Shader OpCapability fallback + fp16 ban adjustments

This minor change in src\video_core\renderer_vulkan\pipeline_helper.h allows renderdoc capture on mhr sunbreak.
Maybe it sanitizes some crashes on old vulkan GPUs.

Device log (nvidia kepler gpu vulkan 1.1.117):
[  17.605262] Render.Vulkan <Info> video_core\vulkan_common\vulkan_device.cpp:1041:GetSuitability: Device doesn't support feature nullDescriptor

VUID-VkWriteDescriptorSet-descriptorType-02997
For sampled/combined/storage image descriptors, Khronos says imageView must be valid or VK_NULL_HANDLE
(https://docs.vulkan.org/refpages/latest/refpages/source/VkPhysicalDeviceRobustness2FeaturesKHR.html)

but then it says: if nullDescriptor is not enabled, imageView must not be VK_NULL_HANDLE.
(https://docs.vulkan.org/spec/latest/chapters/descriptorsets.html)
nullDescriptor is exactly the feature that allows descriptors to be written with null handles.

The fix relies on conditionally replacing the null sampled view with Eden’s dummy null-image-view object (ImageView constructor checks HasNullDescriptor which checks features.robustness2.nullDescriptor. cheap checks, and only for null handles, so no notable cost).

Reviewed-on: https://git.eden-emu.dev/eden-emu/eden/pulls/4056
Reviewed-by: MaranBr <maranbr@eden-emu.dev>
Reviewed-by: CamilleLaVey <camillelavey99@gmail.com>

src/android/app/src/main/java/org/yuzu/yuzu_emu/features/settings/model/BooleanSetting.kt

@@ -31,6 +31,7 @@ enum class BooleanSetting(override val key: String) : AbstractBooleanSetting {
     RENDERER_REACTIVE_FLUSHING("use_reactive_flushing"),
     ENABLE_BUFFER_HISTORY("enable_buffer_history"),
     USE_OPTIMIZED_VERTEX_BUFFERS("use_optimized_vertex_buffers"),
+    ENABLE_GPU_BUFFER_READBACK("enable_gpu_buffer_readback"),
     SYNC_MEMORY_OPERATIONS("sync_memory_operations"),
     BUFFER_REORDER_DISABLE("disable_buffer_reorder"),
     RENDERER_DEBUG("debug"),

src/android/app/src/main/java/org/yuzu/yuzu_emu/features/settings/model/view/SettingsItem.kt

@@ -806,6 +806,13 @@ abstract class SettingsItem(
                     descriptionId = R.string.enable_buffer_history_description
                 )
             )
+            put(
+                SwitchSetting(
+                    BooleanSetting.ENABLE_GPU_BUFFER_READBACK,
+                    titleId = R.string.enable_gpu_buffer_readback,
+                    descriptionId = R.string.enable_gpu_buffer_readback_description
+                )
+            )
             put(
                 SwitchSetting(
                     BooleanSetting.USE_OPTIMIZED_VERTEX_BUFFERS,

src/android/app/src/main/java/org/yuzu/yuzu_emu/features/settings/ui/SettingsFragmentPresenter.kt

@@ -292,6 +292,7 @@ class SettingsFragmentPresenter(
             add(BooleanSetting.RENDERER_FORCE_MAX_CLOCK.key)
             add(BooleanSetting.RENDERER_REACTIVE_FLUSHING.key)
             add(BooleanSetting.ENABLE_BUFFER_HISTORY.key)
+            add(BooleanSetting.ENABLE_GPU_BUFFER_READBACK.key)
             add(BooleanSetting.USE_OPTIMIZED_VERTEX_BUFFERS.key)
 
             add(HeaderSetting(R.string.hacks))

src/android/app/src/main/res/values/strings.xml

@@ -503,6 +503,8 @@
     <string name="renderer_reactive_flushing_description">Improves rendering accuracy in some games at the cost of performance.</string>
     <string name="enable_buffer_history">Enable buffer history</string>
     <string name="enable_buffer_history_description">Enables access to previous buffer states. This option may improve rendering quality and performance consistency in some games.</string>
+    <string name="enable_gpu_buffer_readback">Enable GPU Buffer Readback</string>
+    <string name="enable_gpu_buffer_readback_description">Preserves GPU-modified buffer data by reading it back before uploads. Some games require this to render certain effects properly. May cause issues if the hardware cannot handle the additional workload.</string>
     <string name="use_optimized_vertex_buffers">Optimized Vertex Buffers</string>
     <string name="use_optimized_vertex_buffers_description">Enables optimized vertex buffer binding for improved performance. Requires Mesa 26.0+ Turnip drivers/ QCOM drivers. Will crash on older Turnip drivers (25.3 and below).</string>
 

src/core/core.cpp

@@ -350,7 +350,7 @@ struct System::Impl {
 
         // Register with applet manager
         // All threads are started, begin main process execution, now that we're in the clear
-        applet_manager.CreateAndInsertByFrontendAppletParameters(std::make_unique<Service::Process>(*std::move(process)), params);
+        applet_manager.CreateAndInsertByFrontendAppletParameters(std::move(process), params);
 
         if (Settings::values.gamecard_inserted) {
             if (Settings::values.gamecard_current_game) {

src/core/hle/service/acc/acc.cpp

@@ -729,7 +729,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface(ensure_token_id);
+        rb.PushIpcInterface(ctx, ensure_token_id);
     }
 
     void LoadIdTokenCacheDeprecated(HLERequestContext& ctx) {
@@ -921,7 +921,7 @@ void Module::Interface::GetProfile(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IProfile>(system, user_id, *profile_manager);
+    rb.PushIpcInterface<IProfile>(ctx, system, user_id, *profile_manager);
 }
 
 void Module::Interface::IsUserRegistrationRequestPermitted(HLERequestContext& ctx) {
@@ -993,7 +993,7 @@ void Module::Interface::GetBaasAccountManagerForApplication(HLERequestContext& c
     LOG_DEBUG(Service_ACC, "called");
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IManagerForApplication>(system, profile_manager);
+    rb.PushIpcInterface<IManagerForApplication>(ctx, system, profile_manager);
 }
 
 void Module::Interface::IsUserAccountSwitchLocked(HLERequestContext& ctx) {
@@ -1089,7 +1089,7 @@ void Module::Interface::GetProfileEditor(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IProfileEditor>(system, user_id, *profile_manager);
+    rb.PushIpcInterface<IProfileEditor>(ctx, system, user_id, *profile_manager);
 }
 
 void Module::Interface::GetBaasAccountAdministrator(HLERequestContext &ctx) {
@@ -1100,7 +1100,7 @@ void Module::Interface::GetBaasAccountAdministrator(HLERequestContext &ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IAdministrator>(system, uuid);
+    rb.PushIpcInterface<IAdministrator>(ctx, system, uuid);
 }
 
 void Module::Interface::ListQualifiedUsers(HLERequestContext& ctx) {
@@ -1143,7 +1143,7 @@ void Module::Interface::GetBaasAccountManagerForSystemService(HLERequestContext&
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IManagerForSystemService>(system, uuid);
+    rb.PushIpcInterface<IManagerForSystemService>(ctx, system, uuid);
 }
 
 void Module::Interface::StoreSaveDataThumbnailSystem(HLERequestContext& ctx) {

src/core/hle/service/acc/async_context.cpp

@@ -35,7 +35,7 @@ void IAsyncContext::GetSystemEvent(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(ResultSuccess);
-    rb.PushCopyObjects(completion_event->GetReadableEvent());
+    rb.PushCopyObjects(ctx, completion_event->GetReadableEvent());
 }
 
 void IAsyncContext::Cancel(HLERequestContext& ctx) {

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

@@ -109,6 +109,10 @@ struct Applet {
     std::list<std::shared_ptr<Applet>> child_applets{};
     bool is_completed{};
 
+    std::shared_ptr<Applet> reserved_applet{};
+    bool unwind_after_reserved{};
+    bool is_winding{};
+
     // Self state
     bool exit_locked{};
     s32 fatal_section_count{};

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

@@ -25,6 +25,13 @@ void AppletStorageChannel::Push(Kernel::KernelCore& kernel, std::shared_ptr<ISto
     m_event.Signal(kernel);
 }
 
+void AppletStorageChannel::Unpop(Kernel::KernelCore& kernel, std::shared_ptr<IStorage> storage) {
+    std::scoped_lock lk{m_lock};
+
+    m_data.emplace_front(std::move(storage));
+    m_event.Signal(kernel);
+}
+
 Result AppletStorageChannel::Pop(Kernel::KernelCore& kernel, std::shared_ptr<IStorage>* out_storage) {
     std::scoped_lock lk{m_lock};
 

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

@@ -26,6 +26,7 @@ public:
     ~AppletStorageChannel();
 
     void Push(Kernel::KernelCore& kernel, std::shared_ptr<IStorage> storage);
+    void Unpop(Kernel::KernelCore& kernel, std::shared_ptr<IStorage> storage);
     Result Pop(Kernel::KernelCore& kernel, std::shared_ptr<IStorage>* out_storage);
     Kernel::KReadableEvent* GetEvent();
 

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

@@ -267,7 +267,7 @@ void AppletManager::SetWindowSystem(WindowSystem* window_system) {
 
     if (Settings::values.enable_overlay && m_window_system->GetOverlayDisplayApplet() == nullptr) {
         if (auto overlay_process = CreateProcess(m_system, static_cast<u64>(AppletProgramId::OverlayDisplay), 0, 0)) {
-            auto overlay_applet = std::make_shared<Applet>(m_system, std::make_unique<Service::Process>(*std::move(overlay_process)), false);
+            auto overlay_applet = std::make_shared<Applet>(m_system, std::move(overlay_process), false);
             overlay_applet->program_id = static_cast<u64>(AppletProgramId::OverlayDisplay);
             overlay_applet->applet_id = AppletId::OverlayDisplay;
             overlay_applet->type = AppletType::OverlayApplet;

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

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -12,7 +15,10 @@ namespace Service::AM {
 
 HidRegistration::HidRegistration(Core::System& system, Process& process) : m_process(process) {
     m_hid_server = system.ServiceManager().GetService<HID::IHidServer>("hid", true);
+    this->RegisterCurrentProcess();
+}
 
+void HidRegistration::RegisterCurrentProcess() {
     if (m_process.IsInitialized()) {
         m_hid_server->GetResourceManager()->RegisterAppletResourceUserId(m_process.GetProcessId(),
                                                                          true);

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

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -24,6 +27,7 @@ public:
     explicit HidRegistration(Core::System& system, Process& process);
     ~HidRegistration();
 
+    void RegisterCurrentProcess();
     void EnableAppletToGetInput(bool enable);
 
 private:

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

@@ -171,6 +171,18 @@ void LifecycleManager::SignalSystemEventIfNeeded(Kernel::KernelCore& kernel) {
     }
 }
 
+void LifecycleManager::ResetForRelaunch() {
+    m_unordered_messages.clear();
+
+    m_activity_state = ActivityState::BackgroundVisible;
+    m_requested_focus_state = FocusState{};
+    m_acknowledged_focus_state = FocusState{};
+    m_has_focus_state_changed = true;
+
+    m_suspend_mode = SuspendMode::NoOverride;
+    m_forced_suspend = false;
+}
+
 bool LifecycleManager::PopMessage(Kernel::KernelCore& kernel, AppletMessage* out_message) {
     const auto message = this->PopMessageInOrderOfPriority();
     this->SignalSystemEventIfNeeded(kernel);

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

@@ -138,6 +138,8 @@ public:
     void PushUnorderedMessage(Kernel::KernelCore& kernel, AppletMessage message);
     bool PopMessage(Kernel::KernelCore& kernel, AppletMessage* out_message);
 
+    void ResetForRelaunch();
+
 private:
     FocusState GetFocusStateWhileForegroundObscured() const;
     FocusState GetFocusStateWhileBackground(bool is_obscured) const;

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

@@ -40,23 +40,23 @@ namespace {
     }
 }
 
-[[nodiscard]] inline std::optional<Process> CreateProcessImpl(std::unique_ptr<Loader::AppLoader>& out_loader, Loader::ResultStatus& out_load_result, Core::System& system, FileSys::VirtualFile file, u64 program_id, u64 program_index) {
+[[nodiscard]] inline std::unique_ptr<Process> CreateProcessImpl(std::unique_ptr<Loader::AppLoader>& out_loader, Loader::ResultStatus& out_load_result, Core::System& system, FileSys::VirtualFile file, u64 program_id, u64 program_index) {
     // Get the appropriate loader to parse this NCA.
     out_loader = Loader::GetLoader(system, file, program_id, program_index);
     // Ensure we have a loader which can parse the NCA.
     if (out_loader) {
         // Try to load the process.
-        auto process = std::make_optional<Process>(system);
+        auto process = std::make_unique<Process>(system);
         if (process->Initialize(*out_loader, out_load_result)) {
             return process;
         }
     }
-    return std::nullopt;
+    return nullptr;
 }
 
 } // Anonymous namespace
 
-std::optional<Process> CreateProcess(Core::System& system, u64 program_id, u8 minimum_key_generation, u8 maximum_key_generation) {
+std::unique_ptr<Process> CreateProcess(Core::System& system, u64 program_id, u8 minimum_key_generation, u8 maximum_key_generation) {
     // Attempt to load program NCA.
     FileSys::VirtualFile nca_raw{};
 
@@ -66,7 +66,7 @@ std::optional<Process> CreateProcess(Core::System& system, u64 program_id, u8 mi
 
     // Ensure we retrieved a program NCA.
     if (!nca_raw) {
-        return std::nullopt;
+        return nullptr;
     }
 
     // Ensure we have a suitable version.
@@ -76,7 +76,7 @@ std::optional<Process> CreateProcess(Core::System& system, u64 program_id, u8 mi
             (nca.GetKeyGeneration() < minimum_key_generation ||
              nca.GetKeyGeneration() > maximum_key_generation)) {
             LOG_WARNING(Service_LDR, "Skipping program {:016X} with generation {}", program_id, nca.GetKeyGeneration());
-            return std::nullopt;
+            return nullptr;
         }
     }
 
@@ -85,7 +85,7 @@ std::optional<Process> CreateProcess(Core::System& system, u64 program_id, u8 mi
     return CreateProcessImpl(loader, status, system, nca_raw, program_id, 0);
 }
 
-std::optional<Process> CreateApplicationProcess(std::vector<u8>& out_control, std::unique_ptr<Loader::AppLoader>& out_loader, Loader::ResultStatus& out_load_result, Core::System& system, FileSys::VirtualFile file, u64 program_id, u64 program_index) {
+std::unique_ptr<Process> CreateApplicationProcess(std::vector<u8>& out_control, std::unique_ptr<Loader::AppLoader>& out_loader, Loader::ResultStatus& out_load_result, Core::System& system, FileSys::VirtualFile file, u64 program_id, u64 program_index) {
     if (auto process = CreateProcessImpl(out_loader, out_load_result, system, file, program_id, program_index); process) {
         FileSys::NACP nacp;
         if (out_loader->ReadControlData(nacp) == Loader::ResultStatus::Success) {
@@ -110,7 +110,23 @@ std::optional<Process> CreateApplicationProcess(std::vector<u8>& out_control, st
         system.GetARPManager().Register(launch.title_id, launch, out_control);
         return process;
     }
-    return std::nullopt;
+    return nullptr;
+}
+
+bool ReinitializeProcess(Core::System& system, Process& process, u64 program_id) {
+    auto& storage = system.GetContentProviderUnion();
+    const auto nca_raw = storage.GetEntryRaw(program_id, FileSys::ContentRecordType::Program);
+    if (!nca_raw) {
+        return false;
+    }
+
+    auto loader = Loader::GetLoader(system, nca_raw, program_id, 0);
+    if (!loader) {
+        return false;
+    }
+
+    Loader::ResultStatus status{};
+    return process.Initialize(*loader, status);
 }
 
 } // namespace Service::AM

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

@@ -27,7 +27,9 @@ class Process;
 
 namespace Service::AM {
 
-std::optional<Process> CreateProcess(Core::System& system, u64 program_id, u8 minimum_key_generation, u8 maximum_key_generation);
-std::optional<Process> CreateApplicationProcess(std::vector<u8>& out_control, std::unique_ptr<Loader::AppLoader>& out_loader, Loader::ResultStatus& out_load_result, Core::System& system, FileSys::VirtualFile file, u64 program_id, u64 program_index);
+std::unique_ptr<Process> CreateProcess(Core::System& system, u64 program_id, u8 minimum_key_generation, u8 maximum_key_generation);
+std::unique_ptr<Process> CreateApplicationProcess(std::vector<u8>& out_control, std::unique_ptr<Loader::AppLoader>& out_loader, Loader::ResultStatus& out_load_result, Core::System& system, FileSys::VirtualFile file, u64 program_id, u64 program_index);
+
+bool ReinitializeProcess(Core::System& system, Process& process, u64 program_id);
 
 } // namespace Service::AM

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

@@ -35,9 +35,9 @@ Result CreateGuestApplication(SharedPointer<IApplicationAccessor>* out_applicati
     std::unique_ptr<Loader::AppLoader> loader;
     Loader::ResultStatus result;
     auto process = CreateApplicationProcess(control, loader, result, system, nca_raw, program_id, 0);
-    R_UNLESS(process != std::nullopt, ResultUnknown);
+    R_UNLESS(process != nullptr, ResultUnknown);
 
-    const auto applet = std::make_shared<Applet>(system, std::make_unique<Service::Process>(*std::move(process)), true);
+    const auto applet = std::make_shared<Applet>(system, std::move(process), true);
     applet->program_id = program_id;
     applet->applet_id = AppletId::Application;
     applet->type = AppletType::Application;
@@ -88,9 +88,9 @@ Result IApplicationCreator::CreateSystemApplication(
     std::vector<u8> control;
     std::unique_ptr<Loader::AppLoader> loader;
     auto process = CreateProcess(system, application_id, 1, 22);
-    R_UNLESS(process != std::nullopt, ResultUnknown);
+    R_UNLESS(process != nullptr, ResultUnknown);
 
-    const auto applet = std::make_shared<Applet>(system, std::make_unique<Service::Process>(*std::move(process)), true);
+    const auto applet = std::make_shared<Applet>(system, std::move(process), true);
     applet->program_id = application_id;
     applet->applet_id = AppletId::Starter;
     applet->type = AppletType::LibraryApplet;

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

@@ -75,7 +75,7 @@ ILibraryAppletAccessor::ILibraryAppletAccessor(Core::System& system_,
         {105, D<&ILibraryAppletAccessor::GetPopOutDataEvent>, "GetPopOutDataEvent"},
         {106, D<&ILibraryAppletAccessor::GetPopInteractiveOutDataEvent>, "GetPopInteractiveOutDataEvent"},
         {110, nullptr, "NeedsToExitProcess"},
-        {120, nullptr, "GetLibraryAppletInfo"},
+        {120, D<&ILibraryAppletAccessor::GetLibraryAppletInfo>, "GetLibraryAppletInfo"},
         {150, nullptr, "RequestForAppletToGetForeground"},
         {160, D<&ILibraryAppletAccessor::GetIndirectLayerConsumerHandle>, "GetIndirectLayerConsumerHandle"}, //2.0.0+
         {170, D<&ILibraryAppletAccessor::Unknown170>, "Unknown170"}, //22.0.0+
@@ -218,6 +218,16 @@ Result ILibraryAppletAccessor::GetIndirectLayerConsumerHandle(Out<u64> out_handl
     R_SUCCEED();
 }
 
+Result ILibraryAppletAccessor::GetLibraryAppletInfo(
+    Out<LibraryAppletInfo> out_library_applet_info) {
+    LOG_INFO(Service_AM, "called");
+    *out_library_applet_info = {
+        .applet_id = m_applet->applet_id,
+        .library_applet_mode = m_applet->library_applet_mode,
+    };
+    R_SUCCEED();
+}
+
 Result ILibraryAppletAccessor::Unknown170(OutCopyHandle<Kernel::KReadableEvent> out_event) {
     LOG_WARNING(Service_AM, "(STUBBED) called");
     *out_event = m_applet->unknown_event.GetHandle();

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

@@ -6,6 +6,7 @@
 
 #pragma once
 
+#include "core/hle/service/am/service/library_applet_self_accessor.h"
 #include "core/hle/service/cmif_types.h"
 #include "core/hle/service/service.h"
 
@@ -21,6 +22,10 @@ public:
                                     std::shared_ptr<Applet> applet);
     ~ILibraryAppletAccessor();
 
