, data, jwt);
+        auto result = GenericRequest(method, path, data, accept, jwt);
         if (result.result_string == "401") {
             // Try again with new JWT
             UpdateJWT();
-            result = GenericJson(method, path, data, jwt);
+            result = GenericRequest(method, path, data, accept, jwt);
         }
 
         return result;
@@ -56,12 +59,13 @@ struct Client::Impl {
     /**
      * A generic function with explicit authentication method specified
      * JWT is used if the jwt parameter is not empty
-     * username + token is used if jwt is empty but username and token are not empty
-     * anonymous if all of jwt, username and token are empty
+     * username + token is used if jwt is empty but username and token are
+     * not empty anonymous if all of jwt, username and token are empty
      */
-    Common::WebResult GenericJson(const std::string& method, const std::string& path,
-                                  const std::string& data, const std::string& jwt = "",
-                                  const std::string& username = "", const std::string& token = "") {
+    Common::WebResult GenericRequest(const std::string& method, const std::string& path,
+                                     const std::string& data, const std::string& accept,
+                                     const std::string& jwt = "", const std::string& username = "",
+                                     const std::string& token = "") {
         if (cli == nullptr) {
             auto parsedUrl = LUrlParser::clParseURL::ParseURL(host);
             int port;
@@ -132,8 +136,7 @@ struct Client::Impl {
             return Common::WebResult{Common::WebResult::Code::WrongContent, ""};
         }
 
-        if (content_type->second.find("application/json") == std::string::npos &&
-            content_type->second.find("text/html; charset=utf-8") == std::string::npos) {
+        if (content_type->second.find(accept) == std::string::npos) {
             LOG_ERROR(WebService, "{} to {} returned wrong content: {}", method, host + path,
                       content_type->second);
             return Common::WebResult{Common::WebResult::Code::WrongContent, "Wrong content"};
@@ -147,7 +150,7 @@ struct Client::Impl {
             return;
         }
 
-        auto result = GenericJson("POST", "/jwt/internal", "", "", username, token);
+        auto result = GenericRequest("POST", "/jwt/internal", "", "text/html", "", username, token);
         if (result.result_code != Common::WebResult::Code::Success) {
             LOG_ERROR(WebService, "UpdateJWT failed");
         } else {
@@ -180,16 +183,29 @@ Client::~Client() = default;
 
 Common::WebResult Client::PostJson(const std::string& path, const std::string& data,
                                    bool allow_anonymous) {
-    return impl->GenericJson("POST", path, data, allow_anonymous);
+    return impl->GenericRequest("POST", path, data, allow_anonymous, "application/json");
 }
 
 Common::WebResult Client::GetJson(const std::string& path, bool allow_anonymous) {
-    return impl->GenericJson("GET", path, "", allow_anonymous);
+    return impl->GenericRequest("GET", path, "", allow_anonymous, "application/json");
 }
 
 Common::WebResult Client::DeleteJson(const std::string& path, const std::string& data,
                                      bool allow_anonymous) {
-    return impl->GenericJson("DELETE", path, data, allow_anonymous);
+    return impl->GenericRequest("DELETE", path, data, allow_anonymous, "application/json");
+}
+
+Common::WebResult Client::GetPlain(const std::string& path, bool allow_anonymous) {
+    return impl->GenericRequest("GET", path, "", allow_anonymous, "text/plain");
+}
+
+Common::WebResult Client::GetImage(const std::string& path, bool allow_anonymous) {
+    return impl->GenericRequest("GET", path, "", allow_anonymous, "image/png");
+}
+
+Common::WebResult Client::GetExternalJWT(const std::string& audience) {
+    return impl->GenericRequest("POST", fmt::format("/jwt/external/{}", audience), "", false,
+                                "text/html");
 }
 
 } // namespace WebService

src/web_service/web_backend.h

@@ -46,6 +46,29 @@ public:
     Common::WebResult DeleteJson(const std::string& path, const std::string& data,
                                  bool allow_anonymous);
 
+    /**
+     * Gets a plain string from the specified path.
+     * @param path the URL segment after the host address.
+     * @param allow_anonymous If true, allow anonymous unauthenticated requests.
+     * @return the result of the request.
+     */
+    Common::WebResult GetPlain(const std::string& path, bool allow_anonymous);
+
+    /**
+     * Gets an PNG image from the specified path.
+     * @param path the URL segment after the host address.
+     * @param allow_anonymous If true, allow anonymous unauthenticated requests.
+     * @return the result of the request.
+     */
+    Common::WebResult GetImage(const std::string& path, bool allow_anonymous);
+
+    /**
+     * Requests an external JWT for the specific audience provided.
+     * @param audience the audience of the JWT requested.
+     * @return the result of the request.
+     */
+    Common::WebResult GetExternalJWT(const std::string& audience);
+
 private:
     struct Impl;
     std::unique_ptr<Impl> impl;

src/yuzu/configuration/configure_web.cpp

@@ -11,6 +11,31 @@
 #include "yuzu/configuration/configure_web.h"
 #include "yuzu/uisettings.h"
 
