ShaderDecompiler: Add a debug option to dump the game's shaders.

core: hle: kernel: Implement thread pinning.

CMakeLists.txt

@@ -229,7 +229,7 @@ elseif (${CMAKE_SYSTEM_NAME} STREQUAL "Linux" OR YUZU_USE_BUNDLED_BOOST)
     include_directories(SYSTEM "${Boost_INCLUDE_DIRS}")
 else()
     message(STATUS "Boost 1.73.0 or newer not found, falling back to Conan")
-    list(APPEND CONAN_REQUIRED_LIBS "boost/1.73.0")
+    list(APPEND CONAN_REQUIRED_LIBS "boost/1.78.0")
 endif()
 
 # Attempt to locate any packages that are required and report the missing ones in CONAN_REQUIRED_LIBS
@@ -249,7 +249,7 @@ if(ENABLE_QT)
     # Check for system Qt on Linux, fallback to bundled Qt
     if (${CMAKE_SYSTEM_NAME} STREQUAL "Linux")
         if (NOT YUZU_USE_BUNDLED_QT)
-            find_package(Qt5 ${QT_VERSION} COMPONENTS Widgets)
+            find_package(Qt5 ${QT_VERSION} COMPONENTS Widgets DBus)
         endif()
         if (NOT Qt5_FOUND OR YUZU_USE_BUNDLED_QT)
             # Check for dependencies, then enable bundled Qt download
@@ -508,7 +508,7 @@ set(FFmpeg_COMPONENTS
     avutil
     swscale)
 
-if (${CMAKE_SYSTEM_NAME} STREQUAL "Linux")
+if (UNIX AND NOT APPLE)
     Include(FindPkgConfig REQUIRED)
     pkg_check_modules(LIBVA libva)
 endif()

CMakeModules/GenerateSCMRev.cmake

@@ -11,9 +11,15 @@ find_package(Git QUIET PATHS "${GIT_EXECUTABLE}")
 
 # generate git/build information
 include(GetGitRevisionDescription)
-get_git_head_revision(GIT_REF_SPEC GIT_REV)
-git_describe(GIT_DESC --always --long --dirty)
-git_branch_name(GIT_BRANCH)
+if(NOT GIT_REF_SPEC)
+    get_git_head_revision(GIT_REF_SPEC GIT_REV)
+endif()
+if(NOT GIT_DESC)
+    git_describe(GIT_DESC --always --long --dirty)
+endif()
+if (NOT GIT_BRANCH)
+  git_branch_name(GIT_BRANCH)
+endif()
 get_timestamp(BUILD_DATE)
 
 # Generate cpp with Git revision from template

dist/yuzu.desktop

@@ -8,5 +8,5 @@ Icon=yuzu
 TryExec=yuzu
 Exec=yuzu %f
 Categories=Game;Emulator;Qt;
-MimeType=application/x-nx-nro;application/x-nx-nso;
+MimeType=application/x-nx-nro;application/x-nx-nso;application/x-nx-nsp;application/x-nx-xci;
 Keywords=Switch;Nintendo;

dist/yuzu.xml

@@ -15,4 +15,19 @@
         <glob pattern="*.nso"/>
         <magic><match value="NSO" type="string" offset="0"/></magic>
     </mime-type>
+
+    <mime-type type="application/x-nx-nsp">
+        <comment>Nintendo Switch Package</comment>
+        <acronym>NSP</acronym>
+        <icon name="yuzu"/>
+        <glob pattern="*.nsp"/>
+        <magic><match value="PFS" type="string" offset="0"/></magic>
+    </mime-type>
+
+    <mime-type type="application/x-nx-xci">
+        <comment>Nintendo Switch Card Image</comment>
+        <acronym>XCI</acronym>
+        <icon name="yuzu"/>
+        <glob pattern="*.xci"/>
+    </mime-type>
 </mime-info>

externals/CMakeLists.txt

@@ -52,11 +52,12 @@ endif()
 # SDL2
 if (YUZU_USE_EXTERNAL_SDL2)
     if (NOT WIN32)
-        # Yuzu itself needs: Events Joystick Haptic Sensor Timers Audio
+        # Yuzu itself needs: Atomic Audio Events Joystick Haptic Sensor Threads Timers
+        # Since 2.0.18 Atomic+Threads required for HIDAPI/libusb (see https://github.com/libsdl-org/SDL/issues/5095)
         # Yuzu-cmd also needs: Video (depends on Loadso/Dlopen)
         set(SDL_UNUSED_SUBSYSTEMS
-            Atomic Render Power Threads
-            File CPUinfo Filesystem Locale)
+            CPUinfo File Filesystem
+            Locale Power Render)
         foreach(_SUB ${SDL_UNUSED_SUBSYSTEMS})
           string(TOUPPER ${_SUB} _OPT)
           option(SDL_${_OPT} "" OFF)

src/common/CMakeLists.txt

@@ -22,6 +22,11 @@ add_custom_command(OUTPUT scm_rev.cpp
       -DTITLE_BAR_FORMAT_RUNNING=${TITLE_BAR_FORMAT_RUNNING}
       -DBUILD_TAG=${BUILD_TAG}
       -DBUILD_ID=${DISPLAY_VERSION}
+      -DGIT_REF_SPEC=${GIT_REF_SPEC}
+      -DGIT_REV=${GIT_REV}
+      -DGIT_DESC=${GIT_DESC}
+      -DGIT_BRANCH=${GIT_BRANCH}
+      -DBUILD_FULLNAME=${BUILD_FULLNAME}
       -DGIT_EXECUTABLE=${GIT_EXECUTABLE}
       -P ${CMAKE_SOURCE_DIR}/CMakeModules/GenerateSCMRev.cmake
     DEPENDS

src/common/settings.h

@@ -597,6 +597,7 @@ struct Values {
     BasicSetting<std::string> program_args{std::string(), "program_args"};
     BasicSetting<bool> dump_exefs{false, "dump_exefs"};
     BasicSetting<bool> dump_nso{false, "dump_nso"};
+    BasicSetting<bool> dump_shaders{false, "dump_shaders"};
     BasicSetting<bool> enable_fs_access_log{false, "enable_fs_access_log"};
     BasicSetting<bool> reporting_services{false, "reporting_services"};
     BasicSetting<bool> quest_flag{false, "quest_flag"};

src/core/CMakeLists.txt

@@ -187,6 +187,8 @@ add_library(core STATIC
     hle/kernel/k_event.h
     hle/kernel/k_handle_table.cpp
     hle/kernel/k_handle_table.h
+    hle/kernel/k_interrupt_manager.cpp
+    hle/kernel/k_interrupt_manager.h
     hle/kernel/k_light_condition_variable.cpp
     hle/kernel/k_light_condition_variable.h
     hle/kernel/k_light_lock.cpp
@@ -265,8 +267,6 @@ add_library(core STATIC
     hle/kernel/svc_wrap.h
     hle/kernel/time_manager.cpp
     hle/kernel/time_manager.h
-    hle/lock.cpp
-    hle/lock.h
     hle/result.h
     hle/service/acc/acc.cpp
     hle/service/acc/acc.h

src/core/frontend/applets/controller.cpp

@@ -45,26 +45,26 @@ void DefaultControllerApplet::ReconfigureControllers(std::function<void()> callb
         // Pro Controller -> Dual Joycons -> Left Joycon/Right Joycon -> Handheld
         if (parameters.allow_pro_controller) {
             controller->SetNpadStyleIndex(Core::HID::NpadStyleIndex::ProController);
-            controller->Connect();
+            controller->Connect(true);
         } else if (parameters.allow_dual_joycons) {
             controller->SetNpadStyleIndex(Core::HID::NpadStyleIndex::JoyconDual);
-            controller->Connect();
+            controller->Connect(true);
         } else if (parameters.allow_left_joycon && parameters.allow_right_joycon) {
             // Assign left joycons to even player indices and right joycons to odd player indices.
             // We do this since Captain Toad Treasure Tracker expects a left joycon for Player 1 and
             // a right Joycon for Player 2 in 2 Player Assist mode.
             if (index % 2 == 0) {
                 controller->SetNpadStyleIndex(Core::HID::NpadStyleIndex::JoyconLeft);
-                controller->Connect();
+                controller->Connect(true);
             } else {
                 controller->SetNpadStyleIndex(Core::HID::NpadStyleIndex::JoyconRight);
-                controller->Connect();
+                controller->Connect(true);
             }
         } else if (index == 0 && parameters.enable_single_mode && parameters.allow_handheld &&
                    !Settings::values.use_docked_mode.GetValue()) {
             // We should *never* reach here under any normal circumstances.
             controller->SetNpadStyleIndex(Core::HID::NpadStyleIndex::Handheld);
-            controller->Connect();
+            controller->Connect(true);
         } else {
             UNREACHABLE_MSG("Unable to add a new controller based on the given parameters!");
         }

src/core/hid/emulated_console.cpp

@@ -161,7 +161,10 @@ void EmulatedConsole::SetMotion(const Common::Input::CallbackStatus& callback) {
     motion.rotation = emulated.GetGyroscope();
     motion.orientation = emulated.GetOrientation();
     motion.quaternion = emulated.GetQuaternion();
+    motion.gyro_bias = emulated.GetGyroBias();
     motion.is_at_rest = !emulated.IsMoving(motion_sensitivity);
+    // Find what is this value
+    motion.verticalization_error = 0.0f;
 