+    std::shared_ptr<Applet> GetApplet() const {
+        return m_applet;
+    }
+
 private:
     Result GetAppletStateChangedEvent(OutCopyHandle<Kernel::KReadableEvent> out_event);
     Result IsCompleted(Out<bool> out_is_completed);
@@ -37,6 +42,7 @@ private:
     Result GetPopOutDataEvent(OutCopyHandle<Kernel::KReadableEvent> out_event);
     Result GetPopInteractiveOutDataEvent(OutCopyHandle<Kernel::KReadableEvent> out_event);
     Result GetIndirectLayerConsumerHandle(Out<u64> out_handle);
+    Result GetLibraryAppletInfo(Out<LibraryAppletInfo> out_library_applet_info);
     Result Unknown170(OutCopyHandle<Kernel::KReadableEvent> out_event);
 
     void FrontendExecute();

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

@@ -123,7 +123,7 @@ std::shared_ptr<ILibraryAppletAccessor> CreateGuestApplet(Core::System& system,
 
     auto process = CreateProcess(system, program_id, Firmware1400, Firmware2200);
     if (process) {
-        const auto applet = std::make_shared<Applet>(system, std::make_unique<Service::Process>(*std::move(process)), false);
+        const auto applet = std::make_shared<Applet>(system, std::move(process), false);
         applet->program_id = program_id;
         applet->applet_id = applet_id;
         applet->type = AppletType::LibraryApplet;

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

@@ -233,8 +233,9 @@ Result ILibraryAppletSelfAccessor::ReportVisibleErrorWithErrorContext(
     R_SUCCEED();
 }
 
-Result ILibraryAppletSelfAccessor::UnpopInData() {
-    LOG_WARNING(Service_AM, "(STUBBED) called");
+Result ILibraryAppletSelfAccessor::UnpopInData(SharedPointer<IStorage> storage) {
+    LOG_INFO(Service_AM, "called");
+    m_broker->GetInData().Unpop(system.Kernel(), storage);
     R_SUCCEED();
 }
 

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

@@ -72,7 +72,7 @@ private:
     Result ReportVisibleError(ErrorCode error_code);
     Result ReportVisibleErrorWithErrorContext(
         ErrorCode error_code, InLargeData<ErrorContext, BufferAttr_HipcMapAlias> error_context);
-    Result UnpopInData();
+    Result UnpopInData(SharedPointer<IStorage> storage);
     Result GetMainAppletApplicationDesiredLanguage(Out<u64> out_desired_language);
     Result GetCurrentApplicationId(Out<u64> out_application_id);
     Result GetMainAppletAvailableUsers(Out<bool> out_can_select_any_user, Out<s32> out_users_count,

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

@@ -1,9 +1,11 @@
-// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
 // SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
+#include "core/core.h"
+#include "core/hle/service/am/applet.h"
 #include "core/hle/service/am/frontend/applets.h"
 #include "core/hle/service/am/service/library_applet_accessor.h"
 #include "core/hle/service/am/service/process_winding_controller.h"
@@ -43,6 +45,18 @@ Result IProcessWindingController::OpenCallingLibraryApplet(
     Out<SharedPointer<ILibraryAppletAccessor>> out_calling_library_applet) {
     LOG_INFO(Service_AM, "called");
 
+    std::shared_ptr<Applet> reserved_applet;
+    {
+        std::scoped_lock lk{m_applet->lock};
+        reserved_applet = std::move(m_applet->reserved_applet);
+    }
+
+    if (reserved_applet != nullptr) {
+        *out_calling_library_applet = std::make_shared<ILibraryAppletAccessor>(
+            system, reserved_applet->caller_applet_broker, reserved_applet);
+        R_SUCCEED();
+    }
+
     const auto caller_applet = m_applet->caller_applet.lock();
     if (caller_applet == nullptr) {
         LOG_ERROR(Service_AM, "No caller applet available");
@@ -74,22 +88,82 @@ Result IProcessWindingController::PopContext(Out<SharedPointer<IStorage>> out_co
 }
 
 Result IProcessWindingController::CancelWindingReservation() {
-    LOG_WARNING(Service_AM, "STUBBED");
+    LOG_INFO(Service_AM, "called");
+
+    std::scoped_lock lk{m_applet->lock};
+    m_applet->reserved_applet.reset();
+    m_applet->unwind_after_reserved = false;
+
     R_SUCCEED();
 }
 
 Result IProcessWindingController::WindAndDoReserved() {
-    LOG_WARNING(Service_AM, "STUBBED");
+    LOG_INFO(Service_AM, "called");
+
+    std::shared_ptr<Applet> reserved_applet;
+    {
+        std::scoped_lock lk{m_applet->lock};
+
+        reserved_applet = m_applet->reserved_applet;
+        m_applet->display_layer_manager.SetWindowVisibility(false);
+        m_applet->exit_locked = false;
+        system.SetExitLocked(false);
+    }
+
+    if (reserved_applet) {
+        {
+            std::scoped_lock lk{m_applet->lock};
+            m_applet->is_winding = true;
+        }
+
+        {
+            std::scoped_lock lk{reserved_applet->lock};
+            reserved_applet->window_visible = true;
+            reserved_applet->process->Run();
+        }
+
+        if (reserved_applet->frontend) {
+            reserved_applet->frontend->Initialize();
+            reserved_applet->frontend->Execute();
+        }
+    } else {
+        LOG_WARNING(Service_AM, "called without a reserved applet to start");
+    }
+
+    m_applet->process->Terminate();
+
     R_SUCCEED();
 }
 
-Result IProcessWindingController::ReserveToStartAndWaitAndUnwindThis() {
-    LOG_WARNING(Service_AM, "STUBBED");
+Result IProcessWindingController::ReserveToStartAndWaitAndUnwindThis(
+    SharedPointer<ILibraryAppletAccessor> reserved_applet_accessor) {
+    LOG_INFO(Service_AM, "called");
+
+    if (reserved_applet_accessor == nullptr) {
+        LOG_ERROR(Service_AM, "No applet accessor provided");
+        R_THROW(ResultUnknown);
+    }
+
+    std::scoped_lock lk{m_applet->lock};
+    m_applet->reserved_applet = reserved_applet_accessor->GetApplet();
+    m_applet->unwind_after_reserved = true;
+
     R_SUCCEED();
 }
 
-Result IProcessWindingController::ReserveToStartAndWait() {
-    LOG_WARNING(Service_AM, "STUBBED");
+Result IProcessWindingController::ReserveToStartAndWait(
+    SharedPointer<ILibraryAppletAccessor> reserved_applet_accessor) {
+    LOG_INFO(Service_AM, "called");
+
+    if (reserved_applet_accessor == nullptr) {
+        LOG_ERROR(Service_AM, "No applet accessor provided");
+        R_THROW(ResultUnknown);
+    }
+
+    std::scoped_lock lk{m_applet->lock};
+    m_applet->reserved_applet = reserved_applet_accessor->GetApplet();
+    m_applet->unwind_after_reserved = false;
+
     R_SUCCEED();
 }
 

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

@@ -1,4 +1,4 @@
-// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
 // SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project
@@ -29,8 +29,9 @@ private:
     Result PopContext(Out<SharedPointer<IStorage>> out_context);
     Result CancelWindingReservation();
     Result WindAndDoReserved();
-    Result ReserveToStartAndWaitAndUnwindThis();
-    Result ReserveToStartAndWait();
+    Result ReserveToStartAndWaitAndUnwindThis(
+        SharedPointer<ILibraryAppletAccessor> reserved_applet_accessor);
+    Result ReserveToStartAndWait(SharedPointer<ILibraryAppletAccessor> reserved_applet_accessor);
 
     const std::shared_ptr<Applet> m_applet;
 };

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

@@ -9,6 +9,7 @@
 #include "core/hle/service/am/applet.h"
 #include "core/hle/service/am/applet_manager.h"
 #include "core/hle/service/am/event_observer.h"
+#include "core/hle/service/am/process_creation.h"
 #include "core/hle/service/am/window_system.h"
 
 namespace Service::AM {
@@ -240,6 +241,37 @@ void WindowSystem::PruneTerminatedAppletsLocked() {
             continue;
         }
 
+        // A winding applet has had its own process killed but is kept alive as a transparent slot
+        // while the reserved applet running in its place finishes (see WindAndDoReserved()).
+        if (applet->is_winding) {
+            if (!applet->child_applets.empty()) {
+                it = std::next(it);
+                continue;
+            }
+
+            const bool unwind = applet->unwind_after_reserved;
+            applet->is_winding = false;
+            applet->unwind_after_reserved = false;
+
+            if (unwind && this->RestartAppletProcessLocked(applet.get())) {
+                const u64 new_aruid = applet->aruid.pid;
+                const auto next = std::next(it);
+
+                if (new_aruid != aruid) {
+                    auto node = m_applets.extract(it);
+                    node.key() = new_aruid;
+                    m_applets.insert(std::move(node));
+                }
+
+                applet->process->Run();
+                m_event_observer->RequestUpdate();
+                it = next;
+                continue;
+            }
+
+            applet->reserved_applet.reset();
+        }
+
         // Terminated, so ensure all child applets are terminated.
         if (!applet->child_applets.empty()) {
             this->TerminateChildAppletsLocked(applet.get());
@@ -301,6 +333,28 @@ void WindowSystem::PruneTerminatedAppletsLocked() {
     }
 }
 
+bool WindowSystem::RestartAppletProcessLocked(Applet* applet) {
+    if (!ReinitializeProcess(m_system, *applet->process, applet->program_id)) {
+        LOG_ERROR(Service_AM, "Failed to restart winding applet_id={}",
+                  static_cast<u32>(applet->applet_id));
+        return false;
+    }
+
+    applet->aruid.pid = applet->process->GetProcessId();
+    applet->is_process_running = false;
+    applet->is_completed = false;
+
+    applet->hid_registration.RegisterCurrentProcess();
+
+    applet->lifecycle_manager.ResetForRelaunch();
+    applet->is_activity_runnable = false;
+
+    applet->launch_reason.flag = 1;
+
+    m_event_observer->TrackAppletProcess(*applet);
+    return true;
+}
+
 bool WindowSystem::LockHomeMenuIntoForegroundLocked() {
     // If the home menu is not locked into foreground, then there's nothing to do.
     if (m_home_menu == nullptr || !m_home_menu_foreground_locked) {
@@ -352,6 +406,9 @@ void WindowSystem::UpdateAppletStateLocked(Applet* applet, bool is_foreground, b
     const bool has_obscuring_child_applets = [&] {
         for (const auto& child_applet : applet->child_applets) {
             std::scoped_lock lk2{child_applet->lock};
+            if (child_applet->is_winding) {
+                return true;
+            }
             const auto mode = child_applet->library_applet_mode;
             if (child_applet->is_process_running && child_applet->window_visible &&
                 (mode == LibraryAppletMode::AllForeground ||

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

@@ -61,6 +61,7 @@ public:
 
 private:
     void PruneTerminatedAppletsLocked();
+    bool RestartAppletProcessLocked(Applet* applet);
     bool LockHomeMenuIntoForegroundLocked();
     void TerminateChildAppletsLocked(Applet* applet);
     void UpdateAppletStateLocked(Applet* applet, bool is_foreground, bool overlay_blocking = false);

src/core/hle/service/apm/apm_interface.cpp

@@ -82,7 +82,7 @@ void APM::OpenSession(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<ISession>(system, controller);
+    rb.PushIpcInterface<ISession>(ctx, system, controller);
 }
 
 void APM::GetPerformanceMode(HLERequestContext& ctx) {
@@ -125,7 +125,7 @@ void APM_Sys::GetPerformanceEvent(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<ISession>(system, controller);
+    rb.PushIpcInterface<ISession>(ctx, system, controller);
 }
 
 void APM_Sys::SetCpuBoostMode(HLERequestContext& ctx) {

src/core/hle/service/fgm/fgm.cpp

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2018 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -45,7 +48,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IRequest>(system);
+        rb.PushIpcInterface<IRequest>(ctx, system);
     }
 };
 

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

@@ -184,7 +184,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 1};
         rb.Push(readable_event.Signal(system.Kernel()));
-        rb.PushCopyObjects(readable_event);
+        rb.PushCopyObjects(ctx, readable_event);
     }
 
     void Cancel(HLERequestContext& ctx) {
@@ -400,7 +400,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 1};
         rb.Push(ResultSuccess);
-        rb.PushCopyObjects(notification_event->GetReadableEvent());
+        rb.PushCopyObjects(ctx, notification_event->GetReadableEvent());
     }
 
     void Clear(HLERequestContext& ctx) {
@@ -476,7 +476,7 @@ private:
 void Module::Interface::CreateFriendService(HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IFriendService>(system);
+    rb.PushIpcInterface<IFriendService>(ctx, system);
     LOG_DEBUG(Service_Friend, "called");
 }
 
@@ -488,12 +488,12 @@ void Module::Interface::CreateNotificationService(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<INotificationService>(system, uuid);
+    rb.PushIpcInterface<INotificationService>(ctx, system, uuid);
 }
 
-Module::Interface::Interface(std::shared_ptr<Module> module_, Core::System& system_,
-                             const char* name)
-    : ServiceFramework{system_, name}, module{std::move(module_)} {}
+Module::Interface::Interface(std::shared_ptr<Module> module_, Core::System& system_, const char* name)
+    : ServiceFramework{system_, name}, module{std::move(module_)}
+{}
 
 Module::Interface::~Interface() = default;
 

src/core/hle/service/glue/arp.cpp

@@ -279,7 +279,7 @@ void ARP_W::AcquireRegistrar(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface(registrar);
+    rb.PushIpcInterface(ctx, registrar);
 }
 
 void ARP_W::UnregisterApplicationInstance(HLERequestContext& ctx) {

src/core/hle/service/glue/bgtc.cpp

@@ -28,7 +28,7 @@ void BGTC_T::OpenTaskService(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<ITaskService>(system);
+    rb.PushIpcInterface<ITaskService>(ctx, system);
 }
 
 ITaskService::ITaskService(Core::System& system_) : ServiceFramework{system_, "ITaskService"} {

src/core/hle/service/glue/ectx.cpp

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -56,7 +59,7 @@ ECTX_AW::~ECTX_AW() = default;
 void ECTX_AW::CreateContextRegistrar(HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IContextRegistrar>(std::make_shared<IContextRegistrar>(system));
+    rb.PushIpcInterface<IContextRegistrar>(ctx, system);
 }
 
 } // namespace Service::Glue

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

@@ -726,7 +726,7 @@ void IHidSystemServer::AcquireConnectionTriggerTimeoutEvent(HLERequestContext& c
 
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(ResultSuccess);
-    rb.PushCopyObjects(acquire_device_registered_event->GetReadableEvent());
+    rb.PushCopyObjects(ctx, acquire_device_registered_event->GetReadableEvent());
 }
 
 void IHidSystemServer::AcquireDeviceRegisteredEventForControllerSupport(HLERequestContext& ctx) {
@@ -734,7 +734,7 @@ void IHidSystemServer::AcquireDeviceRegisteredEventForControllerSupport(HLEReque
 
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(ResultSuccess);
-    rb.PushCopyObjects(acquire_device_registered_event->GetReadableEvent());
+    rb.PushCopyObjects(ctx, acquire_device_registered_event->GetReadableEvent());
 }
 
 void IHidSystemServer::GetRegisteredDevices(HLERequestContext& ctx) {
@@ -759,7 +759,7 @@ void IHidSystemServer::AcquireUniquePadConnectionEventHandle(HLERequestContext&
     LOG_WARNING(Service_HID, "(STUBBED) called");
 
     IPC::ResponseBuilder rb{ctx, 2, 1};
-    rb.PushCopyObjects(unique_pad_connection_event->GetReadableEvent());
+    rb.PushCopyObjects(ctx, unique_pad_connection_event->GetReadableEvent());
     rb.Push(ResultSuccess);
 }
 
@@ -776,7 +776,7 @@ void IHidSystemServer::AcquireJoyDetachOnBluetoothOffEventHandle(HLERequestConte
 
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(ResultSuccess);
-    rb.PushCopyObjects(joy_detach_event->GetReadableEvent());
+    rb.PushCopyObjects(ctx, joy_detach_event->GetReadableEvent());
 }
 
 void IHidSystemServer::IsUsbFullKeyControllerEnabled(HLERequestContext& ctx) {

src/core/hle/service/hle_ipc.h

@@ -31,7 +31,7 @@ class Memory;
 }
 
 namespace IPC {
-class ResponseBuilder;
+struct ResponseBuilder;
 }
 
 namespace Service {
@@ -392,7 +392,7 @@ public:
     }
 
 private:
-    friend class IPC::ResponseBuilder;
+    friend struct IPC::ResponseBuilder;
 
     void ParseCommandBuffer(u32_le* src_cmdbuf, bool incoming);
 

src/core/hle/service/ipc_helpers.h

@@ -24,45 +24,7 @@ namespace IPC {
 
 constexpr Result ResultSessionClosed{ErrorModule::HIPC, 301};
 
-class RequestHelperBase {
-protected:
-    Service::HLERequestContext* context = nullptr;
-    u32* cmdbuf;
-    u32 index = 0;
-
-public:
-    explicit RequestHelperBase(u32* command_buffer) : cmdbuf(command_buffer) {}
-
-    explicit RequestHelperBase(Service::HLERequestContext& ctx)
-        : context(&ctx), cmdbuf(ctx.CommandBuffer()) {}
-
-    void Skip(u32 size_in_words, bool set_to_null) {
-        if (set_to_null) {
-            memset(cmdbuf + index, 0, size_in_words * sizeof(u32));
-        }
-        index += size_in_words;
-    }
-
-    /**
-     * Aligns the current position forward to a 16-byte boundary, padding with zeros.
-     */
-    void AlignWithPadding() {
-        if (index & 3) {
-            Skip(static_cast<u32>(4 - (index & 3)), true);
-        }
-    }
-
-    u32 GetCurrentOffset() const {
-        return index;
-    }
-
-    void SetCurrentOffset(u32 offset) {
-        index = offset;
-    }
-};
-
-class ResponseBuilder : public RequestHelperBase {
-public:
+struct ResponseBuilder {
     /// Flags used for customizing the behavior of ResponseBuilder
     enum class Flags : u32 {
         None = 0,
@@ -71,14 +33,13 @@ public:
         AlwaysMoveHandles = 1,
     };
 
-    explicit ResponseBuilder(Service::HLERequestContext& ctx, u32 normal_params_size_,
-                             u32 num_handles_to_copy_ = 0, u32 num_objects_to_move_ = 0,
-                             Flags flags = Flags::None)
-        : RequestHelperBase(ctx), normal_params_size(normal_params_size_),
-          num_handles_to_copy(num_handles_to_copy_),
-          num_objects_to_move(num_objects_to_move_), kernel{ctx.kernel} {
-
-        memset(cmdbuf, 0, sizeof(u32) * IPC::COMMAND_BUFFER_LENGTH);
+    inline explicit ResponseBuilder(Service::HLERequestContext& ctx, u32 normal_params_size_, u32 num_handles_to_copy_ = 0, u32 num_objects_to_move_ = 0, Flags flags = Flags::None)
+        : cmdbuf(ctx.CommandBuffer())
+        , normal_params_size(normal_params_size_)
+        , num_handles_to_copy(num_handles_to_copy_)
+        , num_objects_to_move(num_objects_to_move_)
+    {
+        std::memset(cmdbuf, 0, sizeof(u32) * IPC::COMMAND_BUFFER_LENGTH);
 
         IPC::CommandHeader header{};
         auto const mgr = ctx.GetManager().get();
@@ -117,9 +78,7 @@ public:
             handle_descriptor_header.num_handles_to_copy.Assign(num_handles_to_copy_);
             handle_descriptor_header.num_handles_to_move.Assign(num_handles_to_move);
             PushRaw(handle_descriptor_header);
-
             ctx.handles_offset = index;
-
             Skip(num_handles_to_copy + num_handles_to_move, true);
         }
 
@@ -131,7 +90,6 @@ public:
                 domain_header.num_objects = num_domain_objects;
                 PushRaw(domain_header);
             }
-
             IPC::DataPayloadHeader data_payload_header{};
             data_payload_header.magic = Common::MakeMagic('S', 'F', 'C', 'O');
             PushRaw(data_payload_header);
@@ -141,34 +99,39 @@ public:
 
         ctx.data_payload_offset = index;
         ctx.write_size += index;
-        ctx.domain_offset = static_cast<u32>(index + raw_data_size / sizeof(u32));
+        ctx.domain_offset = u32(index + raw_data_size / sizeof(u32));
     }
 
-    template <class T>
-    void PushIpcInterface(std::shared_ptr<T> iface) {
-        auto manager{context->GetManager()};
+    inline void Skip(u32 size_in_words, bool set_to_null) {
+        if (set_to_null) std::memset(cmdbuf + index, 0, size_in_words * sizeof(u32));
+        index += size_in_words;
+    }
+    /// @brief Aligns the current position forward to a 16-byte boundary, padding with zeros.
+    inline void AlignWithPadding() { if (index & 3) Skip(u32(4 - (index & 3)), true); }
+    inline u32 GetCurrentOffset() const { return index; }
+    inline void SetCurrentOffset(u32 offset) { index = offset; }
 