+static constexpr char token_delimiter{':'};
+
+static std::string GenerateDisplayToken(const std::string& username, const std::string& token) {
+    if (username.empty() || token.empty()) {
+        return {};
+    }
+
+    const std::string unencoded_display_token{username + token_delimiter + token};
+    QByteArray b{unencoded_display_token.c_str()};
+    QByteArray b64 = b.toBase64();
+    return b64.toStdString();
+}
+
+static std::string UsernameFromDisplayToken(const std::string& display_token) {
+    const std::string unencoded_display_token{
+        QByteArray::fromBase64(display_token.c_str()).toStdString()};
+    return unencoded_display_token.substr(0, unencoded_display_token.find(token_delimiter));
+}
+
+static std::string TokenFromDisplayToken(const std::string& display_token) {
+    const std::string unencoded_display_token{
+        QByteArray::fromBase64(display_token.c_str()).toStdString()};
+    return unencoded_display_token.substr(unencoded_display_token.find(token_delimiter) + 1);
+}
+
 ConfigureWeb::ConfigureWeb(QWidget* parent)
     : QWidget(parent), ui(std::make_unique<Ui::ConfigureWeb>()) {
     ui->setupUi(this);
@@ -63,13 +88,18 @@ void ConfigureWeb::SetConfiguration() {
     ui->web_signup_link->setOpenExternalLinks(true);
     ui->web_token_info_link->setOpenExternalLinks(true);
 
+    if (Settings::values.yuzu_username.empty()) {
+        ui->username->setText(tr("Unspecified"));
+    } else {
+        ui->username->setText(QString::fromStdString(Settings::values.yuzu_username));
+    }
+
     ui->toggle_telemetry->setChecked(Settings::values.enable_telemetry);
-    ui->edit_username->setText(QString::fromStdString(Settings::values.yuzu_username));
-    ui->edit_token->setText(QString::fromStdString(Settings::values.yuzu_token));
+    ui->edit_token->setText(QString::fromStdString(
+        GenerateDisplayToken(Settings::values.yuzu_username, Settings::values.yuzu_token)));
 
     // Connect after setting the values, to avoid calling OnLoginChanged now
     connect(ui->edit_token, &QLineEdit::textChanged, this, &ConfigureWeb::OnLoginChanged);
-    connect(ui->edit_username, &QLineEdit::textChanged, this, &ConfigureWeb::OnLoginChanged);
 
     user_verified = true;
 
@@ -80,12 +110,13 @@ void ConfigureWeb::ApplyConfiguration() {
     Settings::values.enable_telemetry = ui->toggle_telemetry->isChecked();
     UISettings::values.enable_discord_presence = ui->toggle_discordrpc->isChecked();
     if (user_verified) {
-        Settings::values.yuzu_username = ui->edit_username->text().toStdString();
-        Settings::values.yuzu_token = ui->edit_token->text().toStdString();
+        Settings::values.yuzu_username =
+            UsernameFromDisplayToken(ui->edit_token->text().toStdString());
+        Settings::values.yuzu_token = TokenFromDisplayToken(ui->edit_token->text().toStdString());
     } else {
-        QMessageBox::warning(this, tr("Username and token not verified"),
-                             tr("Username and token were not verified. The changes to your "
-                                "username and/or token have not been saved."));
+        QMessageBox::warning(
+            this, tr("Token not verified"),
+            tr("Token was not verified. The change to your token has not been saved."));
     }
 }
 
@@ -96,17 +127,15 @@ void ConfigureWeb::RefreshTelemetryID() {
 }
 
 void ConfigureWeb::OnLoginChanged() {
-    if (ui->edit_username->text().isEmpty() && ui->edit_token->text().isEmpty()) {
+    if (ui->edit_token->text().isEmpty()) {
         user_verified = true;
 
         const QPixmap pixmap = QIcon::fromTheme(QStringLiteral("checked")).pixmap(16);
-        ui->label_username_verified->setPixmap(pixmap);
         ui->label_token_verified->setPixmap(pixmap);
     } else {
         user_verified = false;
 
         const QPixmap pixmap = QIcon::fromTheme(QStringLiteral("failed")).pixmap(16);
-        ui->label_username_verified->setPixmap(pixmap);
         ui->label_token_verified->setPixmap(pixmap);
     }
 }
@@ -114,10 +143,11 @@ void ConfigureWeb::OnLoginChanged() {
 void ConfigureWeb::VerifyLogin() {
     ui->button_verify_login->setDisabled(true);
     ui->button_verify_login->setText(tr("Verifying..."));
-    verify_watcher.setFuture(QtConcurrent::run([username = ui->edit_username->text().toStdString(),
-                                                token = ui->edit_token->text().toStdString()] {
-        return Core::VerifyLogin(username, token);
-    }));
+    verify_watcher.setFuture(QtConcurrent::run(
+        [username = UsernameFromDisplayToken(ui->edit_token->text().toStdString()),
+         token = TokenFromDisplayToken(ui->edit_token->text().toStdString())] {
+            return Core::VerifyLogin(username, token);
+        }));
 }
 