     TriggerOnChange(ConsoleTriggerType::Motion);
 }

src/core/hid/emulated_console.h

@@ -50,6 +50,8 @@ struct ConsoleMotion {
     Common::Vec3f rotation{};
     std::array<Common::Vec3f, 3> orientation{};
     Common::Quaternion<f32> quaternion{};
+    Common::Vec3f gyro_bias{};
+    f32 verticalization_error{};
     bool is_at_rest{};
 };
 

src/core/hid/emulated_controller.cpp

@@ -843,23 +843,18 @@ bool EmulatedController::SetVibration(std::size_t device_index, VibrationValue v
 }
 
 bool EmulatedController::TestVibration(std::size_t device_index) {
-    if (device_index >= output_devices.size()) {
-        return false;
-    }
-    if (!output_devices[device_index]) {
-        return false;
-    }
-
-    // Send a slight vibration to test for rumble support
-    constexpr Common::Input::VibrationStatus status = {
+    static constexpr VibrationValue test_vibration = {
         .low_amplitude = 0.001f,
         .low_frequency = 160.0f,
         .high_amplitude = 0.001f,
         .high_frequency = 320.0f,
-        .type = Common::Input::VibrationAmplificationType::Linear,
     };
-    return output_devices[device_index]->SetVibration(status) ==
-           Common::Input::VibrationError::None;
+
+    // Send a slight vibration to test for rumble support
+    SetVibration(device_index, test_vibration);
+
+    // Stop any vibration and return the result
+    return SetVibration(device_index, DEFAULT_VIBRATION_VALUE);
 }
 
 void EmulatedController::SetLedPattern() {
@@ -891,8 +886,9 @@ void EmulatedController::SetSupportedNpadStyleTag(NpadStyleTag supported_styles)
     }
 }
 
-bool EmulatedController::IsControllerSupported() const {
-    switch (npad_type) {
+bool EmulatedController::IsControllerSupported(bool use_temporary_value) const {
+    const auto type = is_configuring && use_temporary_value ? tmp_npad_type : npad_type;
+    switch (type) {
     case NpadStyleIndex::ProController:
         return supported_style_tag.fullkey;
     case NpadStyleIndex::Handheld:
@@ -920,9 +916,10 @@ bool EmulatedController::IsControllerSupported() const {
     }
 }
 
-void EmulatedController::Connect() {
-    if (!IsControllerSupported()) {
-        LOG_ERROR(Service_HID, "Controller type {} is not supported", npad_type);
+void EmulatedController::Connect(bool use_temporary_value) {
+    if (!IsControllerSupported(use_temporary_value)) {
+        const auto type = is_configuring && use_temporary_value ? tmp_npad_type : npad_type;
+        LOG_ERROR(Service_HID, "Controller type {} is not supported", type);
         return;
     }
     {

src/core/hid/emulated_controller.h

@@ -167,8 +167,11 @@ public:
      */
     void SetSupportedNpadStyleTag(NpadStyleTag supported_styles);
 
-    /// Sets the connected status to true
-    void Connect();
+    /**
+     * Sets the connected status to true
+     * @param use_temporary_value If true tmp_npad_type will be used
+     */
+    void Connect(bool use_temporary_value = false);
 
     /// Sets the connected status to false
     void Disconnect();
@@ -319,9 +322,10 @@ private:
 
     /**
      * Checks the current controller type against the supported_style_tag
+     * @param use_temporary_value If true tmp_npad_type will be used
      * @return true if the controller is supported
      */
-    bool IsControllerSupported() const;
+    bool IsControllerSupported(bool use_temporary_value = false) const;
 
     /**
      * Updates the button status of the controller

src/core/hid/hid_types.h

@@ -496,6 +496,13 @@ struct VibrationValue {
 };
 static_assert(sizeof(VibrationValue) == 0x10, "VibrationValue has incorrect size.");
 
+constexpr VibrationValue DEFAULT_VIBRATION_VALUE{
+    .low_amplitude = 0.0f,
+    .low_frequency = 160.0f,
+    .high_amplitude = 0.0f,
+    .high_frequency = 320.0f,
+};
+
 // This is nn::hid::VibrationDeviceInfo
 struct VibrationDeviceInfo {
     VibrationDeviceType type{};

src/core/hid/motion_input.cpp

@@ -23,11 +23,11 @@ void MotionInput::SetAcceleration(const Common::Vec3f& acceleration) {
 }
 
 void MotionInput::SetGyroscope(const Common::Vec3f& gyroscope) {
-    gyro = gyroscope - gyro_drift;
+    gyro = gyroscope - gyro_bias;
 
     // Auto adjust drift to minimize drift
     if (!IsMoving(0.1f)) {
-        gyro_drift = (gyro_drift * 0.9999f) + (gyroscope * 0.0001f);
+        gyro_bias = (gyro_bias * 0.9999f) + (gyroscope * 0.0001f);
     }
 
     if (gyro.Length2() < gyro_threshold) {
@@ -41,8 +41,8 @@ void MotionInput::SetQuaternion(const Common::Quaternion<f32>& quaternion) {
     quat = quaternion;
 }
 
-void MotionInput::SetGyroDrift(const Common::Vec3f& drift) {
-    gyro_drift = drift;
+void MotionInput::SetGyroBias(const Common::Vec3f& bias) {
+    gyro_bias = bias;
 }
 
 void MotionInput::SetGyroThreshold(f32 threshold) {
@@ -192,6 +192,10 @@ Common::Vec3f MotionInput::GetGyroscope() const {
     return gyro;
 }
 
+Common::Vec3f MotionInput::GetGyroBias() const {
+    return gyro_bias;
+}
+
 Common::Quaternion<f32> MotionInput::GetQuaternion() const {
     return quat;
 }

src/core/hid/motion_input.h

@@ -24,7 +24,7 @@ public:
     void SetAcceleration(const Common::Vec3f& acceleration);
     void SetGyroscope(const Common::Vec3f& gyroscope);
     void SetQuaternion(const Common::Quaternion<f32>& quaternion);
-    void SetGyroDrift(const Common::Vec3f& drift);
+    void SetGyroBias(const Common::Vec3f& bias);
     void SetGyroThreshold(f32 threshold);
 
     void EnableReset(bool reset);
@@ -36,6 +36,7 @@ public:
     [[nodiscard]] std::array<Common::Vec3f, 3> GetOrientation() const;
     [[nodiscard]] Common::Vec3f GetAcceleration() const;
     [[nodiscard]] Common::Vec3f GetGyroscope() const;
+    [[nodiscard]] Common::Vec3f GetGyroBias() const;
     [[nodiscard]] Common::Vec3f GetRotations() const;
     [[nodiscard]] Common::Quaternion<f32> GetQuaternion() const;
 
@@ -69,7 +70,7 @@ private:
     Common::Vec3f gyro;
 
     // Vector to be substracted from gyro measurements
-    Common::Vec3f gyro_drift;
+    Common::Vec3f gyro_bias;
 
     // Minimum gyro amplitude to detect if the device is moving
     f32 gyro_threshold = 0.0f;

src/core/hle/kernel/global_scheduler_context.cpp

@@ -9,6 +9,7 @@
 #include "core/hle/kernel/global_scheduler_context.h"
 #include "core/hle/kernel/k_scheduler.h"
 #include "core/hle/kernel/kernel.h"
+#include "core/hle/kernel/physical_core.h"
 
 namespace Kernel {
 
@@ -42,6 +43,11 @@ void GlobalSchedulerContext::PreemptThreads() {
     for (u32 core_id = 0; core_id < Core::Hardware::NUM_CPU_CORES; core_id++) {
         const u32 priority = preemption_priorities[core_id];
         kernel.Scheduler(core_id).RotateScheduledQueue(core_id, priority);
+
+        // Signal an interrupt occurred. For core 3, this is a certainty, as preemption will result
+        // in the rotator thread being scheduled. For cores 0-2, this is to simulate or system
+        // interrupts that may have occurred.
+        kernel.PhysicalCore(core_id).Interrupt();
     }
 }
 

src/core/hle/kernel/k_interrupt_manager.cpp

@@ -0,0 +1,34 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "core/hle/kernel/k_interrupt_manager.h"
+#include "core/hle/kernel/k_process.h"
+#include "core/hle/kernel/k_scheduler.h"
+#include "core/hle/kernel/k_thread.h"
+#include "core/hle/kernel/kernel.h"
+
+namespace Kernel::KInterruptManager {
+
+void HandleInterrupt(KernelCore& kernel, s32 core_id) {
+    auto* process = kernel.CurrentProcess();
+    if (!process) {
+        return;
+    }
+
+    auto& scheduler = kernel.Scheduler(core_id);
+    auto& current_thread = *scheduler.GetCurrentThread();
+
+    // If the user disable count is set, we may need to pin the current thread.
+    if (current_thread.GetUserDisableCount() && !process->GetPinnedThread(core_id)) {
+        KScopedSchedulerLock sl{kernel};
+
+        // Pin the current thread.
+        process->PinCurrentThread(core_id);
+
+        // Set the interrupt flag for the thread.
+        scheduler.GetCurrentThread()->SetInterruptFlag();
+    }
+}
+
+} // namespace Kernel::KInterruptManager

src/core/hle/kernel/k_interrupt_manager.h

@@ -0,0 +1,17 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/common_types.h"
+
+namespace Kernel {
+
+class KernelCore;
+
+namespace KInterruptManager {
+void HandleInterrupt(KernelCore& kernel, s32 core_id);
+}
+
+} // namespace Kernel

src/core/hle/kernel/k_memory_block.h

@@ -120,7 +120,7 @@ static_assert(static_cast<u32>(KMemoryState::CodeOut) == 0x00402015);
 
 enum class KMemoryPermission : u8 {
     None = 0,
-    Mask = static_cast<u8>(~None),
+    All = static_cast<u8>(~None),
 