+    template <class T> inline void PushIpcInterface(Service::HLERequestContext& ctx, std::shared_ptr<T> iface) {
+        auto manager = ctx.GetManager();
         if (manager->IsDomain()) {
-            context->AddDomainObject(std::move(iface));
+            ctx.AddDomainObject(std::move(iface));
         } else {
-            ASSERT(Kernel::GetCurrentProcess(kernel).GetResourceLimit()->Reserve(kernel, Kernel::LimitableResource::SessionCountMax, 1));
+            ASSERT(Kernel::GetCurrentProcess(ctx.kernel).GetResourceLimit()->Reserve(ctx.kernel, Kernel::LimitableResource::SessionCountMax, 1));
 
-            auto* session = Kernel::KSession::Create(kernel);
-            session->Initialize(kernel, nullptr, 0);
-            Kernel::KSession::Register(kernel, session);
+            auto* session = Kernel::KSession::Create(ctx.kernel);
+            session->Initialize(ctx.kernel, nullptr, 0);
+            Kernel::KSession::Register(ctx.kernel, session);
 
-            auto next_manager = std::make_shared<Service::SessionRequestManager>(
-                kernel, manager->GetServerManager());
+            auto next_manager = std::make_shared<Service::SessionRequestManager>(ctx.kernel, manager->GetServerManager());
             next_manager->SetSessionHandler(iface);
             manager->GetServerManager().RegisterSession(&session->GetServerSession(), next_manager);
 
-            context->AddMoveObject(&session->GetClientSession());
+            ctx.AddMoveObject(&session->GetClientSession());
         }
     }
 
-    template <class T, class... Args>
-    void PushIpcInterface(Args&&... args) {
-        PushIpcInterface<T>(std::make_shared<T>(std::forward<Args>(args)...));
+    template <class T, class... Args> inline void PushIpcInterface(Service::HLERequestContext& ctx, Args&&... args) {
+        PushIpcInterface<T>(ctx, std::make_shared<T>(std::forward<Args>(args)...));
     }
 
     void PushImpl(s8 value);
@@ -184,59 +147,38 @@ public:
     void PushImpl(bool value);
     void PushImpl(Result value);
 
-    template <typename T>
-    void Push(T value) {
+    template <typename T> inline void Push(T value) {
         return PushImpl(value);
     }
 
     template <typename First, typename... Other>
     void Push(const First& first_value, const Other&... other_values);
 
-    /**
-     * Helper function for pushing strongly-typed enumeration values.
-     *
-     * @tparam Enum The enumeration type to be pushed
-     *
-     * @param value The value to push.
-     *
-     * @note The underlying size of the enumeration type is the size of the
-     *       data that gets pushed. e.g. "enum class SomeEnum : u16" will
-     *       push a u16-sized amount of data.
-     */
-    template <typename Enum>
-    void PushEnum(Enum value) {
+    /// @brief Helper function for pushing strongly-typed enumeration values.
+    /// @tparam Enum The enumeration type to be pushed
+    /// @param value The value to push.
+    /// @note The underlying size of the enumeration type is the size of the data that gets pushed.
+    /// e.g. "enum class SomeEnum : u16" will push a u16-sized amount of data.
+    template <typename Enum> inline void PushEnum(Enum value) {
         static_assert(std::is_enum_v<Enum>, "T must be an enum type within a PushEnum call.");
-        static_assert(!std::is_convertible_v<Enum, int>,
-                      "enum type in PushEnum must be a strongly typed enum.");
+        static_assert(!std::is_convertible_v<Enum, int>, "enum type in PushEnum must be a strongly typed enum.");
         Push(static_cast<std::underlying_type_t<Enum>>(value));
     }
 
-    /**
-     * @brief Copies the content of the given trivially copyable class to the buffer as a normal
-     * param
-     * @note: The input class must be correctly packed/padded to fit hardware layout.
-     */
-    template <typename T>
-    void PushRaw(const T& value);
+    /// @brief Copies the content of the given trivially copyable class to the buffer as a normal param
+    /// @note: The input class must be correctly packed/padded to fit hardware layout.
+    template <typename T> void PushRaw(const T& value);
+    template <typename... O> void PushMoveObjects(Service::HLERequestContext& ctx, O*... pointers);
+    template <typename... O> void PushMoveObjects(Service::HLERequestContext& ctx, O&... pointers);
+    template <typename... O> void PushCopyObjects(Service::HLERequestContext& ctx, O*... pointers);
+    template <typename... O> void PushCopyObjects(Service::HLERequestContext& ctx, O&... pointers);
 
-    template <typename... O>
-    void PushMoveObjects(O*... pointers);
-
-    template <typename... O>
-    void PushMoveObjects(O&... pointers);
-
-    template <typename... O>
-    void PushCopyObjects(O*... pointers);
-
-    template <typename... O>
-    void PushCopyObjects(O&... pointers);
-
-private:
+    u32* cmdbuf;
+    u32 index = 0;
     u32 normal_params_size{};
     u32 num_handles_to_copy{};
     u32 num_objects_to_move{}; ///< Domain objects or move handles, context dependent
     u32 data_payload_index{};
-    Kernel::KernelCore& kernel;
 };
 
 /// Push ///
@@ -251,8 +193,7 @@ inline void ResponseBuilder::PushImpl(u32 value) {
 
 template <typename T>
 void ResponseBuilder::PushRaw(const T& value) {
-    static_assert(std::is_trivially_copyable_v<T>,
-                  "It's undefined behavior to use memcpy with non-trivially copyable objects");
+    static_assert(std::is_trivially_copyable_v<T>, "It's undefined behavior to use memcpy with non-trivially copyable objects");
     std::memcpy(cmdbuf + index, &value, sizeof(T));
     index += (sizeof(T) + 3) / 4; // round up to word length
 }
@@ -272,8 +213,8 @@ inline void ResponseBuilder::PushImpl(s16 value) {
 }
 
 inline void ResponseBuilder::PushImpl(s64 value) {
-    PushImpl(static_cast<u32>(value));
-    PushImpl(static_cast<u32>(value >> 32));
+    PushImpl(u32(value));
+    PushImpl(u32(value >> 32));
 }
 
 inline void ResponseBuilder::PushImpl(u8 value) {
@@ -285,8 +226,8 @@ inline void ResponseBuilder::PushImpl(u16 value) {
 }
 
 inline void ResponseBuilder::PushImpl(u64 value) {
-    PushImpl(static_cast<u32>(value));
-    PushImpl(static_cast<u32>(value >> 32));
+    PushImpl(u32(value));
+    PushImpl(u32(value >> 32));
 }
 
 inline void ResponseBuilder::PushImpl(float value) {
@@ -312,90 +253,88 @@ void ResponseBuilder::Push(const First& first_value, const Other&... other_value
 }
 
 template <typename... O>
-inline void ResponseBuilder::PushCopyObjects(O*... pointers) {
+inline void ResponseBuilder::PushCopyObjects(Service::HLERequestContext& ctx, O*... pointers) {
     auto objects = {pointers...};
     for (auto& object : objects) {
-        context->AddCopyObject(object);
+        ctx.AddCopyObject(object);
     }
 }
 
 template <typename... O>
-inline void ResponseBuilder::PushCopyObjects(O&... pointers) {
+inline void ResponseBuilder::PushCopyObjects(Service::HLERequestContext& ctx, O&... pointers) {
     auto objects = {&pointers...};
     for (auto& object : objects) {
-        context->AddCopyObject(object);
+        ctx.AddCopyObject(object);
     }
 }
 
 template <typename... O>
-inline void ResponseBuilder::PushMoveObjects(O*... pointers) {
+inline void ResponseBuilder::PushMoveObjects(Service::HLERequestContext& ctx, O*... pointers) {
     auto objects = {pointers...};
     for (auto& object : objects) {
-        context->AddMoveObject(object);
+        ctx.AddMoveObject(object);
     }
 }
 
 template <typename... O>
-inline void ResponseBuilder::PushMoveObjects(O&... pointers) {
+inline void ResponseBuilder::PushMoveObjects(Service::HLERequestContext& ctx, O&... pointers) {
     auto objects = {&pointers...};
     for (auto& object : objects) {
-        context->AddMoveObject(object);
+        ctx.AddMoveObject(object);
     }
 }
 
-class RequestParser : public RequestHelperBase {
-public:
-    explicit RequestParser(u32* command_buffer) : RequestHelperBase(command_buffer) {}
-
-    explicit RequestParser(Service::HLERequestContext& ctx) : RequestHelperBase(ctx) {
+struct RequestParser {
+    inline explicit RequestParser(u32* command_buffer) : cmdbuf(command_buffer) {}
+    inline explicit RequestParser(Service::HLERequestContext& ctx)
+        : cmdbuf(ctx.CommandBuffer())
+    {
         // TIPC does not have data payload offset
         if (!ctx.IsTipc()) {
             ASSERT_MSG(ctx.GetDataPayloadOffset(), "context is incomplete");
             Skip(ctx.GetDataPayloadOffset(), false);
         }
-
         // Skip the u64 command id, it's already stored in the context
         static constexpr u32 CommandIdSize = 2;
         Skip(CommandIdSize, false);
     }
 
-    template <typename T>
-    T Pop();
+    inline void Skip(u32 size_in_words, bool set_to_null) {
+        if (set_to_null) std::memset(cmdbuf + index, 0, size_in_words * sizeof(u32));
+        index += size_in_words;
+    }
+    /// @brief Aligns the current position forward to a 16-byte boundary, padding with zeros.
+    inline void AlignWithPadding() { if (index & 3) Skip(u32(4 - (index & 3)), true); }
+    inline u32 GetCurrentOffset() const { return index; }
+    inline void SetCurrentOffset(u32 offset) { index = offset; }
 
-    template <typename T>
-    void Pop(T& value);
-
-    template <typename First, typename... Other>
-    void Pop(First& first_value, Other&... other_values);
+    template <typename T> T Pop();
+    template <typename T> void Pop(T& value);
+    template <typename First, typename... Other> void Pop(First& first_value, Other&... other_values);
 
     template <typename T>
     T PopEnum() {
         static_assert(std::is_enum_v<T>, "T must be an enum type within a PopEnum call.");
-        static_assert(!std::is_convertible_v<T, int>,
-                      "enum type in PopEnum must be a strongly typed enum.");
-        return static_cast<T>(Pop<std::underlying_type_t<T>>());
+        static_assert(!std::is_convertible_v<T, int>, "enum type in PopEnum must be a strongly typed enum.");
+        return T(Pop<std::underlying_type_t<T>>());
     }
 
-    /**
-     * @brief Reads the next normal parameters as a struct, by copying it
-     * @note: The output class must be correctly packed/padded to fit hardware layout.
-     */
-    template <typename T>
-    void PopRaw(T& value);
+    /// @brief Reads the next normal parameters as a struct, by copying it
+    /// @note: The output class must be correctly packed/padded to fit hardware layout.
+    template <typename T> void PopRaw(T& value);
 
-    /**
-     * @brief Reads the next normal parameters as a struct, by copying it into a new value
-     * @note: The output class must be correctly packed/padded to fit hardware layout.
-     */
-    template <typename T>
-    T PopRaw();
+    /// @brief Reads the next normal parameters as a struct, by copying it into a new value
+    /// @note: The output class must be correctly packed/padded to fit hardware layout.
+    template <typename T> T PopRaw();
 
-    template <class T>
-    std::weak_ptr<T> PopIpcInterface() {
-        ASSERT(context->GetManager()->IsDomain());
-        ASSERT(context->GetDomainMessageHeader().input_object_count > 0);
-        return context->GetDomainHandler<T>(Pop<u32>() - 1);
+    template <class T> [[nodiscard]] std::weak_ptr<T> PopIpcInterface(Service::HLERequestContext& ctx) {
+        ASSERT(ctx.GetManager()->IsDomain());
+        ASSERT(ctx.GetDomainMessageHeader().input_object_count > 0);
+        return ctx.GetDomainHandler<T>(Pop<u32>() - 1);
     }
+
+    u32* cmdbuf;
+    u32 index = 0;
 };
 
 /// Pop ///
@@ -407,7 +346,7 @@ inline u32 RequestParser::Pop() {
 
 template <>
 inline s32 RequestParser::Pop() {
-    return static_cast<s32>(Pop<u32>());
+    return s32(Pop<u32>());
 }
 
 // Ignore the -Wclass-memaccess warning on memcpy for non-trivially default constructible objects.
@@ -417,8 +356,7 @@ inline s32 RequestParser::Pop() {
 #endif
 template <typename T>
 void RequestParser::PopRaw(T& value) {
-    static_assert(std::is_trivially_copyable_v<T>,
-                  "It's undefined behavior to use memcpy with non-trivially copyable objects");
+    static_assert(std::is_trivially_copyable_v<T>, "It's undefined behavior to use memcpy with non-trivially copyable objects");
     std::memcpy(&value, cmdbuf + index, sizeof(T));
     index += (sizeof(T) + 3) / 4; // round up to word length
 }

src/core/hle/service/lm/lm.cpp

@@ -353,7 +353,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<ILogger>(system);
+        rb.PushIpcInterface<ILogger>(ctx, system);
     }
 };
 

src/core/hle/service/nfc/nfc.cpp

@@ -151,7 +151,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IAm>(system);
+        rb.PushIpcInterface<IAm>(ctx, system);
     }
 };
 
@@ -173,7 +173,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<MFIUser>(system);
+        rb.PushIpcInterface<MFIUser>(ctx, system);
     }
 };
 
@@ -195,7 +195,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IUser>(system);
+        rb.PushIpcInterface<IUser>(ctx, system);
     }
 };
 
@@ -217,7 +217,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<ISystem>(system);
+        rb.PushIpcInterface<ISystem>(ctx, system);
     }
 };
 

src/core/hle/service/nfc/nfc_interface.cpp

@@ -143,7 +143,7 @@ void NfcInterface::AttachAvailabilityChangeEvent(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(ResultSuccess);
-    rb.PushCopyObjects(GetManager()->AttachAvailabilityChangeEvent());
+    rb.PushCopyObjects(ctx, GetManager()->AttachAvailabilityChangeEvent());
 }
 
 void NfcInterface::StartDetection(HLERequestContext& ctx) {
@@ -203,7 +203,7 @@ void NfcInterface::AttachActivateEvent(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(result);
-    rb.PushCopyObjects(out_event);
+    rb.PushCopyObjects(ctx, out_event);
 }
 
 void NfcInterface::AttachDeactivateEvent(HLERequestContext& ctx) {
@@ -217,7 +217,7 @@ void NfcInterface::AttachDeactivateEvent(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(result);
-    rb.PushCopyObjects(out_event);
+    rb.PushCopyObjects(ctx, out_event);
 }
 
 void NfcInterface::SetNfcEnabled(HLERequestContext& ctx) {

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

@@ -157,7 +157,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IUser>(system);
+        rb.PushIpcInterface<IUser>(ctx, system);
     }
 };
 
@@ -179,7 +179,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<ISystem>(system);
+        rb.PushIpcInterface<ISystem>(ctx, system);
     }
 };
 
@@ -201,7 +201,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IDebug>(system);
+        rb.PushIpcInterface<IDebug>(ctx, system);
     }
 };
 

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

@@ -307,7 +307,7 @@ private:
     void GetSystemEventReadableHandle(HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 2};
         rb.Push(ResultSuccess);
-        rb.PushCopyObjects(evt_scan_complete->GetReadableEvent(),
+        rb.PushCopyObjects(ctx, evt_scan_complete->GetReadableEvent(),
                            evt_processing->GetReadableEvent());
     }
 
@@ -452,7 +452,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 2};
         rb.Push(ResultSuccess);
-        rb.PushCopyObjects(event1->GetReadableEvent(), event2->GetReadableEvent());
+        rb.PushCopyObjects(ctx, event1->GetReadableEvent(), event2->GetReadableEvent());
     }
 
     void Cancel(HLERequestContext& ctx) {
@@ -528,7 +528,7 @@ void IGeneralService::CreateScanRequest(HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
 
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IScanRequest>(system);
+    rb.PushIpcInterface<IScanRequest>(ctx, system);
 }
 
 void IGeneralService::CreateRequest(HLERequestContext& ctx) {
@@ -537,7 +537,7 @@ void IGeneralService::CreateRequest(HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
 
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IRequest>(system);
+    rb.PushIpcInterface<IRequest>(ctx, system);
 }
 
 void IGeneralService::GetCurrentNetworkProfile(HLERequestContext& ctx) {
@@ -716,7 +716,7 @@ void IGeneralService::GetNetworkProfile(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<INetworkProfile>(system);
+    rb.PushIpcInterface<INetworkProfile>(ctx, system);
 }
 
 void IGeneralService::SetNetworkProfile(HLERequestContext& ctx) {
@@ -869,7 +869,7 @@ void IGeneralService::CreateTemporaryNetworkProfile(HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 6, 0, 1};
 
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<INetworkProfile>(system);
+    rb.PushIpcInterface<INetworkProfile>(ctx, system);
     rb.PushRaw<u128>(uuid);
 }
 
@@ -1124,7 +1124,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IGeneralService>(system);
+        rb.PushIpcInterface<IGeneralService>(ctx, system);
     }
 
     void CreateGeneralService(HLERequestContext& ctx) {
@@ -1132,7 +1132,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IGeneralService>(system);
+        rb.PushIpcInterface<IGeneralService>(ctx, system);
     }
 };
 

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

@@ -53,7 +53,7 @@ private:
         LOG_WARNING(Service_NIM, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IShopServiceAsync>(system);
+        rb.PushIpcInterface<IShopServiceAsync>(ctx, system);
     }
 };
 
@@ -75,7 +75,7 @@ private:
         LOG_WARNING(Service_NIM, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IShopServiceAccessor>(system);
+        rb.PushIpcInterface<IShopServiceAccessor>(ctx, system);
     }
 };
 
@@ -336,7 +336,7 @@ private:
         LOG_DEBUG(Service_NIM, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IShopServiceAccessServer>(system);
+        rb.PushIpcInterface<IShopServiceAccessServer>(ctx, system);
     }
 
     void IsLargeResourceAvailable(HLERequestContext& ctx) {
@@ -356,7 +356,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IShopServiceAccessServer>(system);
+        rb.PushIpcInterface<IShopServiceAccessServer>(ctx, system);
     }
 };
 
@@ -439,7 +439,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 1};
         rb.Push(ResultSuccess);
-        rb.PushCopyObjects(finished_event->GetReadableEvent());
+        rb.PushCopyObjects(ctx, finished_event->GetReadableEvent());
     }
 
     void GetResult(HLERequestContext& ctx) {
@@ -500,7 +500,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IEnsureNetworkClockAvailabilityService>(system);
+        rb.PushIpcInterface<IEnsureNetworkClockAvailabilityService>(ctx, system);
     }
 
     // TODO(ogniK): Do we need these?