 void ConfigureWeb::OnLoginVerified() {
@@ -127,16 +157,15 @@ void ConfigureWeb::OnLoginVerified() {
         user_verified = true;
 
         const QPixmap pixmap = QIcon::fromTheme(QStringLiteral("checked")).pixmap(16);
-        ui->label_username_verified->setPixmap(pixmap);
         ui->label_token_verified->setPixmap(pixmap);
+        ui->username->setText(
+            QString::fromStdString(UsernameFromDisplayToken(ui->edit_token->text().toStdString())));
     } else {
         const QPixmap pixmap = QIcon::fromTheme(QStringLiteral("failed")).pixmap(16);
-        ui->label_username_verified->setPixmap(pixmap);
         ui->label_token_verified->setPixmap(pixmap);
-
-        QMessageBox::critical(
-            this, tr("Verification failed"),
-            tr("Verification failed. Check that you have entered your username and token "
-               "correctly, and that your internet connection is working."));
+        ui->username->setText(tr("Unspecified"));
+        QMessageBox::critical(this, tr("Verification failed"),
+                              tr("Verification failed. Check that you have entered your token "
+                                 "correctly, and that your internet connection is working."));
     }
 }

src/yuzu/configuration/configure_web.ui

@@ -55,11 +55,7 @@
            </widget>
           </item>
           <item row="0" column="1" colspan="3">
-           <widget class="QLineEdit" name="edit_username">
-            <property name="maxLength">
-             <number>36</number>
-            </property>
-           </widget>
+           <widget class="QLabel" name="username" />
           </item>
           <item row="1" column="0">
            <widget class="QLabel" name="label_token">
@@ -79,14 +75,10 @@
             </property>
            </widget>
           </item>
-          <item row="0" column="4">
-           <widget class="QLabel" name="label_username_verified">
-           </widget>
-          </item>
           <item row="1" column="1" colspan="3">
            <widget class="QLineEdit" name="edit_token">
             <property name="maxLength">
-             <number>36</number>
+             <number>80</number>
             </property>
             <property name="echoMode">
              <enum>QLineEdit::Password</enum>

src/yuzu/debugger/wait_tree.cpp

@@ -172,17 +172,6 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeWaitObject::GetChildren() con
     return list;
 }
 
-QString WaitTreeWaitObject::GetResetTypeQString(Kernel::ResetType reset_type) {
-    switch (reset_type) {
-    case Kernel::ResetType::Automatic:
-        return tr("automatic reset");
-    case Kernel::ResetType::Manual:
-        return tr("manual reset");
-    }
-    UNREACHABLE();
-    return {};
-}
-
 WaitTreeObjectList::WaitTreeObjectList(
     const std::vector<Kernel::SharedPtr<Kernel::WaitObject>>& list, bool w_all)
     : object_list(list), wait_all(w_all) {}
@@ -336,16 +325,6 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeThread::GetChildren() const {
 WaitTreeEvent::WaitTreeEvent(const Kernel::ReadableEvent& object) : WaitTreeWaitObject(object) {}
 WaitTreeEvent::~WaitTreeEvent() = default;
 
-std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeEvent::GetChildren() const {
-    std::vector<std::unique_ptr<WaitTreeItem>> list(WaitTreeWaitObject::GetChildren());
-
-    list.push_back(std::make_unique<WaitTreeText>(
-        tr("reset type = %1")
-            .arg(GetResetTypeQString(
-                static_cast<const Kernel::ReadableEvent&>(object).GetResetType()))));
-    return list;
-}
-
 WaitTreeThreadList::WaitTreeThreadList(const std::vector<Kernel::SharedPtr<Kernel::Thread>>& list)
     : thread_list(list) {}
 WaitTreeThreadList::~WaitTreeThreadList() = default;

src/yuzu/debugger/wait_tree.h

@@ -111,8 +111,6 @@ public:
 
 protected:
     const Kernel::WaitObject& object;
-
-    static QString GetResetTypeQString(Kernel::ResetType reset_type);
 };
 
 class WaitTreeObjectList : public WaitTreeExpandableItem {
@@ -146,8 +144,6 @@ class WaitTreeEvent : public WaitTreeWaitObject {
 public:
     explicit WaitTreeEvent(const Kernel::ReadableEvent& object);
     ~WaitTreeEvent() override;
-
-    std::vector<std::unique_ptr<WaitTreeItem>> GetChildren() const override;
 };
 
 class WaitTreeThreadList : public WaitTreeExpandableItem {