     Read = 1 << 0,
     Write = 1 << 1,

src/core/hle/kernel/k_page_table.cpp

@@ -264,9 +264,9 @@ ResultCode KPageTable::InitializeForProcess(FileSys::ProgramAddressSpaceType as_
     ASSERT(heap_last < stack_start || stack_last < heap_start);
     ASSERT(heap_last < kmap_start || kmap_last < heap_start);
 
-    current_heap_addr = heap_region_start;
-    heap_capacity = 0;
-    physical_memory_usage = 0;
+    current_heap_end = heap_region_start;
+    max_heap_size = 0;
+    mapped_physical_memory_size = 0;
     memory_pool = pool;
 
     page_table_impl.Resize(address_space_width, PageBits);
@@ -306,7 +306,7 @@ ResultCode KPageTable::MapProcessCodeMemory(VAddr dst_addr, VAddr src_addr, std:
     KMemoryState state{};
     KMemoryPermission perm{};
     CASCADE_CODE(CheckMemoryState(&state, &perm, nullptr, src_addr, size, KMemoryState::All,
-                                  KMemoryState::Normal, KMemoryPermission::Mask,
+                                  KMemoryState::Normal, KMemoryPermission::All,
                                   KMemoryPermission::ReadAndWrite, KMemoryAttribute::Mask,
                                   KMemoryAttribute::None, KMemoryAttribute::IpcAndDeviceMapped));
 
@@ -465,7 +465,7 @@ ResultCode KPageTable::MapPhysicalMemory(VAddr addr, std::size_t size) {
 
     MapPhysicalMemory(page_linked_list, addr, end_addr);
 
-    physical_memory_usage += remaining_size;
+    mapped_physical_memory_size += remaining_size;
 
     const std::size_t num_pages{size / PageSize};
     block_manager->Update(addr, num_pages, KMemoryState::Free, KMemoryPermission::None,
@@ -507,7 +507,7 @@ ResultCode KPageTable::UnmapPhysicalMemory(VAddr addr, std::size_t size) {
 
     auto process{system.Kernel().CurrentProcess()};
     process->GetResourceLimit()->Release(LimitableResource::PhysicalMemory, mapped_size);
-    physical_memory_usage -= mapped_size;
+    mapped_physical_memory_size -= mapped_size;
 
     return ResultSuccess;
 }
@@ -554,7 +554,7 @@ ResultCode KPageTable::Map(VAddr dst_addr, VAddr src_addr, std::size_t size) {
     KMemoryState src_state{};
     CASCADE_CODE(CheckMemoryState(
         &src_state, nullptr, nullptr, src_addr, size, KMemoryState::FlagCanAlias,
-        KMemoryState::FlagCanAlias, KMemoryPermission::Mask, KMemoryPermission::ReadAndWrite,
+        KMemoryState::FlagCanAlias, KMemoryPermission::All, KMemoryPermission::ReadAndWrite,
         KMemoryAttribute::Mask, KMemoryAttribute::None, KMemoryAttribute::IpcAndDeviceMapped));
 
     if (IsRegionMapped(dst_addr, size)) {
@@ -593,7 +593,7 @@ ResultCode KPageTable::Unmap(VAddr dst_addr, VAddr src_addr, std::size_t size) {
     KMemoryState src_state{};
     CASCADE_CODE(CheckMemoryState(
         &src_state, nullptr, nullptr, src_addr, size, KMemoryState::FlagCanAlias,
-        KMemoryState::FlagCanAlias, KMemoryPermission::Mask, KMemoryPermission::None,
+        KMemoryState::FlagCanAlias, KMemoryPermission::All, KMemoryPermission::None,
         KMemoryAttribute::Mask, KMemoryAttribute::Locked, KMemoryAttribute::IpcAndDeviceMapped));
 
     KMemoryPermission dst_perm{};
@@ -784,7 +784,7 @@ ResultCode KPageTable::ReserveTransferMemory(VAddr addr, std::size_t size, KMemo
     CASCADE_CODE(CheckMemoryState(
         &state, nullptr, &attribute, addr, size,
         KMemoryState::FlagCanTransfer | KMemoryState::FlagReferenceCounted,
-        KMemoryState::FlagCanTransfer | KMemoryState::FlagReferenceCounted, KMemoryPermission::Mask,
+        KMemoryState::FlagCanTransfer | KMemoryState::FlagReferenceCounted, KMemoryPermission::All,
         KMemoryPermission::ReadAndWrite, KMemoryAttribute::Mask, KMemoryAttribute::None,
         KMemoryAttribute::IpcAndDeviceMapped));
 
@@ -806,6 +806,33 @@ ResultCode KPageTable::ResetTransferMemory(VAddr addr, std::size_t size) {
                          KMemoryAttribute::Locked, KMemoryAttribute::IpcAndDeviceMapped));
 
     block_manager->Update(addr, size / PageSize, state, KMemoryPermission::ReadAndWrite);
+    return ResultSuccess;
+}
+
+ResultCode KPageTable::SetMemoryPermission(VAddr addr, std::size_t size,
+                                           Svc::MemoryPermission svc_perm) {
+    const size_t num_pages = size / PageSize;
+
+    // Lock the table.
+    std::lock_guard lock{page_table_lock};
+
+    // Verify we can change the memory permission.
+    KMemoryState old_state;
+    KMemoryPermission old_perm;
+    R_TRY(this->CheckMemoryState(
+        std::addressof(old_state), std::addressof(old_perm), nullptr, addr, size,
+        KMemoryState::FlagCanReprotect, KMemoryState::FlagCanReprotect, KMemoryPermission::None,
+        KMemoryPermission::None, KMemoryAttribute::All, KMemoryAttribute::None));
+
+    // Determine new perm.
+    const KMemoryPermission new_perm = ConvertToKMemoryPermission(svc_perm);
+    R_SUCCEED_IF(old_perm == new_perm);
+
+    // Perform mapping operation.
+    R_TRY(Operate(addr, num_pages, new_perm, OperationType::ChangePermissions));
+
+    // Update the blocks.
+    block_manager->Update(addr, num_pages, old_state, new_perm, KMemoryAttribute::None);
 
     return ResultSuccess;
 }
@@ -832,61 +859,125 @@ ResultCode KPageTable::SetMemoryAttribute(VAddr addr, std::size_t size, KMemoryA
     return ResultSuccess;
 }
 
-ResultCode KPageTable::SetHeapCapacity(std::size_t new_heap_capacity) {
+ResultCode KPageTable::SetMaxHeapSize(std::size_t size) {
+    // Lock the table.
     std::lock_guard lock{page_table_lock};
-    heap_capacity = new_heap_capacity;
+
+    // Only process page tables are allowed to set heap size.
+    ASSERT(!this->IsKernel());
+
+    max_heap_size = size;
+
     return ResultSuccess;
 }
 
-ResultVal<VAddr> KPageTable::SetHeapSize(std::size_t size) {
+ResultCode KPageTable::SetHeapSize(VAddr* out, std::size_t size) {
+    // Try to perform a reduction in heap, instead of an extension.
+    VAddr cur_address{};
+    std::size_t allocation_size{};
+    {
+        // Lock the table.
+        std::lock_guard lk(page_table_lock);
 
-    if (size > heap_region_end - heap_region_start) {
-        return ResultOutOfMemory;
+        // Validate that setting heap size is possible at all.
+        R_UNLESS(!is_kernel, ResultOutOfMemory);
+        R_UNLESS(size <= static_cast<std::size_t>(heap_region_end - heap_region_start),
+                 ResultOutOfMemory);
+        R_UNLESS(size <= max_heap_size, ResultOutOfMemory);
+
+        if (size < GetHeapSize()) {
+            // The size being requested is less than the current size, so we need to free the end of
+            // the heap.
+
+            // Validate memory state.
+            std::size_t num_allocator_blocks;
+            R_TRY(this->CheckMemoryState(std::addressof(num_allocator_blocks),
+                                         heap_region_start + size, GetHeapSize() - size,
+                                         KMemoryState::All, KMemoryState::Normal,
+                                         KMemoryPermission::All, KMemoryPermission::ReadAndWrite,
+                                         KMemoryAttribute::All, KMemoryAttribute::None));
+
+            // Unmap the end of the heap.
+            const auto num_pages = (GetHeapSize() - size) / PageSize;
+            R_TRY(Operate(heap_region_start + size, num_pages, KMemoryPermission::None,
+                          OperationType::Unmap));
+
+            // Release the memory from the resource limit.
+            system.Kernel().CurrentProcess()->GetResourceLimit()->Release(
+                LimitableResource::PhysicalMemory, num_pages * PageSize);
+
+            // Apply the memory block update.
+            block_manager->Update(heap_region_start + size, num_pages, KMemoryState::Free,
+                                  KMemoryPermission::None, KMemoryAttribute::None);
+
+            // Update the current heap end.
+            current_heap_end = heap_region_start + size;
+
+            // Set the output.
+            *out = heap_region_start;
+            return ResultSuccess;
+        } else if (size == GetHeapSize()) {
+            // The size requested is exactly the current size.
+            *out = heap_region_start;
+            return ResultSuccess;
+        } else {
+            // We have to allocate memory. Determine how much to allocate and where while the table
+            // is locked.
+            cur_address = current_heap_end;
+            allocation_size = size - GetHeapSize();
+        }
     }
 