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

@@ -359,8 +359,8 @@ void IReadOnlyApplicationControlDataInterface::ListApplicationTitle(HLERequestCo
 
     IPC::ResponseBuilder rb{ctx, 2, 1, 1};
     rb.Push(ResultSuccess);
-    rb.PushCopyObjects(async_value->ReadableEvent());
-    rb.PushIpcInterface(std::move(async_value));
+    rb.PushCopyObjects(ctx, async_value->ReadableEvent());
+    rb.PushIpcInterface(ctx, std::move(async_value));
 }
 
 Result IReadOnlyApplicationControlDataInterface::GetApplicationControlData3(

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

@@ -199,7 +199,7 @@ void NVDRV::QueryEvent(HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 3, 1};
         rb.Push(ResultSuccess);
         auto& readable_event = event->GetReadableEvent();
-        rb.PushCopyObjects(readable_event);
+        rb.PushCopyObjects(ctx, readable_event);
         rb.PushEnum(NvResult::Success);
     } else {
         LOG_ERROR(Service_NVDRV, "Invalid event request!");

src/core/hle/service/os/process.h

@@ -6,6 +6,8 @@
 
 #pragma once
 
+#include <utility>
+
 #include "common/common_types.h"
 
 namespace Core {
@@ -28,6 +30,15 @@ public:
     inline explicit Process(Core::System& system) noexcept : m_system(system) {}
     inline ~Process() { this->Finalize(); }
 
+    Process(const Process&) = delete;
+    Process& operator=(const Process&) = delete;
+    Process& operator=(Process&&) = delete;
+    inline Process(Process&& other) noexcept
+        : m_system(other.m_system), m_process(std::exchange(other.m_process, nullptr)),
+          m_main_thread_stack_size(std::exchange(other.m_main_thread_stack_size, 0)),
+          m_main_thread_priority(std::exchange(other.m_main_thread_priority, 0)),
+          m_process_started(std::exchange(other.m_process_started, false)) {}
+
     bool Initialize(Loader::AppLoader& loader, Loader::ResultStatus& out_load_result);
     void Finalize();
 

src/core/hle/service/pcv/pcv.cpp

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2018 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -124,7 +127,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IClkrstSession>(system, device_code);
+        rb.PushIpcInterface<IClkrstSession>(ctx, system, device_code);
     }
 };
 

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

@@ -161,7 +161,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 10, 1};
         rb.Push(ResultSuccess);
-        rb.PushCopyObjects(*process);
+        rb.PushCopyObjects(ctx, *process);
         rb.PushRaw(program_location);
         rb.PushRaw(override_status);
     }

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

@@ -147,7 +147,7 @@ void IAlarmService::CreateWakeupAlarm(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<ISteadyClockAlarm>(system, m_alarms, AlarmType::WakeupAlarm);
+    rb.PushIpcInterface<ISteadyClockAlarm>(ctx, system, m_alarms, AlarmType::WakeupAlarm);
 }
 
 void IAlarmService::CreateBackgroundTaskAlarm(HLERequestContext& ctx) {
@@ -155,7 +155,7 @@ void IAlarmService::CreateBackgroundTaskAlarm(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<ISteadyClockAlarm>(system, m_alarms, AlarmType::BackgroundTaskAlarm);
+    rb.PushIpcInterface<ISteadyClockAlarm>(ctx, system, m_alarms, AlarmType::BackgroundTaskAlarm);
 }
 
 ISteadyClockAlarm::ISteadyClockAlarm(Core::System& system_, Alarms& alarms, AlarmType type)
@@ -179,7 +179,7 @@ void ISteadyClockAlarm::GetAlarmEvent(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(ResultSuccess);
-    rb.PushCopyObjects(m_alarm.GetEventHandle());
+    rb.PushCopyObjects(ctx, m_alarm.GetEventHandle());
 }
 
 void ISteadyClockAlarm::Enable(HLERequestContext& ctx) {

src/core/hle/service/ptm/psm.cpp

@@ -67,7 +67,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 1};
         rb.Push(ResultSuccess);
-        rb.PushCopyObjects(state_change_event->GetReadableEvent());
+        rb.PushCopyObjects(ctx, state_change_event->GetReadableEvent());
     }
 
     void UnbindStateChangeEvent(HLERequestContext& ctx) {
@@ -186,7 +186,7 @@ void PSM::OpenSession(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IPsmSession>(system);
+    rb.PushIpcInterface<IPsmSession>(ctx, system);
 }
 
 void PSM::GetBatteryVoltageState(HLERequestContext& ctx) {

src/core/hle/service/ptm/ts.cpp

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
@@ -82,7 +85,7 @@ void TS::OpenSession(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<ISession>(system);
+    rb.PushIpcInterface<ISession>(ctx, system);
 }
 
 } // namespace Service::PTM

src/core/hle/service/sm/sm.cpp

@@ -140,7 +140,7 @@ void SM::GetServiceCmif(HLERequestContext& ctx) {
     if (result == ResultSuccess) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1, IPC::ResponseBuilder::Flags::AlwaysMoveHandles};
         rb.Push(result);
-        rb.PushMoveObjects(client_session);
+        rb.PushMoveObjects(ctx, client_session);
     } else {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(result);
@@ -157,7 +157,7 @@ void SM::GetServiceTipc(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1, IPC::ResponseBuilder::Flags::AlwaysMoveHandles};
     rb.Push(result);
-    rb.PushMoveObjects(result == ResultSuccess ? client_session : nullptr);
+    rb.PushMoveObjects(ctx, result == ResultSuccess ? client_session : nullptr);
 }
 
 static std::string PopServiceName(IPC::RequestParser& rp) {
@@ -230,8 +230,7 @@ void SM::RegisterServiceImpl(HLERequestContext& ctx, std::string name, u32 max_s
               max_session_count, is_light);
 
     Kernel::KServerPort* server_port{};
-    if (const auto result = service_manager.RegisterService(std::addressof(server_port), name,
-                                                            max_session_count, nullptr);
+    if (const auto result = service_manager.RegisterService(std::addressof(server_port), name, max_session_count, nullptr);
         result.IsError()) {
         LOG_ERROR(Service_SM, "failed to register service with error_code={:08X}", result.raw);
         IPC::ResponseBuilder rb{ctx, 2};
@@ -241,7 +240,7 @@ void SM::RegisterServiceImpl(HLERequestContext& ctx, std::string name, u32 max_s
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1, IPC::ResponseBuilder::Flags::AlwaysMoveHandles};
     rb.Push(ResultSuccess);
-    rb.PushMoveObjects(server_port);
+    rb.PushMoveObjects(ctx, server_port);
 }
 
 void SM::UnregisterService(HLERequestContext& ctx) {

src/core/hle/service/sm/sm_controller.cpp

@@ -60,7 +60,7 @@ void Controller::CloneCurrentObject(HLERequestContext& ctx) {
     // We succeeded.
     IPC::ResponseBuilder rb{ctx, 2, 0, 1, IPC::ResponseBuilder::Flags::AlwaysMoveHandles};
     rb.Push(ResultSuccess);
-    rb.PushMoveObjects(session->GetClientSession());
+    rb.PushMoveObjects(ctx, session->GetClientSession());
 }
 
 void Controller::CloneCurrentObjectEx(HLERequestContext& ctx) {

src/core/hle/service/ssl/ssl.cpp

@@ -543,8 +543,7 @@ private:
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(res);
         if (res == ResultSuccess) {
-            rb.PushIpcInterface<ISslConnection>(system, ssl_version, shared_data,
-                                                std::move(backend));
+            rb.PushIpcInterface<ISslConnection>(ctx, system, ssl_version, shared_data, std::move(backend));
         }
     }
 
@@ -624,12 +623,11 @@ private:
         IPC::RequestParser rp{ctx};
         const auto parameters = rp.PopRaw<Parameters>();
 
-        LOG_WARNING(Service_SSL, "(STUBBED) called, api_version={}, pid_placeholder={}",
-                    parameters.ssl_version.api_version, parameters.pid_placeholder);
+        LOG_WARNING(Service_SSL, "(STUBBED) called, api_version={}, pid_placeholder={}", parameters.ssl_version.api_version, parameters.pid_placeholder);
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<ISslContext>(system, parameters.ssl_version);
+        rb.PushIpcInterface<ISslContext>(ctx, system, parameters.ssl_version);
     }
 
     void SetInterfaceVersion(HLERequestContext& ctx) {

src/core/hle/service/usb/usb.cpp

@@ -155,7 +155,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IPdSession>(system);
+        rb.PushIpcInterface<IPdSession>(ctx, system);
     }
 };
 
@@ -199,7 +199,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IPdCradleSession>(system);
+        rb.PushIpcInterface<IPdCradleSession>(ctx, system);
     }
 };
 

src/hid_core/resources/npad/npad.cpp

@@ -50,8 +50,8 @@ NPad::NPad(Core::HID::HIDCore& hid_core_, KernelHelpers::ServiceContext& service
             auto& controller = controller_data[aruid_index][i];
             controller.device = hid_core.GetEmulatedControllerByIndex(i);
             Core::HID::ControllerUpdateCallback engine_callback{
-                .on_change = [this, i](Core::HID::ControllerTriggerType type) {
-                    ControllerUpdate(hid_core.kernel, type, i);
+                .on_change = [this, i, kernel = &hid_core.kernel](Core::HID::ControllerTriggerType type) {
+                    ControllerUpdate(*kernel, type, i);
                 },
                 .is_npad_service = true,
             };

src/shader_recompiler/backend/spirv/emit_spirv.cpp

@@ -332,6 +332,9 @@ void DefineEntryPoint(const IR::Program& program, EmitContext& ctx, Id main) {
 void SetupDenormControl(const Profile& profile, const IR::Program& program, EmitContext& ctx,
                         Id main_func) {
     const Info& info{program.info};
+    if (profile.has_broken_fp16_float_controls && info.uses_fp16) {
+        return;
+    }
     if (info.uses_fp32_denorms_flush && info.uses_fp32_denorms_preserve) {
         LOG_DEBUG(Shader_SPIRV, "Fp32 denorm flush and preserve on the same shader");
     } else if (info.uses_fp32_denorms_flush) {
@@ -432,7 +435,7 @@ void SetupCapabilities(const Profile& profile, const Info& info, EmitContext& ct
     }
     if ((info.uses_subgroup_vote || info.uses_subgroup_invocation_id ||
          info.uses_subgroup_shuffles) &&
-        profile.support_vote) {
+        profile.support_vote && profile.SupportsSubgroupStage(ctx.stage)) {
         ctx.AddCapability(spv::Capability::GroupNonUniformBallot);
         ctx.AddCapability(spv::Capability::GroupNonUniformShuffle);
         if (!profile.warp_size_potentially_larger_than_guest) {

src/shader_recompiler/backend/spirv/emit_spirv_atomic.cpp

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -15,7 +18,7 @@ Id SharedPointer(EmitContext& ctx, Id offset, u32 index_offset = 0) {
     if (index_offset > 0) {
         index = ctx.OpIAdd(ctx.U32[1], index, ctx.Const(index_offset));
     }
-    return ctx.profile.support_explicit_workgroup_layout
+    return ctx.uses_explicit_workgroup_layout
                ? ctx.OpAccessChain(ctx.shared_u32, ctx.shared_memory_u32, ctx.u32_zero_value, index)
                : ctx.OpAccessChain(ctx.shared_u32, ctx.shared_memory_u32, index);
 }
@@ -155,7 +158,7 @@ Id EmitSharedAtomicExchange32(EmitContext& ctx, Id offset, Id value) {
 }
 
 Id EmitSharedAtomicExchange64(EmitContext& ctx, Id offset, Id value) {
-    if (ctx.profile.support_int64_atomics && ctx.profile.support_explicit_workgroup_layout) {
+    if (ctx.profile.support_shared_int64_atomics && ctx.uses_explicit_workgroup_layout) {
         const Id shift_id{ctx.Const(3U)};
         const Id index{ctx.OpShiftRightArithmetic(ctx.U32[1], offset, shift_id)};
         const Id pointer{

src/shader_recompiler/backend/spirv/emit_spirv_memory.cpp

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 

src/shader_recompiler/backend/spirv/emit_spirv_shared_memory.cpp

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -28,7 +31,7 @@ std::pair<Id, Id> ExtractArgs(EmitContext& ctx, Id offset, u32 mask, u32 count)
 } // Anonymous namespace
 
 Id EmitLoadSharedU8(EmitContext& ctx, Id offset) {
-    if (ctx.profile.support_explicit_workgroup_layout) {
+    if (ctx.uses_explicit_workgroup_layout) {
         const Id pointer{
             ctx.OpAccessChain(ctx.shared_u8, ctx.shared_memory_u8, ctx.u32_zero_value, offset)};
         return ctx.OpUConvert(ctx.U32[1], ctx.OpLoad(ctx.U8, pointer));
@@ -39,7 +42,7 @@ Id EmitLoadSharedU8(EmitContext& ctx, Id offset) {
 }
 
 Id EmitLoadSharedS8(EmitContext& ctx, Id offset) {
-    if (ctx.profile.support_explicit_workgroup_layout) {
+    if (ctx.uses_explicit_workgroup_layout) {
         const Id pointer{
             ctx.OpAccessChain(ctx.shared_u8, ctx.shared_memory_u8, ctx.u32_zero_value, offset)};
         return ctx.OpSConvert(ctx.U32[1], ctx.OpLoad(ctx.U8, pointer));
@@ -50,7 +53,7 @@ Id EmitLoadSharedS8(EmitContext& ctx, Id offset) {
 }
 
 Id EmitLoadSharedU16(EmitContext& ctx, Id offset) {
-    if (ctx.profile.support_explicit_workgroup_layout) {
+    if (ctx.uses_explicit_workgroup_layout) {
         const Id pointer{Pointer(ctx, ctx.shared_u16, ctx.shared_memory_u16, offset, 1)};
         return ctx.OpUConvert(ctx.U32[1], ctx.OpLoad(ctx.U16, pointer));
     } else {
@@ -60,7 +63,7 @@ Id EmitLoadSharedU16(EmitContext& ctx, Id offset) {
 }
 
 Id EmitLoadSharedS16(EmitContext& ctx, Id offset) {
-    if (ctx.profile.support_explicit_workgroup_layout) {
+    if (ctx.uses_explicit_workgroup_layout) {
         const Id pointer{Pointer(ctx, ctx.shared_u16, ctx.shared_memory_u16, offset, 1)};
         return ctx.OpSConvert(ctx.U32[1], ctx.OpLoad(ctx.U16, pointer));
     } else {
@@ -70,7 +73,7 @@ Id EmitLoadSharedS16(EmitContext& ctx, Id offset) {
 }
 
 Id EmitLoadSharedU32(EmitContext& ctx, Id offset) {
-    if (ctx.profile.support_explicit_workgroup_layout) {
+    if (ctx.uses_explicit_workgroup_layout) {
         const Id pointer{Pointer(ctx, ctx.shared_u32, ctx.shared_memory_u32, offset, 2)};
         return ctx.OpLoad(ctx.U32[1], pointer);
     } else {
@@ -79,7 +82,7 @@ Id EmitLoadSharedU32(EmitContext& ctx, Id offset) {
 }
 
 Id EmitLoadSharedU64(EmitContext& ctx, Id offset) {
-    if (ctx.profile.support_explicit_workgroup_layout) {
+    if (ctx.uses_explicit_workgroup_layout) {
         const Id pointer{Pointer(ctx, ctx.shared_u32x2, ctx.shared_memory_u32x2, offset, 3)};
         return ctx.OpLoad(ctx.U32[2], pointer);
     } else {
@@ -94,7 +97,7 @@ Id EmitLoadSharedU64(EmitContext& ctx, Id offset) {
 }
 
 Id EmitLoadSharedU128(EmitContext& ctx, Id offset) {
-    if (ctx.profile.support_explicit_workgroup_layout) {
+    if (ctx.uses_explicit_workgroup_layout) {
         const Id pointer{Pointer(ctx, ctx.shared_u32x4, ctx.shared_memory_u32x4, offset, 4)};
         return ctx.OpLoad(ctx.U32[4], pointer);
     }
@@ -110,7 +113,7 @@ Id EmitLoadSharedU128(EmitContext& ctx, Id offset) {
 }
 
 void EmitWriteSharedU8(EmitContext& ctx, Id offset, Id value) {
-    if (ctx.profile.support_explicit_workgroup_layout) {
+    if (ctx.uses_explicit_workgroup_layout) {
         const Id pointer{
             ctx.OpAccessChain(ctx.shared_u8, ctx.shared_memory_u8, ctx.u32_zero_value, offset)};
         ctx.OpStore(pointer, ctx.OpUConvert(ctx.U8, value));
@@ -120,7 +123,7 @@ void EmitWriteSharedU8(EmitContext& ctx, Id offset, Id value) {
 }
 
 void EmitWriteSharedU16(EmitContext& ctx, Id offset, Id value) {
-    if (ctx.profile.support_explicit_workgroup_layout) {
+    if (ctx.uses_explicit_workgroup_layout) {
         const Id pointer{Pointer(ctx, ctx.shared_u16, ctx.shared_memory_u16, offset, 1)};
         ctx.OpStore(pointer, ctx.OpUConvert(ctx.U16, value));
     } else {
@@ -130,7 +133,7 @@ void EmitWriteSharedU16(EmitContext& ctx, Id offset, Id value) {
 
 void EmitWriteSharedU32(EmitContext& ctx, Id offset, Id value) {
     Id pointer{};
-    if (ctx.profile.support_explicit_workgroup_layout) {
+    if (ctx.uses_explicit_workgroup_layout) {
         pointer = Pointer(ctx, ctx.shared_u32, ctx.shared_memory_u32, offset, 2);
     } else {
         const Id shift{ctx.Const(2U)};
@@ -141,7 +144,7 @@ void EmitWriteSharedU32(EmitContext& ctx, Id offset, Id value) {
 }
 
 void EmitWriteSharedU64(EmitContext& ctx, Id offset, Id value) {
-    if (ctx.profile.support_explicit_workgroup_layout) {
+    if (ctx.uses_explicit_workgroup_layout) {
         const Id pointer{Pointer(ctx, ctx.shared_u32x2, ctx.shared_memory_u32x2, offset, 3)};
         ctx.OpStore(pointer, value);
         return;
@@ -156,7 +159,7 @@ void EmitWriteSharedU64(EmitContext& ctx, Id offset, Id value) {
 }
 
 void EmitWriteSharedU128(EmitContext& ctx, Id offset, Id value) {
-    if (ctx.profile.support_explicit_workgroup_layout) {
+    if (ctx.uses_explicit_workgroup_layout) {
         const Id pointer{Pointer(ctx, ctx.shared_u32x4, ctx.shared_memory_u32x4, offset, 4)};
         ctx.OpStore(pointer, value);
         return;

src/shader_recompiler/backend/spirv/emit_spirv_warp.cpp

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -10,7 +13,14 @@ Id SubgroupScope(EmitContext& ctx) {
     return ctx.Const(static_cast<u32>(spv::Scope::Subgroup));
 }
 
+bool StageSupportsSubgroups(EmitContext& ctx) {
+    return ctx.profile.SupportsSubgroupStage(ctx.stage);
+}
+
 Id GetThreadId(EmitContext& ctx) {
+    if (!StageSupportsSubgroups(ctx)) {
+        return ctx.u32_zero_value;
+    }
     return ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id);
 }
 
@@ -68,6 +78,9 @@ Id GetMaxThreadId(EmitContext& ctx, Id thread_id, Id clamp, Id segmentation_mask
 }
 
 Id SelectValue(EmitContext& ctx, Id in_range, Id value, Id src_thread_id) {
+    if (!StageSupportsSubgroups(ctx)) {
+        return value;
+    }
     return ctx.OpSelect(
         ctx.U32[1], in_range,
         ctx.OpGroupNonUniformShuffle(ctx.U32[1], SubgroupScope(ctx), value, src_thread_id), value);
@@ -89,6 +102,9 @@ Id EmitLaneId(EmitContext& ctx) {
 }
 
 Id EmitVoteAll(EmitContext& ctx, Id pred) {
+    if (!StageSupportsSubgroups(ctx)) {
+        return pred;
+    }
     if (!ctx.profile.warp_size_potentially_larger_than_guest) {
         return ctx.OpGroupNonUniformAll(ctx.U1, SubgroupScope(ctx), pred);
     }
@@ -102,6 +118,9 @@ Id EmitVoteAll(EmitContext& ctx, Id pred) {
 }
 
 Id EmitVoteAny(EmitContext& ctx, Id pred) {
+    if (!StageSupportsSubgroups(ctx)) {
+        return pred;
+    }
     if (!ctx.profile.warp_size_potentially_larger_than_guest) {
         return ctx.OpGroupNonUniformAny(ctx.U1, SubgroupScope(ctx), pred);
     }
@@ -115,6 +134,9 @@ Id EmitVoteAny(EmitContext& ctx, Id pred) {
 }
 
 Id EmitVoteEqual(EmitContext& ctx, Id pred) {
+    if (!StageSupportsSubgroups(ctx)) {
+        return ctx.true_value;
+    }
     if (!ctx.profile.warp_size_potentially_larger_than_guest) {
         return ctx.OpGroupNonUniformAllEqual(ctx.U1, SubgroupScope(ctx), pred);
     }
@@ -129,6 +151,9 @@ Id EmitVoteEqual(EmitContext& ctx, Id pred) {
 }
 
 Id EmitSubgroupBallot(EmitContext& ctx, Id pred) {
+    if (!StageSupportsSubgroups(ctx)) {
+        return ctx.OpSelect(ctx.U32[1], pred, ctx.Const(1u), ctx.u32_zero_value);
+    }
     const Id ballot{ctx.OpGroupNonUniformBallot(ctx.U32[4], SubgroupScope(ctx), pred)};
     if (!ctx.profile.warp_size_potentially_larger_than_guest) {
         return ctx.OpCompositeExtract(ctx.U32[1], ballot, 0U);
@@ -137,22 +162,37 @@ Id EmitSubgroupBallot(EmitContext& ctx, Id pred) {
 }
 
 Id EmitSubgroupEqMask(EmitContext& ctx) {
+    if (!StageSupportsSubgroups(ctx)) {
+        return ctx.Const(1u);
+    }
     return LoadMask(ctx, ctx.subgroup_mask_eq);
 }
 
 Id EmitSubgroupLtMask(EmitContext& ctx) {
+    if (!StageSupportsSubgroups(ctx)) {
+        return ctx.u32_zero_value;
+    }
     return LoadMask(ctx, ctx.subgroup_mask_lt);
 }
 
 Id EmitSubgroupLeMask(EmitContext& ctx) {
+    if (!StageSupportsSubgroups(ctx)) {
+        return ctx.Const(1u);
+    }
     return LoadMask(ctx, ctx.subgroup_mask_le);
 }
 
 Id EmitSubgroupGtMask(EmitContext& ctx) {
+    if (!StageSupportsSubgroups(ctx)) {
+        return ctx.u32_zero_value;
+    }
     return LoadMask(ctx, ctx.subgroup_mask_gt);
 }
 