-    const u64 previous_heap_size{GetHeapSize()};
+    // Reserve memory for the heap extension.
+    KScopedResourceReservation memory_reservation(
+        system.Kernel().CurrentProcess()->GetResourceLimit(), LimitableResource::PhysicalMemory,
+        allocation_size);
+    R_UNLESS(memory_reservation.Succeeded(), ResultLimitReached);
 
-    UNIMPLEMENTED_IF_MSG(previous_heap_size > size, "Heap shrink is unimplemented");
+    // Allocate pages for the heap extension.
+    KPageLinkedList page_linked_list;
+    R_TRY(system.Kernel().MemoryManager().Allocate(page_linked_list, allocation_size / PageSize,
+                                                   memory_pool));
 
-    // Increase the heap size
+    // Map the pages.
     {
-        std::lock_guard lock{page_table_lock};
+        // Lock the table.
+        std::lock_guard lk(page_table_lock);
 
-        const u64 delta{size - previous_heap_size};
+        // Ensure that the heap hasn't changed since we began executing.
+        ASSERT(cur_address == current_heap_end);
 
-        // Reserve memory for the heap extension.
-        KScopedResourceReservation memory_reservation(
-            system.Kernel().CurrentProcess()->GetResourceLimit(), LimitableResource::PhysicalMemory,
-            delta);
+        // Check the memory state.
+        std::size_t num_allocator_blocks{};
+        R_TRY(this->CheckMemoryState(std::addressof(num_allocator_blocks), current_heap_end,
+                                     allocation_size, KMemoryState::All, KMemoryState::Free,
+                                     KMemoryPermission::None, KMemoryPermission::None,
+                                     KMemoryAttribute::None, KMemoryAttribute::None));
 
-        if (!memory_reservation.Succeeded()) {
-            LOG_ERROR(Kernel, "Could not reserve heap extension of size {:X} bytes", delta);
-            return ResultLimitReached;
+        // Map the pages.
+        const auto num_pages = allocation_size / PageSize;
+        R_TRY(Operate(current_heap_end, num_pages, page_linked_list, OperationType::MapGroup));
+
+        // Clear all the newly allocated pages.
+        for (std::size_t cur_page = 0; cur_page < num_pages; ++cur_page) {
+            std::memset(system.Memory().GetPointer(current_heap_end + (cur_page * PageSize)), 0,
+                        PageSize);
         }
 
-        KPageLinkedList page_linked_list;
-        const std::size_t num_pages{delta / PageSize};
-
-        CASCADE_CODE(
-            system.Kernel().MemoryManager().Allocate(page_linked_list, num_pages, memory_pool));
-
-        if (IsRegionMapped(current_heap_addr, delta)) {
-            return ResultInvalidCurrentMemory;
-        }
-
-        CASCADE_CODE(
-            Operate(current_heap_addr, num_pages, page_linked_list, OperationType::MapGroup));
-
-        // Succeeded in allocation, commit the resource reservation
+        // We succeeded, so commit our memory reservation.
         memory_reservation.Commit();
 
-        block_manager->Update(current_heap_addr, num_pages, KMemoryState::Normal,
-                              KMemoryPermission::ReadAndWrite);
+        // Apply the memory block update.
+        block_manager->Update(current_heap_end, num_pages, KMemoryState::Normal,
+                              KMemoryPermission::ReadAndWrite, KMemoryAttribute::None);
 
-        current_heap_addr = heap_region_start + size;
+        // Update the current heap end.
+        current_heap_end = heap_region_start + size;
+
+        // Set the output.
+        *out = heap_region_start;
+        return ResultSuccess;
     }
-
-    return heap_region_start;
 }
 
 ResultVal<VAddr> KPageTable::AllocateAndMapMemory(std::size_t needed_num_pages, std::size_t align,
@@ -978,7 +1069,7 @@ ResultCode KPageTable::LockForCodeMemory(VAddr addr, std::size_t size) {
 
     if (const ResultCode result{CheckMemoryState(
             nullptr, &old_perm, nullptr, addr, size, KMemoryState::FlagCanCodeMemory,
-            KMemoryState::FlagCanCodeMemory, KMemoryPermission::Mask,
+            KMemoryState::FlagCanCodeMemory, KMemoryPermission::All,
             KMemoryPermission::UserReadWrite, KMemoryAttribute::All, KMemoryAttribute::None)};
         result.IsError()) {
         return result;
@@ -1031,9 +1122,8 @@ ResultCode KPageTable::InitializeMemoryLayout(VAddr start, VAddr end) {
 
 bool KPageTable::IsRegionMapped(VAddr address, u64 size) {
     return CheckMemoryState(address, size, KMemoryState::All, KMemoryState::Free,
-                            KMemoryPermission::Mask, KMemoryPermission::None,
-                            KMemoryAttribute::Mask, KMemoryAttribute::None,
-                            KMemoryAttribute::IpcAndDeviceMapped)
+                            KMemoryPermission::All, KMemoryPermission::None, KMemoryAttribute::Mask,
+                            KMemoryAttribute::None, KMemoryAttribute::IpcAndDeviceMapped)
         .IsError();
 }
 

src/core/hle/kernel/k_page_table.h

@@ -47,10 +47,11 @@ public:
     KMemoryInfo QueryInfo(VAddr addr);
     ResultCode ReserveTransferMemory(VAddr addr, std::size_t size, KMemoryPermission perm);
     ResultCode ResetTransferMemory(VAddr addr, std::size_t size);
+    ResultCode SetMemoryPermission(VAddr addr, std::size_t size, Svc::MemoryPermission perm);
     ResultCode SetMemoryAttribute(VAddr addr, std::size_t size, KMemoryAttribute mask,
                                   KMemoryAttribute value);
-    ResultCode SetHeapCapacity(std::size_t new_heap_capacity);
-    ResultVal<VAddr> SetHeapSize(std::size_t size);
+    ResultCode SetMaxHeapSize(std::size_t size);
+    ResultCode SetHeapSize(VAddr* out, std::size_t size);
     ResultVal<VAddr> AllocateAndMapMemory(std::size_t needed_num_pages, std::size_t align,
                                           bool is_map_only, VAddr region_start,
                                           std::size_t region_num_pages, KMemoryState state,
@@ -182,14 +183,15 @@ public:
     constexpr VAddr GetAliasCodeRegionSize() const {
         return alias_code_region_end - alias_code_region_start;
     }
+    size_t GetNormalMemorySize() {
+        std::lock_guard lk(page_table_lock);
+        return GetHeapSize() + mapped_physical_memory_size;
+    }
     constexpr std::size_t GetAddressSpaceWidth() const {
         return address_space_width;
     }
-    constexpr std::size_t GetHeapSize() {
-        return current_heap_addr - heap_region_start;
-    }
-    constexpr std::size_t GetTotalHeapSize() {
-        return GetHeapSize() + physical_memory_usage;
+    constexpr std::size_t GetHeapSize() const {
+        return current_heap_end - heap_region_start;
     }
     constexpr bool IsInsideAddressSpace(VAddr address, std::size_t size) const {
         return address_space_start <= address && address + size - 1 <= address_space_end - 1;
@@ -269,10 +271,8 @@ private:
     VAddr code_region_end{};
     VAddr alias_code_region_start{};
     VAddr alias_code_region_end{};
-    VAddr current_heap_addr{};
 
-    std::size_t heap_capacity{};
-    std::size_t physical_memory_usage{};
+    std::size_t mapped_physical_memory_size{};
     std::size_t max_heap_size{};
     std::size_t max_physical_memory_size{};
     std::size_t address_space_width{};

src/core/hle/kernel/k_process.cpp

@@ -28,7 +28,6 @@
 #include "core/hle/kernel/k_thread.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/svc_results.h"
-#include "core/hle/lock.h"
 #include "core/memory.h"
 
 namespace Kernel {
@@ -173,7 +172,7 @@ void KProcess::DecrementThreadCount() {
 
 u64 KProcess::GetTotalPhysicalMemoryAvailable() const {
     const u64 capacity{resource_limit->GetFreeValue(LimitableResource::PhysicalMemory) +
-                       page_table->GetTotalHeapSize() + GetSystemResourceSize() + image_size +
+                       page_table->GetNormalMemorySize() + GetSystemResourceSize() + image_size +
                        main_thread_stack_size};
     if (const auto pool_size = kernel.MemoryManager().GetSize(KMemoryManager::Pool::Application);
         capacity != pool_size) {
@@ -190,7 +189,7 @@ u64 KProcess::GetTotalPhysicalMemoryAvailableWithoutSystemResource() const {
 }
 
 u64 KProcess::GetTotalPhysicalMemoryUsed() const {
-    return image_size + main_thread_stack_size + page_table->GetTotalHeapSize() +
+    return image_size + main_thread_stack_size + page_table->GetNormalMemorySize() +
            GetSystemResourceSize();
 }
 
@@ -221,30 +220,28 @@ bool KProcess::ReleaseUserException(KThread* thread) {
     }
 }
 
-void KProcess::PinCurrentThread() {
+void KProcess::PinCurrentThread(s32 core_id) {
     ASSERT(kernel.GlobalSchedulerContext().IsLocked());
 
     // Get the current thread.
-    const s32 core_id = GetCurrentCoreId(kernel);
-    KThread* cur_thread = GetCurrentThreadPointer(kernel);
+    KThread* cur_thread = kernel.Scheduler(static_cast<std::size_t>(core_id)).GetCurrentThread();
 
     // If the thread isn't terminated, pin it.
     if (!cur_thread->IsTerminationRequested()) {
         // Pin it.
         PinThread(core_id, cur_thread);
-        cur_thread->Pin();
+        cur_thread->Pin(core_id);
 
         // An update is needed.
         KScheduler::SetSchedulerUpdateNeeded(kernel);
     }
 }
 
-void KProcess::UnpinCurrentThread() {
+void KProcess::UnpinCurrentThread(s32 core_id) {
     ASSERT(kernel.GlobalSchedulerContext().IsLocked());
 
     // Get the current thread.
-    const s32 core_id = GetCurrentCoreId(kernel);
-    KThread* cur_thread = GetCurrentThreadPointer(kernel);
+    KThread* cur_thread = kernel.Scheduler(static_cast<std::size_t>(core_id)).GetCurrentThread();
 
     // Unpin it.
     cur_thread->Unpin();
@@ -411,8 +408,8 @@ void KProcess::Run(s32 main_thread_priority, u64 stack_size) {
     resource_limit->Reserve(LimitableResource::Threads, 1);
     resource_limit->Reserve(LimitableResource::PhysicalMemory, main_thread_stack_size);
 
-    const std::size_t heap_capacity{memory_usage_capacity - main_thread_stack_size - image_size};
-    ASSERT(!page_table->SetHeapCapacity(heap_capacity).IsError());
+    const std::size_t heap_capacity{memory_usage_capacity - (main_thread_stack_size + image_size)};
+    ASSERT(!page_table->SetMaxHeapSize(heap_capacity).IsError());
 
     ChangeStatus(ProcessStatus::Running);
 
@@ -543,7 +540,6 @@ void KProcess::FreeTLSRegion(VAddr tls_address) {
 }
 
 void KProcess::LoadModule(CodeSet code_set, VAddr base_addr) {
-    std::lock_guard lock{HLE::g_hle_lock};
     const auto ReprotectSegment = [&](const CodeSet::Segment& segment,
                                       KMemoryPermission permission) {
         page_table->SetProcessMemoryPermission(segment.addr + base_addr, segment.size, permission);

src/core/hle/kernel/k_process.h

@@ -345,8 +345,8 @@ public:
 
     bool IsSignaled() const override;
 
-    void PinCurrentThread();
-    void UnpinCurrentThread();
+    void PinCurrentThread(s32 core_id);
+    void UnpinCurrentThread(s32 core_id);
     void UnpinThread(KThread* thread);
 
     KLightLock& GetStateLock() {

src/core/hle/kernel/k_scheduler.cpp

@@ -15,6 +15,7 @@
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/cpu_manager.h"
+#include "core/hle/kernel/k_interrupt_manager.h"
 #include "core/hle/kernel/k_process.h"
 #include "core/hle/kernel/k_scheduler.h"
 #include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h"
@@ -53,6 +54,13 @@ void KScheduler::RescheduleCores(KernelCore& kernel, u64 cores_pending_reschedul
         }
         cores_pending_reschedule &= ~(1ULL << core);
     }
+
+    for (std::size_t core_id = 0; core_id < Core::Hardware::NUM_CPU_CORES; ++core_id) {
+        if (kernel.PhysicalCore(core_id).IsInterrupted()) {
+            KInterruptManager::HandleInterrupt(kernel, static_cast<s32>(core_id));
+        }
+    }
+
     if (must_context_switch) {
         auto core_scheduler = kernel.CurrentScheduler();
         kernel.ExitSVCProfile();

src/core/hle/kernel/k_thread.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <algorithm>
+#include <atomic>
 #include <cinttypes>
 #include <optional>
 #include <vector>
@@ -26,12 +27,14 @@
 #include "core/hle/kernel/k_resource_limit.h"
 #include "core/hle/kernel/k_scheduler.h"
 #include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h"
+#include "core/hle/kernel/k_system_control.h"
 #include "core/hle/kernel/k_thread.h"
 #include "core/hle/kernel/k_thread_queue.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/svc_results.h"
 #include "core/hle/kernel/time_manager.h"
 #include "core/hle/result.h"
+#include "core/memory.h"
 
 #ifdef ARCHITECTURE_x86_64
 #include "core/arm/dynarmic/arm_dynarmic_32.h"
@@ -50,6 +53,7 @@ static void ResetThreadContext64(Core::ARM_Interface::ThreadContext64& context,
                                  VAddr entry_point, u64 arg) {
     context = {};
     context.cpu_registers[0] = arg;
+    context.cpu_registers[18] = Kernel::KSystemControl::GenerateRandomU64() | 1;
     context.pc = entry_point;
     context.sp = stack_top;
     // TODO(merry): Perform a hardware test to determine the below value.
@@ -61,6 +65,13 @@ namespace Kernel {
 
 namespace {
 
+struct ThreadLocalRegion {
+    static constexpr std::size_t MessageBufferSize = 0x100;
+    std::array<u32, MessageBufferSize / sizeof(u32)> message_buffer;
+    std::atomic_uint16_t disable_count;
+    std::atomic_uint16_t interrupt_flag;
+};
+
 class ThreadQueueImplForKThreadSleep final : public KThreadQueueWithoutEndWait {
 public:
     explicit ThreadQueueImplForKThreadSleep(KernelCore& kernel_)
@@ -344,7 +355,7 @@ void KThread::StartTermination() {
     if (parent != nullptr) {
         parent->ReleaseUserException(this);
         if (parent->GetPinnedThread(GetCurrentCoreId(kernel)) == this) {
-            parent->UnpinCurrentThread();
+            parent->UnpinCurrentThread(core_id);
         }
     }
 
@@ -370,7 +381,7 @@ void KThread::StartTermination() {
     this->Close();
 }
 
-void KThread::Pin() {
+void KThread::Pin(s32 current_core) {
     ASSERT(kernel.GlobalSchedulerContext().IsLocked());
 
     // Set ourselves as pinned.
@@ -387,7 +398,6 @@ void KThread::Pin() {
 
         // Bind ourselves to this core.
         const s32 active_core = GetActiveCore();
-        const s32 current_core = GetCurrentCoreId(kernel);
 
         SetActiveCore(current_core);
         physical_ideal_core_id = current_core;
@@ -480,6 +490,36 @@ void KThread::Unpin() {
     }
 }
 
+u16 KThread::GetUserDisableCount() const {
+    if (!IsUserThread()) {
+        // We only emulate TLS for user threads
+        return {};
+    }
+
+    auto& memory = kernel.System().Memory();
+    return memory.Read16(tls_address + offsetof(ThreadLocalRegion, disable_count));
+}
+
+void KThread::SetInterruptFlag() {
+    if (!IsUserThread()) {
+        // We only emulate TLS for user threads
+        return;
+    }
+
+    auto& memory = kernel.System().Memory();
+    memory.Write16(tls_address + offsetof(ThreadLocalRegion, interrupt_flag), 1);
+}
+
+void KThread::ClearInterruptFlag() {
+    if (!IsUserThread()) {
+        // We only emulate TLS for user threads
+        return;
+    }
+
+    auto& memory = kernel.System().Memory();
+    memory.Write16(tls_address + offsetof(ThreadLocalRegion, interrupt_flag), 0);
+}
+
 ResultCode KThread::GetCoreMask(s32* out_ideal_core, u64* out_affinity_mask) {
     KScopedSchedulerLock sl{kernel};
 

src/core/hle/kernel/k_thread.h

@@ -307,6 +307,10 @@ public:
         return parent != nullptr;
     }
 
+    u16 GetUserDisableCount() const;
+    void SetInterruptFlag();
+    void ClearInterruptFlag();
+
     [[nodiscard]] KThread* GetLockOwner() const {
         return lock_owner;
     }
@@ -490,7 +494,7 @@ public:
         this->GetStackParameters().disable_count--;
     }
 
-    void Pin();
+    void Pin(s32 current_core);
 
     void Unpin();
 

src/core/hle/kernel/kernel.cpp

@@ -182,7 +182,10 @@ struct KernelCore::Impl {
         // Shutdown all processes.
         if (current_process) {
             current_process->Finalize();
-            current_process->Close();
+            // current_process->Close();
+            // TODO: The current process should be destroyed based on accurate ref counting after
+            // calling Close(). Adding a manual Destroy() call instead to avoid a memory leak.
+            current_process->Destroy();
             current_process = nullptr;
         }
 

src/core/hle/kernel/svc.cpp

@@ -41,7 +41,6 @@
 #include "core/hle/kernel/svc_results.h"
 #include "core/hle/kernel/svc_types.h"
 #include "core/hle/kernel/svc_wrap.h"
-#include "core/hle/lock.h"
 #include "core/hle/result.h"
 #include "core/memory.h"
 #include "core/reporter.h"
@@ -136,25 +135,15 @@ enum class ResourceLimitValueType {
 } // Anonymous namespace
 