 Id EmitSubgroupGeMask(EmitContext& ctx) {
+    if (!StageSupportsSubgroups(ctx)) {
+        return ctx.Const(1u);
+    }
     return LoadMask(ctx, ctx.subgroup_mask_ge);
 }
 
@@ -222,7 +262,7 @@ Id EmitShuffleButterfly(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id
 
 Id EmitFSwizzleAdd(EmitContext& ctx, Id op_a, Id op_b, Id swizzle) {
     const Id three{ctx.Const(3U)};
-    Id mask{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
+    Id mask{GetThreadId(ctx)};
     mask = ctx.OpBitwiseAnd(ctx.U32[1], mask, three);
     mask = ctx.OpShiftLeftLogical(ctx.U32[1], mask, ctx.Const(1U));
     mask = ctx.OpShiftRightLogical(ctx.U32[1], swizzle, mask);

src/shader_recompiler/backend/spirv/spirv_emit_context.cpp

@@ -371,7 +371,7 @@ Id CasFunction(EmitContext& ctx, Operation operation, Id value_type) {
 Id CasLoop(EmitContext& ctx, Operation operation, Id array_pointer, Id element_pointer,
            Id value_type, Id memory_type, spv::Scope scope) {
     const bool is_shared{scope == spv::Scope::Workgroup};
-    const bool is_struct{!is_shared || ctx.profile.support_explicit_workgroup_layout};
+    const bool is_struct{!is_shared || ctx.uses_explicit_workgroup_layout};
     const Id cas_func{CasFunction(ctx, operation, value_type)};
     const Id zero{ctx.u32_zero_value};
     const Id scope_id{ctx.Const(static_cast<u32>(scope))};
@@ -620,7 +620,11 @@ void EmitContext::DefineSharedMemory(const IR::Program& program) {
 
         return std::make_tuple(variable, element_pointer, pointer);
     }};
-    if (profile.support_explicit_workgroup_layout) {
+    uses_explicit_workgroup_layout =
+        profile.support_explicit_workgroup_layout &&
+        (!program.info.uses_int8 || profile.support_workgroup_layout_8bit_access) &&
+        (!program.info.uses_int16 || profile.support_workgroup_layout_16bit_access);
+    if (uses_explicit_workgroup_layout) {
         AddExtension("SPV_KHR_workgroup_memory_explicit_layout");
         AddCapability(spv::Capability::WorkgroupMemoryExplicitLayoutKHR);
         if (program.info.uses_int8) {
@@ -938,6 +942,10 @@ void EmitContext::DefineGlobalMemoryFunctions(const Info& info) {
     if (!info.uses_global_memory || !profile.support_int64) {
         return;
     }
+    if (!profile.support_descriptor_aliasing) {
+        DefineGlobalMemoryFunctionsU32Fallback(info);
+        return;
+    }
     using DefPtr = Id StorageDefinitions::*;
     const Id zero{u32_zero_value};
     const auto define_body{[&](DefPtr ssbo_member, Id addr, Id element_pointer, u32 shift,
@@ -1015,6 +1023,111 @@ void EmitContext::DefineGlobalMemoryFunctions(const Info& info) {
         define(&StorageDefinitions::U32x4, storage_types.U32x4, U32[4], sizeof(u32[4]));
 }
 
+void EmitContext::DefineGlobalMemoryFunctionsU32Fallback(const Info& info) {
+    const Id zero{u32_zero_value};
+    const auto define_body{[&](Id addr, u32 num_words, auto&& callback) {
+        AddLabel();
+        const size_t num_buffers{info.storage_buffers_descriptors.size()};
+        for (size_t index = 0; index < num_buffers; ++index) {
+            if (!info.nvn_buffer_used[index]) {
+                continue;
+            }
+            const auto& ssbo{info.storage_buffers_descriptors[index]};
+            const u32 addr_word{ssbo.cbuf_offset / 4};
+            const u32 addr_lo_comp{addr_word % 4};
+            const Id cbuf{cbufs[ssbo.cbuf_index].U32x4};
+            const Id addr_vec_pointer{
+                OpAccessChain(uniform_types.U32x4, cbuf, zero, Const(addr_word / 4))};
+            const Id addr_vec{OpLoad(U32[4], addr_vec_pointer)};
+            const Id addr_lo{OpCompositeExtract(U32[1], addr_vec, addr_lo_comp)};
+            const Id addr_hi{OpCompositeExtract(U32[1], addr_vec, addr_lo_comp + 1U)};
+            const Id unaligned_addr{
+                OpBitcast(U64, OpCompositeConstruct(U32[2], addr_lo, addr_hi))};
+
+            const u64 ssbo_align_mask{~(profile.min_ssbo_alignment - 1U)};
+            const Id ssbo_addr{OpBitwiseAnd(U64, unaligned_addr, Constant(U64, ssbo_align_mask))};
+
+            const u32 size_word{addr_word + 2};
+            Id size_vec{addr_vec};
+            if (size_word / 4 != addr_word / 4) {
+                const Id size_vec_pointer{
+                    OpAccessChain(uniform_types.U32x4, cbuf, zero, Const(size_word / 4))};
+                size_vec = OpLoad(U32[4], size_vec_pointer);
+            }
+            const Id ssbo_size{
+                OpUConvert(U64, OpCompositeExtract(U32[1], size_vec, size_word % 4))};
+            const Id ssbo_end{OpIAdd(U64, ssbo_addr, ssbo_size)};
+            const Id cond{OpLogicalAnd(U1, OpUGreaterThanEqual(U1, addr, ssbo_addr),
+                                       OpULessThan(U1, addr, ssbo_end))};
+            const Id then_label{OpLabel()};
+            const Id else_label{OpLabel()};
+            OpSelectionMerge(else_label, spv::SelectionControlMask::MaskNone);
+            OpBranchConditional(cond, then_label, else_label);
+            AddLabel(then_label);
+            const Id ssbo_id{ssbos[index].U32};
+            const Id ssbo_offset{OpUConvert(U32[1], OpISub(U64, addr, ssbo_addr))};
+            const Id base_word{OpShiftRightLogical(U32[1], ssbo_offset, Const(2U))};
+            std::array<Id, 4> word_pointers{};
+            for (u32 word = 0; word < num_words; ++word) {
+                const Id word_index{word == 0 ? base_word
+                                              : OpIAdd(U32[1], base_word, Const(word))};
+                word_pointers[word] =
+                    OpAccessChain(storage_types.U32.element, ssbo_id, zero, word_index);
+            }
+            callback(word_pointers);
+            AddLabel(else_label);
+        }
+    }};
+    const auto define_load{[&](Id type, u32 num_words) {
+        const Id function_type{TypeFunction(type, U64)};
+        const Id func_id{OpFunction(type, spv::FunctionControlMask::MaskNone, function_type)};
+        const Id addr{OpFunctionParameter(U64)};
+        define_body(addr, num_words, [&](const std::array<Id, 4>& pointers) {
+            std::array<Id, 4> words{};
+            for (u32 word = 0; word < num_words; ++word) {
+                words[word] = OpLoad(U32[1], pointers[word]);
+            }
+            switch (num_words) {
+            case 1:
+                OpReturnValue(words[0]);
+                break;
+            case 2:
+                OpReturnValue(OpCompositeConstruct(type, words[0], words[1]));
+                break;
+            default:
+                OpReturnValue(
+                    OpCompositeConstruct(type, words[0], words[1], words[2], words[3]));
+                break;
+            }
+        });
+        OpReturnValue(ConstantNull(type));
+        OpFunctionEnd();
+        return func_id;
+    }};
+    const auto define_write{[&](Id type, u32 num_words) {
+        const Id function_type{TypeFunction(void_id, U64, type)};
+        const Id func_id{OpFunction(void_id, spv::FunctionControlMask::MaskNone, function_type)};
+        const Id addr{OpFunctionParameter(U64)};
+        const Id data{OpFunctionParameter(type)};
+        define_body(addr, num_words, [&](const std::array<Id, 4>& pointers) {
+            for (u32 word = 0; word < num_words; ++word) {
+                const Id value{num_words == 1 ? data : OpCompositeExtract(U32[1], data, word)};
+                OpStore(pointers[word], value);
+            }
+            OpReturn();
+        });
+        OpReturn();
+        OpFunctionEnd();
+        return func_id;
+    }};
+    load_global_func_u32 = define_load(U32[1], 1);
+    load_global_func_u32x2 = define_load(U32[2], 2);
+    load_global_func_u32x4 = define_load(U32[4], 4);
+    write_global_func_u32 = define_write(U32[1], 1);
+    write_global_func_u32x2 = define_write(U32[2], 2);
+    write_global_func_u32x4 = define_write(U32[4], 4);
+}
+
 void EmitContext::DefineRescalingInput(const Info& info) {
     if (!info.uses_rescaling_uniform) {
         return;
@@ -1438,7 +1551,7 @@ void EmitContext::DefineInputs(const IR::Program& program) {
     if (info.uses_is_helper_invocation) {
         is_helper_invocation = DefineInput(*this, U1, false, spv::BuiltIn::HelperInvocation);
     }
-    if (info.uses_subgroup_mask) {
+    if (info.uses_subgroup_mask && profile.SupportsSubgroupStage(stage)) {
         subgroup_mask_eq = DefineInput(*this, U32[4], false, spv::BuiltIn::SubgroupEqMaskKHR);
         subgroup_mask_lt = DefineInput(*this, U32[4], false, spv::BuiltIn::SubgroupLtMaskKHR);
         subgroup_mask_le = DefineInput(*this, U32[4], false, spv::BuiltIn::SubgroupLeMaskKHR);
@@ -1452,9 +1565,10 @@ void EmitContext::DefineInputs(const IR::Program& program) {
             Decorate(subgroup_mask_ge, spv::Decoration::Flat);
         }
     }
-    if (info.uses_fswzadd || info.uses_subgroup_invocation_id || info.uses_subgroup_shuffles ||
-        (profile.warp_size_potentially_larger_than_guest &&
-         (info.uses_subgroup_vote || info.uses_subgroup_mask))) {
+    if ((info.uses_fswzadd || info.uses_subgroup_invocation_id || info.uses_subgroup_shuffles ||
+         (profile.warp_size_potentially_larger_than_guest &&
+          (info.uses_subgroup_vote || info.uses_subgroup_mask))) &&
+        profile.SupportsSubgroupStage(stage)) {
         AddCapability(spv::Capability::GroupNonUniform);
         subgroup_local_invocation_id =
             DefineInput(*this, U32[1], false, spv::BuiltIn::SubgroupLocalInvocationId);

src/shader_recompiler/backend/spirv/spirv_emit_context.h

@@ -310,6 +310,10 @@ public:
 
     Id local_memory{};
 
+    /// True when this shader's shared memory uses SPV_KHR_workgroup_memory_explicit_layout.
+    /// False when the host lacks the extension or a width this shader accesses natively.
+    bool uses_explicit_workgroup_layout{};
+
     Id shared_memory_u8{};
     Id shared_memory_u16{};
     Id shared_memory_u32{};
@@ -387,6 +391,7 @@ private:
     void DefineAttributeMemAccess(const Info& info);
     void DefineWriteStorageCasLoopFunction(const Info& info);
     void DefineGlobalMemoryFunctions(const Info& info);
+    void DefineGlobalMemoryFunctionsU32Fallback(const Info& info);
     void DefineRescalingInput(const Info& info);
     void DefineRescalingInputPushConstant();
     void DefineRescalingInputUniformConstant();

src/shader_recompiler/frontend/maxwell/translate_program.cpp

@@ -236,8 +236,11 @@ void LowerGeometryPassthrough(const IR::Program& program, const HostTranslateInf
 
 IR::Program TranslateProgram(ObjectPool<IR::Inst>& inst_pool, ObjectPool<IR::Block>& block_pool,
                              Environment& env, Flow::CFG& cfg, const HostTranslateInfo& host_info) {
+    HostTranslateInfo normalized_host_info{host_info};
+    normalized_host_info.ApplyDescriptorLimitPolicy();
+
     IR::Program program;
-    program.syntax_list = BuildASL(inst_pool, block_pool, env, cfg, host_info);
+    program.syntax_list = BuildASL(inst_pool, block_pool, env, cfg, normalized_host_info);
     program.blocks = GenerateBlocks(program.syntax_list);
     program.post_order_blocks = PostOrder(program.syntax_list.front());
     program.stage = env.ShaderStage();
@@ -260,9 +263,9 @@ IR::Program TranslateProgram(ObjectPool<IR::Inst>& inst_pool, ObjectPool<IR::Blo
                 program.info.passthrough.mask[i] = ((mask[i / 32] >> (i % 32)) & 1) == 0;
             }
 
-            if (!host_info.support_geometry_shader_passthrough) {
+            if (!normalized_host_info.support_geometry_shader_passthrough) {
                 program.output_vertices = GetOutputTopologyVertices(program.output_topology);
-                LowerGeometryPassthrough(program, host_info);
+                LowerGeometryPassthrough(program, normalized_host_info);
             }
         }
         break;
@@ -277,16 +280,16 @@ IR::Program TranslateProgram(ObjectPool<IR::Inst>& inst_pool, ObjectPool<IR::Blo
     RemoveUnreachableBlocks(program);
 
     // Replace instructions before the SSA rewrite
-    if (!host_info.support_float64) {
+    if (!normalized_host_info.support_float64) {
         Optimization::LowerFp64ToFp32(program);
     }
-    if (!host_info.support_float16) {
+    if (!normalized_host_info.support_float16) {
         Optimization::LowerFp16ToFp32(program);
     }
-    if (!host_info.support_int64) {
+    if (!normalized_host_info.support_int64) {
         Optimization::LowerInt64ToInt32(program);
     }
-    if (!host_info.support_conditional_barrier) {
+    if (!normalized_host_info.support_conditional_barrier) {
         Optimization::ConditionalBarrierPass(program);
     }
     Optimization::SsaRewritePass(program);
@@ -295,8 +298,8 @@ IR::Program TranslateProgram(ObjectPool<IR::Inst>& inst_pool, ObjectPool<IR::Blo
 
     Optimization::PositionPass(env, program);
 
-    Optimization::GlobalMemoryToStorageBufferPass(program, host_info);
-    Optimization::TexturePass(env, program, host_info);
+    Optimization::GlobalMemoryToStorageBufferPass(program, normalized_host_info);
+    Optimization::TexturePass(env, program, normalized_host_info);
 
     if (Settings::values.resolution_info.active || Settings::values.rescale_hack.GetValue()) {
         Optimization::RescalingPass(program);
@@ -306,7 +309,7 @@ IR::Program TranslateProgram(ObjectPool<IR::Inst>& inst_pool, ObjectPool<IR::Blo
         Optimization::VerificationPass(program);
     }
     Optimization::CollectShaderInfoPass(env, program);
-    Optimization::LayerPass(program, host_info);
+    Optimization::LayerPass(program, normalized_host_info);
     Optimization::VendorWorkaroundPass(program);
 
     CollectInterpolationInfo(env, program);

src/shader_recompiler/host_translate_info.h

@@ -6,6 +6,8 @@
 
 #pragma once
 
+#include "common/common_types.h"
+
 namespace Shader {
 
 // Try to keep entries here to a minimum
@@ -13,20 +15,52 @@ namespace Shader {
 
 /// Misc information about the host
 struct HostTranslateInfo {
+    static constexpr u32 DEFAULT_DESCRIPTOR_LIMIT = 1024;
+
+    u64 min_ssbo_alignment{};            ///< Minimum alignment supported by the device for SSBOs
+    u32 max_per_stage_descriptor_sampled_images{}; ///< maximum sampled descriptors per stage
+    u32 max_per_stage_resources{};                 ///< maximum resources per stage
+    u32 max_descriptor_set_samplers{};
+    u32 max_descriptor_set_uniform_buffers{};
+    u32 max_descriptor_set_uniform_buffers_dynamic{};
+    u32 max_descriptor_set_storage_buffers{};
+    u32 max_descriptor_set_storage_buffers_dynamic{};
+    u32 max_descriptor_set_sampled_images{};
+    u32 max_descriptor_set_storage_images{};
+    u32 max_descriptor_set_input_attachements{};
     bool support_float64{};      ///< True when the device supports 64-bit floats
     bool support_float16{};      ///< True when the device supports 16-bit floats
     bool support_int64{};        ///< True when the device supports 64-bit integers
     bool needs_demote_reorder{}; ///< True when the device needs DemoteToHelperInvocation reordered
     bool support_snorm_render_buffer{};  ///< True when the device supports SNORM render buffers
     bool support_viewport_index_layer{}; ///< True when the device supports gl_Layer in VS
-    u32 min_ssbo_alignment{};            ///< Minimum alignment supported by the device for SSBOs
-    u32 max_per_stage_descriptor_sampled_images{1024}; ///< maximum sampled descriptors per stage
-    u32 max_per_stage_resources{4096};                 ///< maximum resources per stage
-    u32 max_descriptor_set_sampled_images{1024};        ///< maximum sampled descriptors per set
     bool support_geometry_shader_passthrough{}; ///< True when the device supports geometry
                                                 ///< passthrough shaders
     bool support_conditional_barrier{}; ///< True when the device supports barriers in conditional
                                         ///< control flow
+
+    void ApplyDescriptorLimitPolicy() noexcept {
+        if (min_ssbo_alignment == 0) {
+            min_ssbo_alignment = 1;
+        }
+        ApplyDescriptorLimitFallback(max_per_stage_descriptor_sampled_images);
+        ApplyDescriptorLimitFallback(max_per_stage_resources);
+        ApplyDescriptorLimitFallback(max_descriptor_set_samplers);
+        ApplyDescriptorLimitFallback(max_descriptor_set_uniform_buffers);
+        ApplyDescriptorLimitFallback(max_descriptor_set_uniform_buffers_dynamic);
+        ApplyDescriptorLimitFallback(max_descriptor_set_storage_buffers);
+        ApplyDescriptorLimitFallback(max_descriptor_set_storage_buffers_dynamic);
+        ApplyDescriptorLimitFallback(max_descriptor_set_sampled_images);
+        ApplyDescriptorLimitFallback(max_descriptor_set_storage_images);
+        ApplyDescriptorLimitFallback(max_descriptor_set_input_attachements);
+    }
+
+private:
+    static void ApplyDescriptorLimitFallback(u32& limit) noexcept {
+        if (limit == 0) {
+            limit = DEFAULT_DESCRIPTOR_LIMIT;
+        }
+    }
 };
 
 } // namespace Shader

src/shader_recompiler/ir_opt/global_memory_to_storage_buffer_pass.cpp

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -545,7 +548,7 @@ void GlobalMemoryToStorageBufferPass(IR::Program& program, const HostTranslateIn
         IR::Block* const block{storage_inst.block};
         IR::Inst* const inst{storage_inst.inst};
         const IR::U32 offset{
-            StorageOffset(*block, *inst, storage_buffer, host_info.min_ssbo_alignment)};
+            StorageOffset(*block, *inst, storage_buffer, u32(host_info.min_ssbo_alignment))};
         Replace(*block, *inst, index, offset);
     }
 }

src/shader_recompiler/ir_opt/texture_pass.cpp

@@ -32,71 +32,62 @@ struct TextureInst {
 using TextureInstVector = boost::container::small_vector<TextureInst, 24>;
 
 constexpr u32 DESCRIPTOR_SIZE = 8;
-constexpr u32 DESCRIPTOR_SIZE_SHIFT = static_cast<u32>(std::countr_zero(DESCRIPTOR_SIZE));
-constexpr u32 DYNAMIC_DESCRIPTOR_CBUF_BYTES = 16 * 1024;
-constexpr u32 MAX_DYNAMIC_DESCRIPTOR_COUNT = 1024;
+constexpr u32 DESCRIPTOR_SIZE_SHIFT = u32(std::countr_zero(DESCRIPTOR_SIZE));
+constexpr u32 DESCRIPTOR_MAX_COUNT = 1024;
 
 u32 DynamicDescriptorSizeShift(const IR::U32& dynamic_offset) {
-    const IR::Inst* const inst{dynamic_offset.InstRecursive()};
-    if (!inst || inst->GetOpcode() != IR::Opcode::ShiftLeftLogical32) {
+    const IR::Inst* const inst = dynamic_offset.InstRecursive();
+    if (!inst || inst->GetOpcode() != IR::Opcode::ShiftLeftLogical32)
         return DESCRIPTOR_SIZE_SHIFT;
-    }
-    const IR::Value shift{inst->Arg(1)};
-    if (!shift.IsImmediate()) {
+    const IR::Value shift = inst->Arg(1);
+    if (!shift.IsImmediate())
         return DESCRIPTOR_SIZE_SHIFT;
-    }
-    const u32 size_shift{shift.U32()};
-    return size_shift >= DESCRIPTOR_SIZE_SHIFT && size_shift < 31 ? size_shift
-                                                                  : DESCRIPTOR_SIZE_SHIFT;
+    const u32 size_shift = shift.U32();
+    return size_shift >= DESCRIPTOR_SIZE_SHIFT && size_shift < 31 ? size_shift : DESCRIPTOR_SIZE_SHIFT;
 }
 