 /// Set the process heap to a given Size. It can both extend and shrink the heap.
-static ResultCode SetHeapSize(Core::System& system, VAddr* heap_addr, u64 heap_size) {
-    std::lock_guard lock{HLE::g_hle_lock};
-    LOG_TRACE(Kernel_SVC, "called, heap_size=0x{:X}", heap_size);
+static ResultCode SetHeapSize(Core::System& system, VAddr* out_address, u64 size) {
+    LOG_TRACE(Kernel_SVC, "called, heap_size=0x{:X}", size);
 
-    // Size must be a multiple of 0x200000 (2MB) and be equal to or less than 8GB.
-    if ((heap_size % 0x200000) != 0) {
-        LOG_ERROR(Kernel_SVC, "The heap size is not a multiple of 2MB, heap_size=0x{:016X}",
-                  heap_size);
-        return ResultInvalidSize;
-    }
+    // Validate size.
+    R_UNLESS(Common::IsAligned(size, HeapSizeAlignment), ResultInvalidSize);
+    R_UNLESS(size < MainMemorySizeMax, ResultInvalidSize);
 
-    if (heap_size >= 0x200000000) {
-        LOG_ERROR(Kernel_SVC, "The heap size is not less than 8GB, heap_size=0x{:016X}", heap_size);
-        return ResultInvalidSize;
-    }
-
-    auto& page_table{system.Kernel().CurrentProcess()->PageTable()};
-
-    CASCADE_RESULT(*heap_addr, page_table.SetHeapSize(heap_size));
+    // Set the heap size.
+    R_TRY(system.Kernel().CurrentProcess()->PageTable().SetHeapSize(out_address, size));
 
     return ResultSuccess;
 }
@@ -166,9 +155,38 @@ static ResultCode SetHeapSize32(Core::System& system, u32* heap_addr, u32 heap_s
     return result;
 }
 
+constexpr bool IsValidSetMemoryPermission(MemoryPermission perm) {
+    switch (perm) {
+    case MemoryPermission::None:
+    case MemoryPermission::Read:
+    case MemoryPermission::ReadWrite:
+        return true;
+    default:
+        return false;
+    }
+}
+
+static ResultCode SetMemoryPermission(Core::System& system, VAddr address, u64 size,
+                                      MemoryPermission perm) {
+    // Validate address / size.
+    R_UNLESS(Common::IsAligned(address, PageSize), ResultInvalidAddress);
+    R_UNLESS(Common::IsAligned(size, PageSize), ResultInvalidSize);
+    R_UNLESS(size > 0, ResultInvalidSize);
+    R_UNLESS((address < address + size), ResultInvalidCurrentMemory);
+
+    // Validate the permission.
+    R_UNLESS(IsValidSetMemoryPermission(perm), ResultInvalidNewMemoryPermission);
+
+    // Validate that the region is in range for the current process.
+    auto& page_table = system.Kernel().CurrentProcess()->PageTable();
+    R_UNLESS(page_table.Contains(address, size), ResultInvalidCurrentMemory);
+
+    // Set the memory attribute.
+    return page_table.SetMemoryPermission(address, size, perm);
+}
+
 static ResultCode SetMemoryAttribute(Core::System& system, VAddr address, u64 size, u32 mask,
                                      u32 attribute) {
-    std::lock_guard lock{HLE::g_hle_lock};
     LOG_DEBUG(Kernel_SVC,
               "called, address=0x{:016X}, size=0x{:X}, mask=0x{:08X}, attribute=0x{:08X}", address,
               size, mask, attribute);
@@ -212,7 +230,6 @@ static ResultCode SetMemoryAttribute32(Core::System& system, u32 address, u32 si
 
 /// Maps a memory range into a different range.
 static ResultCode MapMemory(Core::System& system, VAddr dst_addr, VAddr src_addr, u64 size) {
-    std::lock_guard lock{HLE::g_hle_lock};
     LOG_TRACE(Kernel_SVC, "called, dst_addr=0x{:X}, src_addr=0x{:X}, size=0x{:X}", dst_addr,
               src_addr, size);
 
@@ -232,7 +249,6 @@ static ResultCode MapMemory32(Core::System& system, u32 dst_addr, u32 src_addr,
 
 /// Unmaps a region that was previously mapped with svcMapMemory
 static ResultCode UnmapMemory(Core::System& system, VAddr dst_addr, VAddr src_addr, u64 size) {
-    std::lock_guard lock{HLE::g_hle_lock};
     LOG_TRACE(Kernel_SVC, "called, dst_addr=0x{:X}, src_addr=0x{:X}, size=0x{:X}", dst_addr,
               src_addr, size);
 
@@ -642,7 +658,6 @@ static void OutputDebugString(Core::System& system, VAddr address, u64 len) {
 /// Gets system/memory information for the current process
 static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, Handle handle,
                           u64 info_sub_id) {
-    std::lock_guard lock{HLE::g_hle_lock};
     LOG_TRACE(Kernel_SVC, "called info_id=0x{:X}, info_sub_id=0x{:X}, handle=0x{:08X}", info_id,
               info_sub_id, handle);
 
@@ -886,22 +901,17 @@ static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, Handle
         return ResultSuccess;
     }
     case GetInfoType::IdleTickCount: {
-        if (handle == 0) {
-            LOG_ERROR(Kernel_SVC, "Thread handle does not exist, handle=0x{:08X}",
-                      static_cast<Handle>(handle));
-            return ResultInvalidHandle;
-        }
+        // Verify the input handle is invalid.
+        R_UNLESS(handle == InvalidHandle, ResultInvalidHandle);
 
-        if (info_sub_id != 0xFFFFFFFFFFFFFFFF &&
-            info_sub_id != system.Kernel().CurrentPhysicalCoreIndex()) {
-            LOG_ERROR(Kernel_SVC, "Core is not the current core, got {}", info_sub_id);
-            return ResultInvalidCombination;
-        }
+        // Verify the requested core is valid.
+        const bool core_valid =
+            (info_sub_id == static_cast<u64>(-1ULL)) ||
+            (info_sub_id == static_cast<u64>(system.Kernel().CurrentPhysicalCoreIndex()));
+        R_UNLESS(core_valid, ResultInvalidCombination);
 
-        const auto& scheduler = *system.Kernel().CurrentScheduler();
-        const auto* const idle_thread = scheduler.GetIdleThread();
-
-        *result = idle_thread->GetCpuTime();
+        // Get the idle tick count.
+        *result = system.Kernel().CurrentScheduler()->GetIdleThread()->GetCpuTime();
         return ResultSuccess;
     }
     default:
@@ -924,7 +934,6 @@ static ResultCode GetInfo32(Core::System& system, u32* result_low, u32* result_h
 
 /// Maps memory at a desired address
 static ResultCode MapPhysicalMemory(Core::System& system, VAddr addr, u64 size) {
-    std::lock_guard lock{HLE::g_hle_lock};
     LOG_DEBUG(Kernel_SVC, "called, addr=0x{:016X}, size=0x{:X}", addr, size);
 
     if (!Common::Is4KBAligned(addr)) {
@@ -978,7 +987,6 @@ static ResultCode MapPhysicalMemory32(Core::System& system, u32 addr, u32 size)
 
 /// Unmaps memory previously mapped via MapPhysicalMemory
 static ResultCode UnmapPhysicalMemory(Core::System& system, VAddr addr, u64 size) {
-    std::lock_guard lock{HLE::g_hle_lock};
     LOG_DEBUG(Kernel_SVC, "called, addr=0x{:016X}, size=0x{:X}", addr, size);
 
     if (!Common::Is4KBAligned(addr)) {
@@ -1520,7 +1528,6 @@ static ResultCode ControlCodeMemory(Core::System& system, Handle code_memory_han
 static ResultCode QueryProcessMemory(Core::System& system, VAddr memory_info_address,
                                      VAddr page_info_address, Handle process_handle,
                                      VAddr address) {
-    std::lock_guard lock{HLE::g_hle_lock};
     LOG_TRACE(Kernel_SVC, "called process=0x{:08X} address={:X}", process_handle, address);
     const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable();
     KScopedAutoObject process = handle_table.GetObject<KProcess>(process_handle);
@@ -2020,6 +2027,25 @@ static ResultCode SignalToAddress(Core::System& system, VAddr address, Svc::Sign
                                                                   count);
 }
 
+static void SynchronizePreemptionState(Core::System& system) {
+    auto& kernel = system.Kernel();
+
+    // Lock the scheduler.
+    KScopedSchedulerLock sl{kernel};
+
+    // If the current thread is pinned, unpin it.
+    KProcess* cur_process = system.Kernel().CurrentProcess();
+    const auto core_id = GetCurrentCoreId(kernel);
+
+    if (cur_process->GetPinnedThread(core_id) == GetCurrentThreadPointer(kernel)) {
+        // Clear the current thread's interrupt flag.
+        GetCurrentThread(kernel).ClearInterruptFlag();
+
+        // Unpin the current thread.
+        cur_process->UnpinCurrentThread(core_id);
+    }
+}
+
 static ResultCode SignalToAddress32(Core::System& system, u32 address, Svc::SignalType signal_type,
                                     s32 value, s32 count) {
     return SignalToAddress(system, address, signal_type, value, count);
@@ -2738,7 +2764,7 @@ static const FunctionDef SVC_Table_32[] = {
 static const FunctionDef SVC_Table_64[] = {
     {0x00, nullptr, "Unknown"},
     {0x01, SvcWrap64<SetHeapSize>, "SetHeapSize"},
-    {0x02, nullptr, "SetMemoryPermission"},
+    {0x02, SvcWrap64<SetMemoryPermission>, "SetMemoryPermission"},
     {0x03, SvcWrap64<SetMemoryAttribute>, "SetMemoryAttribute"},
     {0x04, SvcWrap64<MapMemory>, "MapMemory"},
     {0x05, SvcWrap64<UnmapMemory>, "UnmapMemory"},
@@ -2790,7 +2816,7 @@ static const FunctionDef SVC_Table_64[] = {
     {0x33, SvcWrap64<GetThreadContext>, "GetThreadContext"},
     {0x34, SvcWrap64<WaitForAddress>, "WaitForAddress"},
     {0x35, SvcWrap64<SignalToAddress>, "SignalToAddress"},
-    {0x36, nullptr, "SynchronizePreemptionState"},
+    {0x36, SvcWrap64<SynchronizePreemptionState>, "SynchronizePreemptionState"},
     {0x37, nullptr, "Unknown"},
     {0x38, nullptr, "Unknown"},
     {0x39, nullptr, "Unknown"},

src/core/hle/kernel/svc_common.h

@@ -5,6 +5,7 @@
 #pragma once
 
 #include "common/common_types.h"
+#include "common/literals.h"
 
 namespace Kernel {
 using Handle = u32;
@@ -12,9 +13,13 @@ using Handle = u32;
 
 namespace Kernel::Svc {
 
+using namespace Common::Literals;
+
 constexpr s32 ArgumentHandleCountMax = 0x40;
 constexpr u32 HandleWaitMask{1u << 30};
 
+constexpr inline std::size_t HeapSizeAlignment = 2_MiB;
+
 constexpr inline Handle InvalidHandle = Handle(0);
 
 enum PseudoHandle : Handle {

src/core/hle/kernel/svc_wrap.h

@@ -249,6 +249,14 @@ void SvcWrap64(Core::System& system) {
         func(system, Param(system, 0), Param(system, 1), static_cast<u32>(Param(system, 2))).raw);
 }
 
+// Used by SetMemoryPermission
+template <ResultCode func(Core::System&, u64, u64, Svc::MemoryPermission)>
+void SvcWrap64(Core::System& system) {
+    FuncReturn(system, func(system, Param(system, 0), Param(system, 1),
+                            static_cast<Svc::MemoryPermission>(Param(system, 2)))
+                           .raw);
+}
+
 // Used by MapSharedMemory
 template <ResultCode func(Core::System&, Handle, u64, u64, Svc::MemoryPermission)>
 void SvcWrap64(Core::System& system) {

src/core/hle/lock.cpp

@@ -1,9 +0,0 @@
-// Copyright 2017 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <core/hle/lock.h>
-
-namespace HLE {
-std::recursive_mutex g_hle_lock;
-}

src/core/hle/lock.h

@@ -1,18 +0,0 @@
-// Copyright 2017 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <mutex>
-
-namespace HLE {
-/*
- * Synchronizes access to the internal HLE kernel structures, it is acquired when a guest
- * application thread performs a syscall. It should be acquired by any host threads that read or
- * modify the HLE kernel state. Note: Any operation that directly or indirectly reads from or writes
- * to the emulated memory is not protected by this mutex, and should be avoided in any threads other
- * than the CPU thread.
- */
-extern std::recursive_mutex g_hle_lock;
-} // namespace HLE

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

@@ -6,7 +6,6 @@
 #include "common/logging/log.h"
 #include "core/core.h"
 #include "core/hle/kernel/k_event.h"
-#include "core/hle/lock.h"
 #include "core/hle/service/bcat/backend/backend.h"
 
 namespace Service::BCAT {
@@ -29,10 +28,6 @@ DeliveryCacheProgressImpl& ProgressServiceBackend::GetImpl() {
     return impl;
 }
 
-void ProgressServiceBackend::SetNeedHLELock(bool need) {
-    need_hle_lock = need;
-}
-
 void ProgressServiceBackend::SetTotalSize(u64 size) {
     impl.total_bytes = size;
     SignalUpdate();
@@ -88,12 +83,7 @@ void ProgressServiceBackend::FinishDownload(ResultCode result) {
 }
 
 void ProgressServiceBackend::SignalUpdate() {
-    if (need_hle_lock) {
-        std::lock_guard lock(HLE::g_hle_lock);
-        update_event->GetWritableEvent().Signal();
-    } else {
-        update_event->GetWritableEvent().Signal();
-    }
+    update_event->GetWritableEvent().Signal();
 }
 
 Backend::Backend(DirectoryGetter getter) : dir_getter(std::move(getter)) {}

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

@@ -71,10 +71,6 @@ class ProgressServiceBackend {
 public:
     ~ProgressServiceBackend();
 
-    // Clients should call this with true if any of the functions are going to be called from a
-    // non-HLE thread and this class need to lock the hle mutex. (default is false)
-    void SetNeedHLELock(bool need);
-
     // Sets the number of bytes total in the entire download.
     void SetTotalSize(u64 size);
 
@@ -109,7 +105,6 @@ private:
 
     DeliveryCacheProgressImpl impl{};
     Kernel::KEvent* update_event;
-    bool need_hle_lock = false;
 };
 
 // A class representing an abstract backend for BCAT functionality.

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

@@ -33,15 +33,14 @@ void Controller_ConsoleSixAxis::OnUpdate(const Core::Timing::CoreTiming& core_ti
     const auto& last_entry = seven_sixaxis_lifo.ReadCurrentEntry().state;
     next_seven_sixaxis_state.sampling_number = last_entry.sampling_number + 1;
 
-    // Try to read sixaxis sensor states
     const auto motion_status = console->GetMotion();
+    last_global_timestamp = core_timing.GetGlobalTimeNs().count();
 
-    console_six_axis.is_seven_six_axis_sensor_at_rest = motion_status.is_at_rest;
-
+    // This value increments every time the switch goes to sleep
+    next_seven_sixaxis_state.unknown = 1;
+    next_seven_sixaxis_state.timestamp = last_global_timestamp - last_saved_timestamp;
     next_seven_sixaxis_state.accel = motion_status.accel;
-    // Zero gyro values as they just mess up with the camera
-    // Note: Probably a correct sensivity setting must be set
-    next_seven_sixaxis_state.gyro = {};
+    next_seven_sixaxis_state.gyro = motion_status.gyro;
     next_seven_sixaxis_state.quaternion = {
         {
             motion_status.quaternion.xyz.y,
@@ -52,9 +51,9 @@ void Controller_ConsoleSixAxis::OnUpdate(const Core::Timing::CoreTiming& core_ti
     };
 
     console_six_axis.sampling_number++;
-    // TODO(German77): Find the purpose of those values
-    console_six_axis.verticalization_error = 0.0f;
-    console_six_axis.gyro_bias = {0.0f, 0.0f, 0.0f};
+    console_six_axis.is_seven_six_axis_sensor_at_rest = motion_status.is_at_rest;
+    console_six_axis.verticalization_error = motion_status.verticalization_error;
+    console_six_axis.gyro_bias = motion_status.gyro_bias;
 
     // Update console six axis shared memory
     std::memcpy(data + SHARED_MEMORY_OFFSET, &console_six_axis, sizeof(console_six_axis));
@@ -69,7 +68,6 @@ void Controller_ConsoleSixAxis::SetTransferMemoryPointer(u8* t_mem) {
 }
 
 void Controller_ConsoleSixAxis::ResetTimestamp() {
-    seven_sixaxis_lifo.buffer_count = 0;
-    seven_sixaxis_lifo.buffer_tail = 0;
+    last_saved_timestamp = last_global_timestamp;
 }
 } // namespace Service::HID

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

@@ -39,8 +39,9 @@ public:
 
 private:
     struct SevenSixAxisState {
-        INSERT_PADDING_WORDS(4); // unused
-        s64 sampling_number{};
+        INSERT_PADDING_WORDS(2); // unused
+        u64 timestamp{};
+        u64 sampling_number{};
         u64 unknown{};
         Common::Vec3f accel{};
         Common::Vec3f gyro{};
@@ -52,9 +53,10 @@ private:
     struct ConsoleSharedMemory {
         u64 sampling_number{};
         bool is_seven_six_axis_sensor_at_rest{};
-        INSERT_PADDING_BYTES(4); // padding
+        INSERT_PADDING_BYTES(3); // padding
         f32 verticalization_error{};
         Common::Vec3f gyro_bias{};
+        INSERT_PADDING_BYTES(4); // padding
     };
     static_assert(sizeof(ConsoleSharedMemory) == 0x20, "ConsoleSharedMemory is an invalid size");
 