-u32 DynamicDescriptorCount(u32 base_offset, u32 size_shift) {
-    if (size_shift >= 31 || base_offset >= DYNAMIC_DESCRIPTOR_CBUF_BYTES) {
+u32 DynamicDescriptorCount(u32 base_offset, u32 size_shift, u32 max_descriptors) {
+    auto const descriptor_limit = (std::max)(1U, max_descriptors);
+    auto const max_cbuf_bytes = 16 * descriptor_limit;
+    if (size_shift >= 31 || base_offset >= max_cbuf_bytes)
         return 1;
-    }
-    const u32 stride{1U << size_shift};
-    const u32 available{DYNAMIC_DESCRIPTOR_CBUF_BYTES - base_offset};
-    if (available < DESCRIPTOR_SIZE) {
+    auto const stride = 1U << size_shift;
+    auto const available = max_cbuf_bytes - base_offset;
+    if (available < DESCRIPTOR_SIZE)
         return 1;
-    }
-    const u32 available_count{1U + (available - DESCRIPTOR_SIZE) / stride};
-    return std::min(MAX_DYNAMIC_DESCRIPTOR_COUNT, available_count);
+    auto const available_count = 1U + (available - DESCRIPTOR_SIZE) / stride;
+    return std::min(descriptor_limit, available_count);
 }
 
 u32 SaturatingSub(u32 lhs, u32 rhs) {
     return lhs > rhs ? lhs - rhs : 0;
 }
 
-template <typename Descriptors>
-u32 StaticDescriptorCount(const Descriptors& descriptors) {
-    u32 count{};
-    for (const auto& desc : descriptors) {
-        if (desc.count <= 1) {
-            count += desc.count;
-        }
-    }
-    return count;
+template <typename T>
+[[nodiscard]] u32 StaticDescriptorCount(T const& descriptors) noexcept {
+    return std::accumulate(descriptors.cbegin(), descriptors.cend(), 0U, [](auto const& acc, auto const& e) {
+        return acc + (e.count <= 1 ? e.count : 0);
+    });
 }
 
-u32 DynamicSampledTextureCap(const Info& info, const HostTranslateInfo& host_info,
-                             u32 dynamic_arrays) {
-    if (dynamic_arrays == 0) {
-        return MAX_DYNAMIC_DESCRIPTOR_COUNT;
+u32 DynamicSampledTextureCap(const Info& info, const HostTranslateInfo& host_info, u32 dynamic_arrays) {
+    auto const sampled_limit = (std::max)(1U, std::min(host_info.max_per_stage_descriptor_sampled_images,
+                                                       host_info.max_descriptor_set_sampled_images));
+    auto const resource_limit = (std::max)(1U, host_info.max_per_stage_resources);
+    if (dynamic_arrays > 0) {
+        auto const sampled_static_count = StaticDescriptorCount(info.texture_buffer_descriptors) + StaticDescriptorCount(info.texture_descriptors);
+        auto const resource_static_count =
+            NumDescriptors(info.constant_buffer_descriptors)
+            + NumDescriptors(info.storage_buffers_descriptors)
+            + sampled_static_count + NumDescriptors(info.image_buffer_descriptors)
+            + NumDescriptors(info.image_descriptors);
+        auto const sampled_budget = SaturatingSub(sampled_limit, sampled_static_count);
+        auto const resource_budget = SaturatingSub(resource_limit, resource_static_count);
+        auto const sampled_cap = sampled_budget / dynamic_arrays;
+        auto const resource_cap = resource_budget / dynamic_arrays;
+        return (std::max)(1U, (std::min)(sampled_cap, resource_cap));
     }
-    const u32 sampled_static_count{StaticDescriptorCount(info.texture_buffer_descriptors) +
-                                   StaticDescriptorCount(info.texture_descriptors)};
-    const u32 resource_static_count{
-        NumDescriptors(info.constant_buffer_descriptors) +
-        NumDescriptors(info.storage_buffers_descriptors) + sampled_static_count +
-        NumDescriptors(info.image_buffer_descriptors) + NumDescriptors(info.image_descriptors)};
-    const u32 sampled_limit{std::min(host_info.max_per_stage_descriptor_sampled_images,
-                                     host_info.max_descriptor_set_sampled_images)};
-    const u32 sampled_budget{SaturatingSub(sampled_limit, sampled_static_count)};
-    const u32 resource_budget{SaturatingSub(host_info.max_per_stage_resources,
-                                            resource_static_count)};
-    const u32 sampled_cap{sampled_budget / dynamic_arrays};
-    const u32 resource_cap{resource_budget / dynamic_arrays};
-    return std::max(1U, std::min({MAX_DYNAMIC_DESCRIPTOR_COUNT, sampled_cap, resource_cap}));
+    return (std::min)({DESCRIPTOR_MAX_COUNT, sampled_limit, resource_limit});
 }
 
 IR::Opcode IndexedInstruction(const IR::Inst& inst) {
@@ -304,21 +295,23 @@ static inline bool IsTexturePixelFormatIntegerCached(Environment& env,
 }
 
 
-std::optional<ConstBufferAddr> Track(const IR::Value& value, Environment& env);
-static inline std::optional<ConstBufferAddr> TrackCached(const IR::Value& v, Environment& env) {
+std::optional<ConstBufferAddr> Track(const IR::Value& value, Environment& env, const HostTranslateInfo& host_info);
+static inline std::optional<ConstBufferAddr> TrackCached(const IR::Value& v, Environment& env, const HostTranslateInfo& host_info) {
     if (const IR::Inst* key = v.InstRecursive()) {
         if (auto it = env.track_cache.find(key); it != env.track_cache.end()) return it->second;
-        auto found = Track(v, env);
+        auto found = Track(v, env, host_info);
         if (found) env.track_cache.emplace(key, *found);
         return found;
     }
-    return Track(v, env);
+    return Track(v, env, host_info);
 }
 
-std::optional<ConstBufferAddr> TryGetConstBuffer(const IR::Inst* inst, Environment& env);
+std::optional<ConstBufferAddr> TryGetConstBuffer(const IR::Inst* inst, Environment& env, const HostTranslateInfo& host_info);
 
-std::optional<ConstBufferAddr> Track(const IR::Value& value, Environment& env) {
-    return IR::BreadthFirstSearch(value, [&env](const IR::Inst* inst) { return TryGetConstBuffer(inst, env); });
+std::optional<ConstBufferAddr> Track(const IR::Value& value, Environment& env, const HostTranslateInfo& host_info) {
+    return IR::BreadthFirstSearch(value, [&env, &host_info](const IR::Inst* inst) {
+        return TryGetConstBuffer(inst, env, host_info);
+    });
 }
 
 std::optional<u32> TryGetConstant(IR::Value& value, Environment& env) {
@@ -342,13 +335,13 @@ std::optional<u32> TryGetConstant(IR::Value& value, Environment& env) {
     return ReadCbufCached(env, index_number, offset_number);
 }
 
-std::optional<ConstBufferAddr> TryGetConstBuffer(const IR::Inst* inst, Environment& env) {
+std::optional<ConstBufferAddr> TryGetConstBuffer(const IR::Inst* inst, Environment& env, const HostTranslateInfo& host_info) {
     switch (inst->GetOpcode()) {
     default:
         return std::nullopt;
     case IR::Opcode::BitwiseOr32: {
-        std::optional lhs{TrackCached(inst->Arg(0), env)};
-        std::optional rhs{TrackCached(inst->Arg(1), env)};
+        std::optional lhs{TrackCached(inst->Arg(0), env, host_info)};
+        std::optional rhs{TrackCached(inst->Arg(1), env, host_info)};
         if (!lhs || !rhs) {
             return std::nullopt;
         }
@@ -378,12 +371,11 @@ std::optional<ConstBufferAddr> TryGetConstBuffer(const IR::Inst* inst, Environme
         if (!shift.IsImmediate()) {
             return std::nullopt;
         }
-        std::optional lhs{TrackCached(inst->Arg(0), env)};
+        std::optional lhs{TrackCached(inst->Arg(0), env, host_info)};
         if (lhs) {
             lhs->shift_left = shift.U32();
         }
         return lhs;
-        break;
     }
     case IR::Opcode::BitwiseAnd32: {
         IR::Value op1{inst->Arg(0)};
@@ -407,7 +399,7 @@ std::optional<ConstBufferAddr> TryGetConstBuffer(const IR::Inst* inst, Environme
                 return std::nullopt;
             } while (false);
         }
-        std::optional lhs{TrackCached(op1, env)};
+        std::optional lhs{TrackCached(op1, env, host_info)};
         if (lhs) {
             lhs->shift_left = static_cast<u32>(std::countr_zero(op2.U32()));
         }
@@ -453,7 +445,10 @@ std::optional<ConstBufferAddr> TryGetConstBuffer(const IR::Inst* inst, Environme
     } else {
         return std::nullopt;
     }
-    const u32 size_shift{DynamicDescriptorSizeShift(dynamic_offset)};
+    auto const size_shift = DynamicDescriptorSizeShift(dynamic_offset);
+    auto const sampled_limit = (std::max)(1U, (std::min)(host_info.max_per_stage_descriptor_sampled_images,
+                                                         host_info.max_descriptor_set_sampled_images));
+    auto const resource_limit = (std::max)(1U, host_info.max_per_stage_resources);
     return ConstBufferAddr{
         .index = index.U32(),
         .offset = base_offset,
@@ -462,15 +457,15 @@ std::optional<ConstBufferAddr> TryGetConstBuffer(const IR::Inst* inst, Environme
         .secondary_offset = 0,
         .secondary_shift_left = 0,
         .dynamic_offset = dynamic_offset,
-        .count = DynamicDescriptorCount(base_offset, size_shift),
+        .count = DynamicDescriptorCount(base_offset, size_shift, (std::min)({DESCRIPTOR_MAX_COUNT, sampled_limit, resource_limit})),
         .has_secondary = false,
     };
 }
 
-TextureInst MakeInst(Environment& env, IR::Block* block, IR::Inst& inst) {
+TextureInst MakeInst(Environment& env, IR::Block* block, IR::Inst& inst, const HostTranslateInfo& host_info) {
     ConstBufferAddr addr;
     if (IsBindless(inst)) {
-        const std::optional<ConstBufferAddr> track_addr{TrackCached(inst.Arg(0), env)};
+        const std::optional<ConstBufferAddr> track_addr{TrackCached(inst.Arg(0), env, host_info)};
 
         if (!track_addr) {
             throw NotImplementedException("Failed to track bindless texture constant buffer");
@@ -506,15 +501,15 @@ u32 GetTextureHandle(Environment& env, const ConstBufferAddr& cbuf) {
     return lhs_raw | rhs_raw;
 }
 
-    [[maybe_unused]]TextureType ReadTextureType(Environment& env, const ConstBufferAddr& cbuf) {
+[[maybe_unused]] TextureType ReadTextureType(Environment& env, const ConstBufferAddr& cbuf) {
     return env.ReadTextureType(GetTextureHandle(env, cbuf));
 }
 
-    [[maybe_unused]]TexturePixelFormat ReadTexturePixelFormat(Environment& env, const ConstBufferAddr& cbuf) {
+[[maybe_unused]] TexturePixelFormat ReadTexturePixelFormat(Environment& env, const ConstBufferAddr& cbuf) {
     return env.ReadTexturePixelFormat(GetTextureHandle(env, cbuf));
 }
 
-    [[maybe_unused]]bool IsTexturePixelFormatInteger(Environment& env, const ConstBufferAddr& cbuf) {
+[[maybe_unused]] bool IsTexturePixelFormatInteger(Environment& env, const ConstBufferAddr& cbuf) {
     return env.IsTexturePixelFormatInteger(GetTextureHandle(env, cbuf));
 }
 
@@ -675,7 +670,7 @@ void TexturePass(Environment& env, IR::Program& program, const HostTranslateInfo
             if (!IsTextureInstruction(inst)) {
                 continue;
             }
-            to_replace.push_back(MakeInst(env, block, inst));
+            to_replace.push_back(MakeInst(env, block, inst, host_info));
         }
     }
     // Sort instructions to visit textures by constant buffer index, then by offset
@@ -689,8 +684,7 @@ void TexturePass(Environment& env, IR::Program& program, const HostTranslateInfo
         program.info.texture_descriptors,
         program.info.image_descriptors,
     };
-    const u32 sampled_dynamic_cap{
-        DynamicSampledTextureCap(program.info, host_info, DynamicSampledTextureArrayCount(to_replace))};
+    const u32 sampled_dynamic_cap = DynamicSampledTextureCap(program.info, host_info, DynamicSampledTextureArrayCount(to_replace));
     for (TextureInst& texture_inst : to_replace) {
         // TODO: Handle arrays
         IR::Inst* const inst{texture_inst.inst};

src/shader_recompiler/profile.h

@@ -10,6 +10,8 @@
 
 namespace Shader {
 
+enum class Stage : u32;
+
 struct Profile {
     u32 supported_spirv{0x00010000};
     bool unified_descriptor_binding{};
@@ -29,12 +31,16 @@ struct Profile {
     bool support_fp32_signed_zero_nan_preserve{};
     bool support_fp64_signed_zero_nan_preserve{};
     bool support_explicit_workgroup_layout{};
+    bool support_workgroup_layout_8bit_access{};
+    bool support_workgroup_layout_16bit_access{};
     bool support_vote{};
+    u32 supported_subgroup_stages{0x7F};
     bool support_viewport_index_layer_non_geometry{};
     bool support_viewport_mask{};
     bool support_typeless_image_loads{};
     bool support_demote_to_helper_invocation{};
     bool support_int64_atomics{};
+    bool support_shared_int64_atomics{};
     bool support_derivative_control{};
     bool support_geometry_shader_passthrough{};
     bool support_native_ndc{};
@@ -92,8 +98,11 @@ struct Profile {
     bool has_broken_robust{};
 
     u64 min_ssbo_alignment{};
-
     u32 max_user_clip_distances{};
+
+    bool SupportsSubgroupStage(Stage stage) const {
+        return (supported_subgroup_stages & (1u << static_cast<u32>(stage))) != 0;
+    }
 };
 
 } // namespace Shader

src/tests/common/host_memory.cpp

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -15,12 +18,19 @@ static constexpr auto PERMS = Common::MemoryPermission::ReadWrite;
 static constexpr auto HEAP = false;
 
 TEST_CASE("HostMemory: Initialize and deinitialize", "[common]") {
-    { HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE); }
-    { HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE); }
+    {
+        HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+        REQUIRE(mem.BackingBasePointer() != nullptr);
+    }
+    {
+        HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+        REQUIRE(mem.BackingBasePointer() != nullptr);
+    }
 }
 
 TEST_CASE("HostMemory: Simple map", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x5000, 0x8000, 0x1000, PERMS, HEAP);
 
     volatile u8* const data = mem.VirtualBasePointer() + 0x5000;
@@ -30,6 +40,7 @@ TEST_CASE("HostMemory: Simple map", "[common]") {
 
 TEST_CASE("HostMemory: Simple mirror map", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x5000, 0x3000, 0x2000, PERMS, HEAP);
     mem.Map(0x8000, 0x4000, 0x1000, PERMS, HEAP);
 
@@ -41,6 +52,7 @@ TEST_CASE("HostMemory: Simple mirror map", "[common]") {
 
 TEST_CASE("HostMemory: Simple unmap", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x5000, 0x3000, 0x2000, PERMS, HEAP);
 
     volatile u8* const data = mem.VirtualBasePointer() + 0x5000;
@@ -52,6 +64,7 @@ TEST_CASE("HostMemory: Simple unmap", "[common]") {
 
 TEST_CASE("HostMemory: Simple unmap and remap", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x5000, 0x3000, 0x2000, PERMS, HEAP);
 
     volatile u8* const data = mem.VirtualBasePointer() + 0x5000;
@@ -69,6 +82,7 @@ TEST_CASE("HostMemory: Simple unmap and remap", "[common]") {
 
 TEST_CASE("HostMemory: Nieche allocation", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x0000, 0, 0x20000, PERMS, HEAP);
     mem.Unmap(0x0000, 0x4000, HEAP);
     mem.Map(0x1000, 0, 0x2000, PERMS, HEAP);
@@ -78,6 +92,7 @@ TEST_CASE("HostMemory: Nieche allocation", "[common]") {
 
 TEST_CASE("HostMemory: Full unmap", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x8000, 0, 0x4000, PERMS, HEAP);
     mem.Unmap(0x8000, 0x4000, HEAP);
     mem.Map(0x6000, 0, 0x16000, PERMS, HEAP);
@@ -85,6 +100,7 @@ TEST_CASE("HostMemory: Full unmap", "[common]") {
 
 TEST_CASE("HostMemory: Right out of bounds unmap", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x0000, 0, 0x4000, PERMS, HEAP);
     mem.Unmap(0x2000, 0x4000, HEAP);
     mem.Map(0x2000, 0x80000, 0x4000, PERMS, HEAP);
@@ -92,6 +108,7 @@ TEST_CASE("HostMemory: Right out of bounds unmap", "[common]") {
 
 TEST_CASE("HostMemory: Left out of bounds unmap", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+
     mem.Map(0x8000, 0, 0x4000, PERMS, HEAP);
     mem.Unmap(0x6000, 0x4000, HEAP);
     mem.Map(0x8000, 0, 0x2000, PERMS, HEAP);
@@ -99,6 +116,7 @@ TEST_CASE("HostMemory: Left out of bounds unmap", "[common]") {
 
 TEST_CASE("HostMemory: Multiple placeholder unmap", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x0000, 0, 0x4000, PERMS, HEAP);
     mem.Map(0x4000, 0, 0x1b000, PERMS, HEAP);
     mem.Unmap(0x3000, 0x1c000, HEAP);
@@ -107,6 +125,7 @@ TEST_CASE("HostMemory: Multiple placeholder unmap", "[common]") {
 
 TEST_CASE("HostMemory: Unmap between placeholders", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x0000, 0, 0x4000, PERMS, HEAP);
     mem.Map(0x4000, 0, 0x4000, PERMS, HEAP);
     mem.Unmap(0x2000, 0x4000, HEAP);
@@ -115,6 +134,7 @@ TEST_CASE("HostMemory: Unmap between placeholders", "[common]") {
 
 TEST_CASE("HostMemory: Unmap to origin", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x4000, 0, 0x4000, PERMS, HEAP);
     mem.Map(0x8000, 0, 0x4000, PERMS, HEAP);
     mem.Unmap(0x4000, 0x4000, HEAP);
@@ -124,6 +144,7 @@ TEST_CASE("HostMemory: Unmap to origin", "[common]") {
 
 TEST_CASE("HostMemory: Unmap to right", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x4000, 0, 0x4000, PERMS, HEAP);
     mem.Map(0x8000, 0, 0x4000, PERMS, HEAP);
     mem.Unmap(0x8000, 0x4000, HEAP);
@@ -132,6 +153,7 @@ TEST_CASE("HostMemory: Unmap to right", "[common]") {
 
 TEST_CASE("HostMemory: Partial right unmap check bindings", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x4000, 0x10000, 0x4000, PERMS, HEAP);
 
     volatile u8* const ptr = mem.VirtualBasePointer() + 0x4000;
@@ -144,6 +166,7 @@ TEST_CASE("HostMemory: Partial right unmap check bindings", "[common]") {
 
 TEST_CASE("HostMemory: Partial left unmap check bindings", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x4000, 0x10000, 0x4000, PERMS, HEAP);
 
     volatile u8* const ptr = mem.VirtualBasePointer() + 0x4000;
@@ -158,6 +181,7 @@ TEST_CASE("HostMemory: Partial left unmap check bindings", "[common]") {
 
 TEST_CASE("HostMemory: Partial middle unmap check bindings", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x4000, 0x10000, 0x4000, PERMS, HEAP);
 
     volatile u8* const ptr = mem.VirtualBasePointer() + 0x4000;
@@ -172,6 +196,7 @@ TEST_CASE("HostMemory: Partial middle unmap check bindings", "[common]") {
 
 TEST_CASE("HostMemory: Partial sparse middle unmap and check bindings", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x4000, 0x10000, 0x2000, PERMS, HEAP);
     mem.Map(0x6000, 0x20000, 0x2000, PERMS, HEAP);
 

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -245,16 +245,31 @@ ShaderCache::ShaderCache(Tegra::MaxwellDeviceMemoryManager& device_memory_,
               std::min<u32>(device.GetMaxUserClipDistances(), Maxwell::Regs::NumClipDistances),
       },
       host_info{
-          .support_float64 = true,
-          .support_float16 = false,
-          .support_int64 = device.HasShaderInt64(),
-          .needs_demote_reorder = device.IsAmd(),
-          .support_snorm_render_buffer = false,
-          .support_viewport_index_layer = device.HasVertexViewportLayer(),
-          .min_ssbo_alignment = static_cast<u32>(device.GetShaderStorageBufferAlignment()),
-          .support_geometry_shader_passthrough = device.HasGeometryShaderPassthrough(),
-          .support_conditional_barrier = device.SupportsConditionalBarriers(),
+        .min_ssbo_alignment = static_cast<u32>(device.GetShaderStorageBufferAlignment()),
+        .max_per_stage_descriptor_sampled_images =
+            Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .max_per_stage_resources = Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .max_descriptor_set_samplers = Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .max_descriptor_set_uniform_buffers = Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .max_descriptor_set_uniform_buffers_dynamic =
+            Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .max_descriptor_set_storage_buffers = Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .max_descriptor_set_storage_buffers_dynamic =
+            Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .max_descriptor_set_sampled_images = Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .max_descriptor_set_storage_images = Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .max_descriptor_set_input_attachements =
+            Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .support_float64 = true,
+        .support_float16 = false,
+        .support_int64 = device.HasShaderInt64(),
+        .needs_demote_reorder = device.IsAmd(),
+        .support_snorm_render_buffer = false,
+        .support_viewport_index_layer = device.HasVertexViewportLayer(),
+        .support_geometry_shader_passthrough = device.HasGeometryShaderPassthrough(),
+        .support_conditional_barrier = device.SupportsConditionalBarriers(),
       } {
+    host_info.ApplyDescriptorLimitPolicy();
     if (use_asynchronous_shaders) {
         workers = CreateWorkers();
     }

src/video_core/renderer_vulkan/pipeline_helper.h

@@ -22,6 +22,15 @@ namespace Vulkan {
 
 using Shader::Backend::SPIRV::NUM_TEXTURE_AND_IMAGE_SCALING_WORDS;
 
+[[nodiscard]] inline u32 NumDescriptorEntries(const Shader::Info& info) {
+    return Shader::NumDescriptors(info.constant_buffer_descriptors) +
+           Shader::NumDescriptors(info.storage_buffers_descriptors) +
+           Shader::NumDescriptors(info.texture_buffer_descriptors) +
+           Shader::NumDescriptors(info.image_buffer_descriptors) +
+           Shader::NumDescriptors(info.texture_descriptors) +
+           Shader::NumDescriptors(info.image_descriptors);
+}
+
 class DescriptorLayoutBuilder {
 public:
     DescriptorLayoutBuilder(const Device& device_) : device{&device_} {}
@@ -194,7 +203,11 @@ inline void PushImageDescriptors(TextureCache& texture_cache,
             const VideoCommon::ImageViewId image_view_id{(views++)->id};
             const VideoCommon::SamplerId sampler_id{*(samplers++)};
             ImageView& image_view{texture_cache.GetImageView(image_view_id)};
-            const VkImageView vk_image_view{image_view.Handle(desc.type)};
+            VkImageView vk_image_view{image_view.Handle(desc.type)};
+            if (vk_image_view == VK_NULL_HANDLE) {
+                const VkImageView null_image_view{texture_cache.GetImageView(VideoCommon::NULL_IMAGE_VIEW_ID).Handle(desc.type)};
+                if (null_image_view != VK_NULL_HANDLE) vk_image_view = null_image_view;
+            }
             const Sampler& sampler{texture_cache.GetSampler(sampler_id)};
             const bool use_fallback_sampler{sampler.HasAddedAnisotropy() &&
                                             !image_view.SupportsAnisotropy()};

src/video_core/renderer_vulkan/vk_compute_pass.cpp

@@ -326,7 +326,7 @@ std::pair<VkBuffer, VkDeviceSize> Uint8Pass::Assemble(u32 num_vertices, VkBuffer
     const u32 staging_size = static_cast<u32>(num_vertices * sizeof(u16));
     const auto staging = staging_buffer_pool.Request(staging_size, MemoryUsage::DeviceLocal);
 
-    compute_pass_descriptor_queue.Acquire();
+    compute_pass_descriptor_queue.Acquire(scheduler, 2);
     compute_pass_descriptor_queue.AddBuffer(src_buffer, src_offset, num_vertices);
     compute_pass_descriptor_queue.AddBuffer(staging.buffer, staging.offset, staging_size);
     const void* const descriptor_data{compute_pass_descriptor_queue.UpdateData()};
@@ -384,7 +384,7 @@ std::pair<VkBuffer, VkDeviceSize> QuadIndexedPass::Assemble(
     const std::size_t staging_size = num_tri_vertices * sizeof(u32);
     const auto staging = staging_buffer_pool.Request(staging_size, MemoryUsage::DeviceLocal);
 
-    compute_pass_descriptor_queue.Acquire();
+    compute_pass_descriptor_queue.Acquire(scheduler, 2);
     compute_pass_descriptor_queue.AddBuffer(src_buffer, src_offset, input_size);
     compute_pass_descriptor_queue.AddBuffer(staging.buffer, staging.offset, staging_size);
     const void* const descriptor_data{compute_pass_descriptor_queue.UpdateData()};
@@ -429,7 +429,7 @@ void ConditionalRenderingResolvePass::Resolve(VkBuffer dst_buffer, VkBuffer src_
     }
     const size_t compare_size = compare_to_zero ? 8 : 24;
 
-    compute_pass_descriptor_queue.Acquire();
+    compute_pass_descriptor_queue.Acquire(scheduler, 2);
     compute_pass_descriptor_queue.AddBuffer(src_buffer, src_offset, compare_size);
     compute_pass_descriptor_queue.AddBuffer(dst_buffer, 0, sizeof(u32));
     const void* const descriptor_data{compute_pass_descriptor_queue.UpdateData()};
@@ -498,7 +498,7 @@ void QueriesPrefixScanPass::Run(VkBuffer accumulation_buffer, VkBuffer dst_buffe
         static constexpr size_t DISPATCH_SIZE = 2048U;
         size_t runs_to_do = std::min<size_t>(current_runs, DISPATCH_SIZE);
         current_runs -= runs_to_do;
-        compute_pass_descriptor_queue.Acquire();
+        compute_pass_descriptor_queue.Acquire(scheduler, 3);
         compute_pass_descriptor_queue.AddBuffer(src_buffer, 0, number_of_sums * sizeof(u64));
         compute_pass_descriptor_queue.AddBuffer(dst_buffer, 0, number_of_sums * sizeof(u64));
         compute_pass_descriptor_queue.AddBuffer(accumulation_buffer, 0, sizeof(u64));
@@ -600,7 +600,7 @@ void ASTCDecoderPass::Assemble(Image& image, const StagingBufferRef& map,
         const u32 num_dispatches_y = Common::DivCeil(swizzle.num_tiles.height, 8U);
         const u32 num_dispatches_z = image.info.resources.layers;
 
-        compute_pass_descriptor_queue.Acquire();
+        compute_pass_descriptor_queue.Acquire(scheduler, 2);
         compute_pass_descriptor_queue.AddBuffer(map.buffer, input_offset,
                                                 image.guest_size_bytes - swizzle.buffer_offset);
         compute_pass_descriptor_queue.AddImage(image.StorageImageView(swizzle.level));
@@ -821,7 +821,7 @@ void BlockLinearUnswizzle3DPass::UnswizzleChunk(
     pc.blocks_dim[1] = blocks_y;
     pc.blocks_dim[2] = z_count; // Only process the count
 
-    compute_pass_descriptor_queue.Acquire();
+    compute_pass_descriptor_queue.Acquire(scheduler, 3);
     compute_pass_descriptor_queue.AddBuffer(*image.runtime->swizzle_table_buffer, 0,
                                            image.runtime->swizzle_table_size);
     compute_pass_descriptor_queue.AddBuffer(swizzled.buffer,
@@ -989,7 +989,7 @@ void MSAACopyPass::CopyImage(Image& dst_image, Image& src_image,
         ASSERT(copy.dst_subresource.base_layer == 0);
         ASSERT(copy.dst_subresource.num_layers == 1);
 
-        compute_pass_descriptor_queue.Acquire();
+        compute_pass_descriptor_queue.Acquire(scheduler, 2);
         compute_pass_descriptor_queue.AddImage(
             src_image.StorageImageView(copy.src_subresource.base_level));
         compute_pass_descriptor_queue.AddImage(

src/video_core/renderer_vulkan/vk_compute_pipeline.cpp

@@ -45,6 +45,7 @@ ComputePipeline::ComputePipeline(const Device& device_, Scheduler& scheduler, vk
     }
     std::copy_n(info.constant_buffer_used_sizes.begin(), uniform_buffer_sizes.size(),
                 uniform_buffer_sizes.begin());
+    num_descriptor_entries = NumDescriptorEntries(info);
 
     auto func{[this, &scheduler, &descriptor_pool, shader_notify, pipeline_statistics] {
         DescriptorLayoutBuilder builder{device};
@@ -113,7 +114,7 @@ ComputePipeline::ComputePipeline(const Device& device_, Scheduler& scheduler, vk
 void ComputePipeline::Configure(Tegra::Engines::KeplerCompute& kepler_compute,
                                 Tegra::MemoryManager& gpu_memory, Scheduler& scheduler,
                                 BufferCache& buffer_cache, TextureCache& texture_cache) {
-    guest_descriptor_queue.Acquire();
+    guest_descriptor_queue.Acquire(scheduler, num_descriptor_entries);
 
     buffer_cache.SetComputeUniformBufferState(info.constant_buffer_mask, &uniform_buffer_sizes);
     buffer_cache.UnbindComputeStorageBuffers();

src/video_core/renderer_vulkan/vk_compute_pipeline.h

@@ -53,6 +53,7 @@ private:
     vk::PipelineCache& pipeline_cache;
     GuestDescriptorQueue& guest_descriptor_queue;
     Shader::Info info;
+    u32 num_descriptor_entries{};
 
     VideoCommon::ComputeUniformBufferSizes uniform_buffer_sizes{};
 

src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp

@@ -268,6 +268,7 @@ GraphicsPipeline::GraphicsPipeline(
         num_textures += Shader::NumDescriptors(info->texture_descriptors);
         num_image_elements += Shader::NumDescriptors(info->texture_descriptors);
         num_image_elements += Shader::NumDescriptors(info->image_descriptors);
+        num_descriptor_entries += NumDescriptorEntries(*info);
     }
     fragment_has_color0_output = stage_infos[NUM_STAGES - 1].stores_frag_color[0];
     auto func{[this, shader_notify, &render_pass_cache, &descriptor_pool, pipeline_statistics] {
@@ -473,7 +474,7 @@ bool GraphicsPipeline::ConfigureImpl(bool is_indexed) {
     buffer_cache.UpdateGraphicsBuffers(is_indexed);
     buffer_cache.BindHostGeometryBuffers(is_indexed);
 
-    guest_descriptor_queue.Acquire();
+    guest_descriptor_queue.Acquire(scheduler, num_descriptor_entries);
 
     RescalingPushConstant rescaling;
     RenderAreaPushConstant render_area;

src/video_core/renderer_vulkan/vk_graphics_pipeline.h

@@ -159,6 +159,7 @@ private:
     std::array<Shader::Info, NUM_STAGES> stage_infos;
     std::array<u32, 5> enabled_uniform_buffer_masks{};
     VideoCommon::UniformBufferSizes uniform_buffer_sizes{};
+    u32 num_descriptor_entries{};
     size_t num_image_elements{};
     u32 num_textures{};
     bool fragment_has_color0_output{};

src/video_core/renderer_vulkan/vk_pipeline_cache.cpp

@@ -378,6 +378,20 @@ PipelineCache::PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_,
       serialization_thread(1, "VkPipelineSerialization") {
     const auto& float_control{device.FloatControlProperties()};
     const VkDriverId driver_id{device.GetDriverID()};
+    const VkShaderStageFlags subgroup_stages{device.GetSubgroupSupportedStages()};
+    const auto subgroup_stage_bit{[subgroup_stages](VkShaderStageFlags flag, Shader::Stage stage) {
+        return (subgroup_stages & flag) != 0 ? (1u << static_cast<u32>(stage)) : 0u;
+    }};
+    const u32 supported_subgroup_stages{
+        subgroup_stage_bit(VK_SHADER_STAGE_VERTEX_BIT, Shader::Stage::VertexA) |
+        subgroup_stage_bit(VK_SHADER_STAGE_VERTEX_BIT, Shader::Stage::VertexB) |
+        subgroup_stage_bit(VK_SHADER_STAGE_TESSELLATION_CONTROL_BIT,
+                           Shader::Stage::TessellationControl) |
+        subgroup_stage_bit(VK_SHADER_STAGE_TESSELLATION_EVALUATION_BIT,
+                           Shader::Stage::TessellationEval) |
+        subgroup_stage_bit(VK_SHADER_STAGE_GEOMETRY_BIT, Shader::Stage::Geometry) |
+        subgroup_stage_bit(VK_SHADER_STAGE_FRAGMENT_BIT, Shader::Stage::Fragment) |
+        subgroup_stage_bit(VK_SHADER_STAGE_COMPUTE_BIT, Shader::Stage::Compute)};
     profile = Shader::Profile{
         .supported_spirv = device.SupportedSpirvVersion(),
         .unified_descriptor_binding = true,
@@ -402,7 +416,12 @@ PipelineCache::PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_,
         .support_fp64_signed_zero_nan_preserve =
             float_control.shaderSignedZeroInfNanPreserveFloat64 != VK_FALSE,
         .support_explicit_workgroup_layout = device.IsKhrWorkgroupMemoryExplicitLayoutSupported(),
+        .support_workgroup_layout_8bit_access =
+            device.IsWorkgroupMemoryExplicitLayout8BitSupported(),
+        .support_workgroup_layout_16bit_access =
+            device.IsWorkgroupMemoryExplicitLayout16BitSupported(),
         .support_vote = device.IsSubgroupFeatureSupported(VK_SUBGROUP_FEATURE_VOTE_BIT),
+        .supported_subgroup_stages = supported_subgroup_stages,
         .support_viewport_index_layer_non_geometry =
             device.IsExtShaderViewportIndexLayerSupported(),
         .support_viewport_mask = device.IsNvViewportArray2Supported(),
@@ -410,6 +429,7 @@ PipelineCache::PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_,
         .support_demote_to_helper_invocation =
             device.IsExtShaderDemoteToHelperInvocationSupported(),
         .support_int64_atomics = device.IsExtShaderAtomicInt64Supported(),
+        .support_shared_int64_atomics = device.IsSharedInt64AtomicsSupported(),
         .support_derivative_control = true,
         .support_geometry_shader_passthrough = device.IsNvGeometryShaderPassthroughSupported(),
         .support_native_ndc = device.IsExtDepthClipControlSupported(),
@@ -433,16 +453,28 @@ PipelineCache::PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_,
         .has_broken_spirv_position_input = driver_id == false,
         .has_broken_unsigned_image_offsets = false,
         .has_broken_signed_operations = false,
-        .has_broken_fp16_float_controls = driver_id == VK_DRIVER_ID_NVIDIA_PROPRIETARY,
+        .has_broken_fp16_float_controls = driver_id == VK_DRIVER_ID_NVIDIA_PROPRIETARY ||
+                                          driver_id == VK_DRIVER_ID_QUALCOMM_PROPRIETARY,
         .ignore_nan_fp_comparisons = false,
         .has_broken_spirv_subgroup_mask_vector_extract_dynamic = false,
         .has_broken_robust =
             device.IsNvidia() && device.GetNvidiaArch() <= NvidiaArchitecture::Arch_Pascal,
         .min_ssbo_alignment = device.GetStorageBufferAlignment(),
-        .max_user_clip_distances = device.GetMaxUserClipDistances(),
+        .max_user_clip_distances = device.GetMaxUserClipDistances()
     };
 
     host_info = Shader::HostTranslateInfo{
+        .min_ssbo_alignment = device.GetStorageBufferAlignment(),
+        .max_per_stage_descriptor_sampled_images = device.GetMaxPerStageDescriptorSampledImages(),
+        .max_per_stage_resources = device.GetMaxPerStageResources(),
+        .max_descriptor_set_samplers = device.GetMaxDescriptorSetSamplers(),
+        .max_descriptor_set_uniform_buffers = device.GetMaxDescriptorSetUniformBuffers(),
+        .max_descriptor_set_uniform_buffers_dynamic = device.GetMaxDescriptorSetUniformBuffersDynamic(),
+        .max_descriptor_set_storage_buffers = device.GetMaxDescriptorSetStorageBuffers(),
+        .max_descriptor_set_storage_buffers_dynamic = device.GetMaxDescriptorSetStorageBuffersDynamic(),
+        .max_descriptor_set_sampled_images = device.GetMaxDescriptorSetSampledImages(),
+        .max_descriptor_set_storage_images = device.GetMaxDescriptorSetStorageImages(),
+        .max_descriptor_set_input_attachements = device.GetMaxDescriptorSetInputAttachments(),
         .support_float64 = device.IsFloat64Supported(),
         .support_float16 = device.IsFloat16Supported(),
         .support_int64 = device.IsShaderInt64Supported(),
@@ -451,13 +483,10 @@ PipelineCache::PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_,
                                 driver_id == VK_DRIVER_ID_SAMSUNG_PROPRIETARY,
         .support_snorm_render_buffer = true,
         .support_viewport_index_layer = device.IsExtShaderViewportIndexLayerSupported(),
-        .min_ssbo_alignment = static_cast<u32>(device.GetStorageBufferAlignment()),
-        .max_per_stage_descriptor_sampled_images = device.GetMaxPerStageDescriptorSampledImages(),
-        .max_per_stage_resources = device.GetMaxPerStageResources(),
-        .max_descriptor_set_sampled_images = device.GetMaxDescriptorSetSampledImages(),
         .support_geometry_shader_passthrough = device.IsNvGeometryShaderPassthroughSupported(),
         .support_conditional_barrier = device.SupportsConditionalBarriers(),
     };
+    host_info.ApplyDescriptorLimitPolicy();
 
     if (device.GetMaxVertexInputAttributes() < Maxwell::NumVertexAttributes) {
         LOG_WARNING(Render_Vulkan, "maxVertexInputAttributes is too low: {} < {}",
@@ -492,10 +521,14 @@ PipelineCache::PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_,
         device.IsExtExtendedDynamicState3BlendingSupported();
     dynamic_features.has_extended_dynamic_state_3_enables =
         device.IsExtExtendedDynamicState3EnablesSupported();
-    dynamic_features.has_dynamic_state3_depth_clamp_enable = false;
+    dynamic_features.has_dynamic_state3_depth_clamp_enable =
+        dynamic_features.has_extended_dynamic_state_3_enables &&
+        device.SupportsDynamicState3DepthClampEnable();
     dynamic_features.has_dynamic_state3_logic_op_enable =
+        dynamic_features.has_extended_dynamic_state_3_enables &&
         device.SupportsDynamicState3LogicOpEnable();
     dynamic_features.has_dynamic_state3_line_stipple_enable =
+        dynamic_features.has_extended_dynamic_state_3_enables &&
         device.SupportsDynamicState3LineStippleEnable();
 
     // VIDS: Independent toggle (not affected by dyna_state levels)
@@ -892,12 +925,8 @@ std::unique_ptr<ComputePipeline> PipelineCache::CreateComputePipeline(
     }
 
     auto program{TranslateProgram(pools.inst, pools.block, env, cfg, host_info)};
-    const VkDriverIdKHR driver_id = device.GetDriverID();
-    const bool needs_shared_mem_clamp =
-        driver_id == VK_DRIVER_ID_QUALCOMM_PROPRIETARY ||
-        driver_id == VK_DRIVER_ID_ARM_PROPRIETARY;
     const u32 max_shared_memory = device.GetMaxComputeSharedMemorySize();
-    if (needs_shared_mem_clamp && program.shared_memory_size > max_shared_memory) {
+    if (program.shared_memory_size > max_shared_memory) {
         LOG_WARNING(Render_Vulkan,
                     "Compute shader 0x{:016x} requests {}KB shared memory but device max is {}KB - clamping",
                     key.unique_hash,

src/video_core/renderer_vulkan/vk_query_cache.cpp

@@ -218,8 +218,7 @@ public:
         }
         PauseCounter();
         const auto driver_id = device.GetDriverID();
-        if (driver_id == VK_DRIVER_ID_QUALCOMM_PROPRIETARY ||
-            driver_id == VK_DRIVER_ID_ARM_PROPRIETARY || driver_id == VK_DRIVER_ID_MESA_TURNIP) {
+        if (driver_id == VK_DRIVER_ID_ARM_PROPRIETARY || driver_id == VK_DRIVER_ID_MESA_TURNIP) {
             pending_sync.clear();
             sync_values_stash.clear();
             return;

src/video_core/renderer_vulkan/vk_rasterizer.cpp

@@ -203,7 +203,7 @@ RasterizerVulkan::RasterizerVulkan(Core::Frontend::EmuWindow& emu_window_, Tegra
     : gpu{gpu_}, device_memory{device_memory_}, device{device_},
       memory_allocator{memory_allocator_}, state_tracker{state_tracker_}, scheduler{scheduler_},
       staging_pool(device, memory_allocator, scheduler), descriptor_pool(device, scheduler),
-      guest_descriptor_queue(device, scheduler), compute_pass_descriptor_queue(device, scheduler),
+      guest_descriptor_queue(device), compute_pass_descriptor_queue(device),
       blit_image(device, scheduler, state_tracker, descriptor_pool), render_pass_cache(device),
       texture_cache_runtime{
           device,     scheduler,         memory_allocator, staging_pool,
@@ -1255,6 +1255,18 @@ void RasterizerVulkan::UpdateDepthBias(Tegra::Engines::Maxwell3D::Regs& regs) {
         }
     }
 
+    const bool is_float_depth =
+        regs.zeta.format == Tegra::DepthFormat::Z32_FLOAT ||
+        regs.zeta.format == Tegra::DepthFormat::Z32_FLOAT_X24S8_UINT;
+    if (is_float_depth && units != 0.0f && !device.IsExtDepthBiasControlSupported()) {
+        static bool logged_float_bias_warning = false;
+        if (!logged_float_bias_warning) {
+            logged_float_bias_warning = true;
+            LOG_WARNING(Render_Vulkan,
+                        "Depth bias on a float depth target without VK_EXT_depth_bias_control");
+        }
+    }
+
     scheduler.Record([constant = units, clamp = regs.depth_bias_clamp,
                       factor = regs.slope_scale_depth_bias, this](vk::CommandBuffer cmdbuf) {
         if (device.IsExtDepthBiasControlSupported()) {

src/video_core/renderer_vulkan/vk_scheduler.cpp

@@ -155,15 +155,14 @@ void Scheduler::WaitWorker() {
 }
 
 void Scheduler::DispatchWork() {
-    if (chunk->Empty()) {
-        return;
+    if (chunk && !chunk->Empty()) {
+        {
+            std::scoped_lock ql{queue_mutex};
+            work_queue.push(std::move(chunk));
+        }
+        event_cv.notify_all();
+        AcquireNewChunk();
     }
-    {
-        std::scoped_lock ql{queue_mutex};
-        work_queue.push(std::move(chunk));
-    }
-    event_cv.notify_all();
-    AcquireNewChunk();
 }
 
 void Scheduler::RequestRenderpass(const Framebuffer* framebuffer) {

src/video_core/renderer_vulkan/vk_update_descriptor.cpp

@@ -7,6 +7,7 @@
 #include <variant>
 #include <boost/container/static_vector.hpp>
 
+#include "common/assert.h"
 #include "common/logging.h"
 #include "video_core/renderer_vulkan/vk_scheduler.h"
 #include "video_core/renderer_vulkan/vk_update_descriptor.h"
@@ -15,8 +16,9 @@
 
 namespace Vulkan {
 
-UpdateDescriptorQueue::UpdateDescriptorQueue(const Device& device_, Scheduler& scheduler_)
-    : device{device_}, scheduler{scheduler_} {
+UpdateDescriptorQueue::UpdateDescriptorQueue(const Device& device_)
+    : device{device_}
+{
     payload_start = payload.data();
     payload_cursor = payload.data();
 }
@@ -31,13 +33,15 @@ void UpdateDescriptorQueue::TickFrame() {
     payload_cursor = payload_start;
 }
 
-void UpdateDescriptorQueue::Acquire() {
-    // Minimum number of entries required.
-    // This is the maximum number of entries a single draw call might use.
-    static constexpr size_t MIN_ENTRIES = 0x400;
-
-    if (std::distance(payload_start, payload_cursor) + MIN_ENTRIES >= FRAME_PAYLOAD_SIZE) {
-        LOG_WARNING(Render_Vulkan, "Payload overflow, waiting for worker thread");
+void UpdateDescriptorQueue::Acquire(Scheduler& scheduler, size_t required_entries) {
+    static constexpr size_t DEFAULT_REQUIRED_ENTRIES = 0x400;
+    const size_t reserve = required_entries > 0 ? required_entries : DEFAULT_REQUIRED_ENTRIES;
+    ASSERT_MSG(reserve < FRAME_PAYLOAD_SIZE, "Descriptor reservation {} >= frame capacity {}",
+               reserve, FRAME_PAYLOAD_SIZE);
+    const size_t used = static_cast<size_t>(std::distance(payload_start, payload_cursor));
+    if (used + reserve >= FRAME_PAYLOAD_SIZE) {
+        LOG_WARNING(Render_Vulkan, "Payload overflow (used={}, reserve={}, capacity={})",
+                    used, reserve, FRAME_PAYLOAD_SIZE);
         scheduler.WaitWorker();
         payload_cursor = payload_start;
     }

src/video_core/renderer_vulkan/vk_update_descriptor.h

@@ -34,12 +34,11 @@ class UpdateDescriptorQueue final {
     static constexpr size_t PAYLOAD_SIZE = FRAME_PAYLOAD_SIZE * FRAMES_IN_FLIGHT;
 
 public:
-    explicit UpdateDescriptorQueue(const Device& device_, Scheduler& scheduler_);
+    explicit UpdateDescriptorQueue(const Device& device_);
     ~UpdateDescriptorQueue();
 
     void TickFrame();
-
-    void Acquire();
+    void Acquire(Scheduler& scheduler, size_t required_entries = 0);
 
     const DescriptorUpdateEntry* UpdateData() const noexcept {
         return upload_start;
@@ -75,8 +74,6 @@ public:
 
 private:
     const Device& device;
-    Scheduler& scheduler;
-
     size_t frame_index{0};
     DescriptorUpdateEntry* payload_cursor = nullptr;
     DescriptorUpdateEntry* payload_start = nullptr;

src/video_core/shader_environment.cpp

@@ -255,8 +255,7 @@ std::optional<u64> GenericEnvironment::TryFindSize() {
     static constexpr u64 SELF_BRANCH_A = 0xE2400FFFFF87000FULL;
     static constexpr u64 SELF_BRANCH_B = 0xE2400FFFFF07000FULL;
 
-    static constexpr u64 MESA_EXIT_MASK = 0xFFF00000000F001FULL;
-    static constexpr u64 MESA_EXIT_VALUE = (0xE30ULL << 52) | (0x7ULL << 16) | 0xFULL;
+    static constexpr u64 EXIT_VALUE = 0xE30000000007000FULL;
 
     code.resize(MAXIMUM_SIZE / INST_SIZE);
 
@@ -271,7 +270,7 @@ std::optional<u64> GenericEnvironment::TryFindSize() {
             if (inst == SELF_BRANCH_A || inst == SELF_BRANCH_B) {
                 return offset + index;
             }
-            if ((inst & MESA_EXIT_MASK) == MESA_EXIT_VALUE) {
+            if (!is_proprietary_driver && inst == EXIT_VALUE) {
                 return offset + index + INST_SIZE;
             }
         }

src/video_core/texture_cache/texture_cache.h

@@ -1915,13 +1915,16 @@ void TextureCache<P>::TrimInactiveSamplers(size_t budget) {
         ankerl::unordered_dense::set<SamplerId> active_sampler_ids;
         for (auto const& e : channel_state->sampler_ids)
             active_sampler_ids.insert(e.second);
+        if constexpr (requires { runtime.Finish(); }) {
+            runtime.Finish();
+        }
         // Elements in the map must be necesarily valid
         size_t removed = 0;
         for (auto it = channel_state->samplers.begin(); it != channel_state->samplers.end();) {
             const SamplerId sampler_id = it->second;
             if (!sampler_id || sampler_id == CORRUPT_ID) {
                 it = channel_state->samplers.erase(it);
-            } else if (std::ranges::find(active_sampler_ids, sampler_id) != active_sampler_ids.end()) {
+            } else if (active_sampler_ids.contains(sampler_id)) {
                 ++it;
             } else {
                 slot_samplers.erase(sampler_id);

src/video_core/vulkan_common/vulkan_device.cpp

@@ -501,14 +501,6 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
         LOG_WARNING(Render_Vulkan,
                     "Qualcomm drivers have slow push descriptor implementation");
         RemoveExtension(extensions.push_descriptor, VK_KHR_PUSH_DESCRIPTOR_EXTENSION_NAME);
-        LOG_WARNING(Render_Vulkan,
-                    "Disabling shader float controls and 64-bit integer features on Qualcomm proprietary drivers");
-        RemoveExtension(extensions.shader_float_controls, VK_KHR_SHADER_FLOAT_CONTROLS_EXTENSION_NAME);
-        RemoveExtensionFeature(extensions.shader_atomic_int64, features.shader_atomic_int64,
-                               VK_KHR_SHADER_ATOMIC_INT64_EXTENSION_NAME);
-        features.shader_atomic_int64.shaderBufferInt64Atomics = false;
-        features.shader_atomic_int64.shaderSharedInt64Atomics = false;
-        features.features.shaderInt64 = false;
 
 #if defined(__ANDROID__) && defined(ARCHITECTURE_arm64)
         // BCn patching only safe on Android 9+ (API 28+). Older versions crash on driver load.
@@ -561,7 +553,6 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
             features.shader_float16_int8.shaderFloat16 = false;
         }
 
-        // Mali/ NVIDIA proprietary drivers: Shader stencil export not supported
         // Use hardware depth/stencil blits instead when available
         if (!extensions.shader_stencil_export) {
             LOG_INFO(Render_Vulkan,
@@ -574,8 +565,8 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
 
             if (!is_blit_depth24_stencil8_supported && !is_blit_depth32_stencil8_supported) {
                 LOG_WARNING(Render_Vulkan,
-                            "NVIDIA: Neither shader export nor hardware blits available for "
-                            "depth/stencil. Performance may be degraded.");
+                            "Neither shader export nor hardware blits available for "
+                            "depth/stencil.");
             }
         }
     }
@@ -653,21 +644,13 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
 
     if (is_turnip || is_qualcomm) {
         LOG_WARNING(Render_Vulkan, "Driver requires higher-than-reported binding limits");
-        properties.properties.limits.maxVertexInputBindings = 32;
+        properties.properties.limits.maxVertexInputBindings =
+            (std::max)(properties.properties.limits.maxVertexInputBindings, 32U);
     }
 
     const auto dyna_state = Settings::values.dyna_state.GetValue();
-
-    // Base dynamic states (VIEWPORT, SCISSOR, DEPTH_BIAS, etc.) are ALWAYS active in vk_graphics_pipeline.cpp
-    // This slider controls EXTENDED dynamic states with accumulative levels per Vulkan specs:
-    //   Level 0 = Core Dynamic States only (Vulkan 1.0)
-    //   Level 1 = Core + VK_EXT_extended_dynamic_state
-    //   Level 2 = Core + VK_EXT_extended_dynamic_state + VK_EXT_extended_dynamic_state2
-    //   Level 3 = Core + VK_EXT_extended_dynamic_state + VK_EXT_extended_dynamic_state2 + VK_EXT_extended_dynamic_state3
-
     switch (dyna_state) {
     case Settings::ExtendedDynamicState::Disabled:
-        // Level 0: Disable all extended dynamic state extensions
         RemoveExtensionFeature(extensions.extended_dynamic_state, features.extended_dynamic_state,
                               VK_EXT_EXTENDED_DYNAMIC_STATE_EXTENSION_NAME);
         RemoveExtensionFeature(extensions.extended_dynamic_state2, features.extended_dynamic_state2,
@@ -678,7 +661,6 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
         dynamic_state3_enables = false;
         break;
     case Settings::ExtendedDynamicState::EDS1:
-        // Level 1: Enable EDS1, disable EDS2 and EDS3
         RemoveExtensionFeature(extensions.extended_dynamic_state2, features.extended_dynamic_state2,
                               VK_EXT_EXTENDED_DYNAMIC_STATE_2_EXTENSION_NAME);
         RemoveExtensionFeature(extensions.extended_dynamic_state3, features.extended_dynamic_state3,
@@ -687,7 +669,6 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
         dynamic_state3_enables = false;
         break;
     case Settings::ExtendedDynamicState::EDS2:
-        // Level 2: Enable EDS1 + EDS2, disable EDS3
         RemoveExtensionFeature(extensions.extended_dynamic_state3, features.extended_dynamic_state3,
                               VK_EXT_EXTENDED_DYNAMIC_STATE_3_EXTENSION_NAME);
         dynamic_state3_blending = false;
@@ -695,12 +676,9 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
         break;
     case Settings::ExtendedDynamicState::EDS3:
     default:
-        // Level 3: Enable all (EDS1 + EDS2 + EDS3)
         break;
     }
 
-    // VK_EXT_vertex_input_dynamic_state is independent from EDS
-    // It can be enabled even without extended_dynamic_state
     if (!Settings::values.vertex_input_dynamic_state.GetValue()) {
         RemoveExtensionFeature(extensions.vertex_input_dynamic_state, features.vertex_input_dynamic_state, VK_EXT_VERTEX_INPUT_DYNAMIC_STATE_EXTENSION_NAME);
     }
@@ -1282,8 +1260,7 @@ void Device::RemoveUnsuitableExtensions() {
                                        VK_EXT_IMAGE_ROBUSTNESS_EXTENSION_NAME);
 
     // VK_KHR_shader_atomic_int64
-    extensions.shader_atomic_int64 = features.shader_atomic_int64.shaderBufferInt64Atomics &&
-                                     features.shader_atomic_int64.shaderSharedInt64Atomics;
+    extensions.shader_atomic_int64 = features.shader_atomic_int64.shaderBufferInt64Atomics;
     RemoveExtensionFeatureIfUnsuitable(extensions.shader_atomic_int64, features.shader_atomic_int64,
                                        VK_KHR_SHADER_ATOMIC_INT64_EXTENSION_NAME);
 
@@ -1332,10 +1309,7 @@ void Device::RemoveUnsuitableExtensions() {
 
     // VK_KHR_workgroup_memory_explicit_layout
     extensions.workgroup_memory_explicit_layout =
-        features.features.shaderInt16 &&
         features.workgroup_memory_explicit_layout.workgroupMemoryExplicitLayout &&
-        features.workgroup_memory_explicit_layout.workgroupMemoryExplicitLayout8BitAccess &&
-        features.workgroup_memory_explicit_layout.workgroupMemoryExplicitLayout16BitAccess &&
         features.workgroup_memory_explicit_layout.workgroupMemoryExplicitLayoutScalarBlockLayout;
     RemoveExtensionFeatureIfUnsuitable(extensions.workgroup_memory_explicit_layout,
                                        features.workgroup_memory_explicit_layout,

src/video_core/vulkan_common/vulkan_device.h

@@ -321,32 +321,23 @@ public:
         return properties.properties.limits.maxPushConstantsSize;
     }
 
-    /// Returns the maximum size for shared memory.
-    u32 GetMaxComputeSharedMemorySize() const {
-        return properties.properties.limits.maxComputeSharedMemorySize;
-    }
-
-    /// Returns the maximum number of dynamic storage buffer descriptors per set.
-    u32 GetMaxDescriptorSetStorageBuffersDynamic() const {
-        return properties.properties.limits.maxDescriptorSetStorageBuffersDynamic;
-    }
-
-    /// Returns the maximum number of dynamic uniform buffer descriptors per set.
-    u32 GetMaxDescriptorSetUniformBuffersDynamic() const {
-        return properties.properties.limits.maxDescriptorSetUniformBuffersDynamic;
-    }
-
-    u32 GetMaxPerStageDescriptorSampledImages() const {
-        return properties.properties.limits.maxPerStageDescriptorSampledImages;
-    }
-
-    u32 GetMaxPerStageResources() const {
-        return properties.properties.limits.maxPerStageResources;
-    }
-
-    u32 GetMaxDescriptorSetSampledImages() const {
-        return properties.properties.limits.maxDescriptorSetSampledImages;
-    }
+#define FN_MAX_LIMIT_LIST \
+    FN_MAX_LIMIT_ELEM(ComputeSharedMemorySize) \
+    FN_MAX_LIMIT_ELEM(PerStageDescriptorSampledImages) \
+    FN_MAX_LIMIT_ELEM(PerStageResources) \
+    FN_MAX_LIMIT_ELEM(DescriptorSetSamplers) \
+    FN_MAX_LIMIT_ELEM(DescriptorSetUniformBuffers) \
+    FN_MAX_LIMIT_ELEM(DescriptorSetUniformBuffersDynamic) \
+    FN_MAX_LIMIT_ELEM(DescriptorSetStorageBuffers) \
+    FN_MAX_LIMIT_ELEM(DescriptorSetStorageBuffersDynamic) \
+    FN_MAX_LIMIT_ELEM(DescriptorSetSampledImages) \
+    FN_MAX_LIMIT_ELEM(DescriptorSetStorageImages) \
+    FN_MAX_LIMIT_ELEM(DescriptorSetInputAttachments)
+#define FN_MAX_LIMIT_ELEM(name) \
+    u32 GetMax##name() const { return properties.properties.limits.max##name; }
+FN_MAX_LIMIT_LIST
+#undef FN_MAX_LIMIT_ELEM
+#undef FN_MAX_LIMIT_LIST
 
     /// Returns float control properties of the device.
     const VkPhysicalDeviceFloatControlsPropertiesKHR& FloatControlProperties() const {
@@ -412,6 +403,11 @@ public:
         return properties.subgroup_properties.supportedOperations & feature;
     }
 
+    /// Returns the shader stages that support subgroup operations.
+    VkShaderStageFlags GetSubgroupSupportedStages() const {
+        return properties.subgroup_properties.supportedStages;
+    }
+
     /// Returns the maximum number of push descriptors.
     u32 MaxPushDescriptors() const {
         return properties.push_descriptor.maxPushDescriptors;
@@ -492,6 +488,18 @@ public:
         return extensions.workgroup_memory_explicit_layout;
     }
 
+    /// Returns true if the device supports 8-bit accesses to workgroup explicit layout memory.
+    bool IsWorkgroupMemoryExplicitLayout8BitSupported() const {
+        return extensions.workgroup_memory_explicit_layout &&
+               features.workgroup_memory_explicit_layout.workgroupMemoryExplicitLayout8BitAccess;
+    }
+
+    /// Returns true if the device supports 16-bit accesses to workgroup explicit layout memory.
+    bool IsWorkgroupMemoryExplicitLayout16BitSupported() const {
+        return extensions.workgroup_memory_explicit_layout && features.features.shaderInt16 &&
+               features.workgroup_memory_explicit_layout.workgroupMemoryExplicitLayout16BitAccess;
+    }
+
     /// Returns true if the device supports VK_KHR_image_format_list.
     bool IsKhrImageFormatListSupported() const {
         return extensions.image_format_list || instance_version >= VK_API_VERSION_1_2;
@@ -720,9 +728,16 @@ public:
                features.provoking_vertex.transformFeedbackPreservesProvokingVertex;
     }
 
-    /// Returns true if the device supports VK_KHR_shader_atomic_int64.
+    /// Returns true if the device supports int64 atomics on storage buffers.
     bool IsExtShaderAtomicInt64Supported() const {
-        return extensions.shader_atomic_int64;
+        return extensions.shader_atomic_int64 &&
+               features.shader_atomic_int64.shaderBufferInt64Atomics;
+    }
+
+    /// Returns true if the device supports int64 atomics on workgroup (shared) memory.
+    bool IsSharedInt64AtomicsSupported() const {
+        return extensions.shader_atomic_int64 &&
+               features.shader_atomic_int64.shaderSharedInt64Atomics;
     }
 
     bool IsExtConditionalRendering() const {

tools/maxwell-ir/main.cpp

@@ -44,6 +44,7 @@ int IrShaderRecompilerImpl(int argc, char *argv[]) {
     host_info.support_geometry_shader_passthrough = true;
     host_info.support_conditional_barrier = true;
     host_info.min_ssbo_alignment = 0;
+    host_info.ApplyDescriptorLimitPolicy();
     auto program = Shader::Maxwell::TranslateProgram(inst_pool, block_pool, env, cfg, host_info);
     auto const dumped_ir = Shader::IR::DumpProgram(program);
     std::printf("%s\n", dumped_ir.c_str());

tools/maxwell-spirv/spirv_recompiler_impl.cpp

@@ -52,6 +52,7 @@ int SpirvShaderRecompilerImpl(int argc, char *argv[]) {
     host_info.support_geometry_shader_passthrough = true;
     host_info.support_conditional_barrier = true;
     host_info.min_ssbo_alignment = 0;
+    host_info.ApplyDescriptorLimitPolicy();
     auto program = Shader::Maxwell::TranslateProgram(inst_pool, block_pool, env, cfg, host_info);
 
     // IR::Program TranslateProgram(ObjectPool<IR::Inst>& inst_pool, ObjectPool<IR::Block>& block_pool,