@@ -64,6 +66,8 @@ private:
     Core::HID::EmulatedConsole* console;
     u8* transfer_memory = nullptr;
     bool is_transfer_memory_set = false;
+    u64 last_saved_timestamp{};
+    u64 last_global_timestamp{};
     ConsoleSharedMemory console_six_axis{};
     SevenSixAxisState next_seven_sixaxis_state{};
 };

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

@@ -66,9 +66,9 @@ Controller_NPad::Controller_NPad(Core::HID::HIDCore& hid_core_,
         auto& controller = controller_data[i];
         controller.device = hid_core.GetEmulatedControllerByIndex(i);
         controller.vibration[Core::HID::EmulatedDeviceIndex::LeftIndex].latest_vibration_value =
-            DEFAULT_VIBRATION_VALUE;
+            Core::HID::DEFAULT_VIBRATION_VALUE;
         controller.vibration[Core::HID::EmulatedDeviceIndex::RightIndex].latest_vibration_value =
-            DEFAULT_VIBRATION_VALUE;
+            Core::HID::DEFAULT_VIBRATION_VALUE;
         Core::HID::ControllerUpdateCallback engine_callback{
             .on_change = [this,
                           i](Core::HID::ControllerTriggerType type) { ControllerUpdate(type, i); },
@@ -781,7 +781,8 @@ bool Controller_NPad::VibrateControllerAtIndex(Core::HID::NpadIdType npad_id,
             Core::HID::VibrationValue vibration{0.0f, 160.0f, 0.0f, 320.0f};
             controller.device->SetVibration(device_index, vibration);
             // Then reset the vibration value to its default value.
-            controller.vibration[device_index].latest_vibration_value = DEFAULT_VIBRATION_VALUE;
+            controller.vibration[device_index].latest_vibration_value =
+                Core::HID::DEFAULT_VIBRATION_VALUE;
         }
 
         return false;

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

@@ -90,13 +90,6 @@ public:
         Default = 3,
     };
 
-    static constexpr Core::HID::VibrationValue DEFAULT_VIBRATION_VALUE{
-        .low_amplitude = 0.0f,
-        .low_frequency = 160.0f,
-        .high_amplitude = 0.0f,
-        .high_frequency = 320.0f,
-    };
-
     void SetSupportedStyleSet(Core::HID::NpadStyleTag style_set);
     Core::HID::NpadStyleTag GetSupportedStyleSet() const;
 

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

@@ -1404,7 +1404,7 @@ void Hid::SendVibrationGcErmCommand(Kernel::HLERequestContext& ctx) {
                 .high_frequency = 0.0f,
             };
         default:
-            return Controller_NPad::DEFAULT_VIBRATION_VALUE;
+            return Core::HID::DEFAULT_VIBRATION_VALUE;
         }
     }();
 

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

@@ -9,7 +9,6 @@
 #include "core/core.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/k_event.h"
-#include "core/hle/lock.h"
 #include "core/hle/service/nfp/nfp.h"
 #include "core/hle/service/nfp/nfp_user.h"
 
@@ -337,7 +336,6 @@ void Module::Interface::CreateUserInterface(Kernel::HLERequestContext& ctx) {
 }
 
 bool Module::Interface::LoadAmiibo(const std::vector<u8>& buffer) {
-    std::lock_guard lock{HLE::g_hle_lock};
     if (buffer.size() < sizeof(AmiiboFile)) {
         return false;
     }

src/core/loader/kip.cpp

@@ -71,7 +71,6 @@ AppLoader::LoadResult AppLoader_KIP::Load(Kernel::KProcess& process,
                         kip->GetTitleID(), 0xFFFFFFFFFFFFFFFF, 0x1FE00000,
                         kip->GetKernelCapabilities());
 
-    const VAddr base_address = process.PageTable().GetCodeRegionStart();
     Kernel::CodeSet codeset;
     Kernel::PhysicalMemory program_image;
 
@@ -91,7 +90,14 @@ AppLoader::LoadResult AppLoader_KIP::Load(Kernel::KProcess& process,
     program_image.resize(PageAlignSize(kip->GetBSSOffset()) + kip->GetBSSSize());
     codeset.DataSegment().size += kip->GetBSSSize();
 
+    // Setup the process code layout
+    if (process.LoadFromMetadata(FileSys::ProgramMetadata::GetDefault(), program_image.size())
+            .IsError()) {
+        return {ResultStatus::ErrorNotInitialized, {}};
+    }
+
     codeset.memory = std::move(program_image);
+    const VAddr base_address = process.PageTable().GetCodeRegionStart();
     process.LoadModule(std::move(codeset), base_address);
 
     LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", kip->GetName(), base_address);

src/input_common/drivers/udp_client.cpp

@@ -536,42 +536,46 @@ CalibrationConfigurationJob::CalibrationConfigurationJob(
     std::function<void(u16, u16, u16, u16)> data_callback) {
 
     std::thread([=, this] {
+        u16 min_x{UINT16_MAX};
+        u16 min_y{UINT16_MAX};
+        u16 max_x{};
+        u16 max_y{};
+
         Status current_status{Status::Initialized};
-        SocketCallback callback{
-            [](Response::Version) {}, [](Response::PortInfo) {},
-            [&](Response::PadData data) {
-                static constexpr u16 CALIBRATION_THRESHOLD = 100;
-                static constexpr u16 MAX_VALUE = UINT16_MAX;
+        SocketCallback callback{[](Response::Version) {}, [](Response::PortInfo) {},
+                                [&](Response::PadData data) {
+                                    constexpr u16 CALIBRATION_THRESHOLD = 100;
 
-                if (current_status == Status::Initialized) {
-                    // Receiving data means the communication is ready now
-                    current_status = Status::Ready;
-                    status_callback(current_status);
-                }
-                const auto& touchpad_0 = data.touch[0];
-                if (touchpad_0.is_active == 0) {
-                    return;
-                }
-                LOG_DEBUG(Input, "Current touch: {} {}", touchpad_0.x, touchpad_0.y);
-                const u16 min_x = std::min(MAX_VALUE, static_cast<u16>(touchpad_0.x));
-                const u16 min_y = std::min(MAX_VALUE, static_cast<u16>(touchpad_0.y));
-                if (current_status == Status::Ready) {
-                    // First touch - min data (min_x/min_y)
-                    current_status = Status::Stage1Completed;
-                    status_callback(current_status);
-                }
-                if (touchpad_0.x - min_x > CALIBRATION_THRESHOLD &&
-                    touchpad_0.y - min_y > CALIBRATION_THRESHOLD) {
-                    // Set the current position as max value and finishes configuration
-                    const u16 max_x = touchpad_0.x;
-                    const u16 max_y = touchpad_0.y;
-                    current_status = Status::Completed;
-                    data_callback(min_x, min_y, max_x, max_y);
-                    status_callback(current_status);
+                                    if (current_status == Status::Initialized) {
+                                        // Receiving data means the communication is ready now
+                                        current_status = Status::Ready;
+                                        status_callback(current_status);
+                                    }
+                                    if (data.touch[0].is_active == 0) {
+                                        return;
+                                    }
+                                    LOG_DEBUG(Input, "Current touch: {} {}", data.touch[0].x,
+                                              data.touch[0].y);
+                                    min_x = std::min(min_x, static_cast<u16>(data.touch[0].x));
+                                    min_y = std::min(min_y, static_cast<u16>(data.touch[0].y));
+                                    if (current_status == Status::Ready) {
+                                        // First touch - min data (min_x/min_y)
+                                        current_status = Status::Stage1Completed;
+                                        status_callback(current_status);
+                                    }
+                                    if (data.touch[0].x - min_x > CALIBRATION_THRESHOLD &&
+                                        data.touch[0].y - min_y > CALIBRATION_THRESHOLD) {
+                                        // Set the current position as max value and finishes
+                                        // configuration
+                                        max_x = data.touch[0].x;
+                                        max_y = data.touch[0].y;
+                                        current_status = Status::Completed;
+                                        data_callback(min_x, min_y, max_x, max_y);
+                                        status_callback(current_status);
 
-                    complete_event.Set();
-                }
-            }};
+                                        complete_event.Set();
+                                    }
+                                }};
         Socket socket{host, port, std::move(callback)};
         std::thread worker_thread{SocketLoop, &socket};
         complete_event.Wait();

src/input_common/helpers/udp_protocol.h

@@ -54,6 +54,18 @@ struct Message {
 template <typename T>
 constexpr Type GetMessageType();
 
+template <typename T>
+Message<T> CreateMessage(const u32 magic, const T data, const u32 sender_id) {
+    boost::crc_32_type crc;
+    Header header{
+        magic, PROTOCOL_VERSION, sizeof(T) + sizeof(Type), 0, sender_id, GetMessageType<T>(),
+    };
+    Message<T> message{header, data};
+    crc.process_bytes(&message, sizeof(Message<T>));
+    message.header.crc = crc.checksum();
+    return message;
+}
+
 namespace Request {
 
 enum RegisterFlags : u8 {
@@ -101,14 +113,7 @@ static_assert(std::is_trivially_copyable_v<PadData>,
  */
 template <typename T>
 Message<T> Create(const T data, const u32 client_id = 0) {
-    boost::crc_32_type crc;
-    Header header{
-        CLIENT_MAGIC, PROTOCOL_VERSION, sizeof(T) + sizeof(Type), 0, client_id, GetMessageType<T>(),
-    };
-    Message<T> message{header, data};
-    crc.process_bytes(&message, sizeof(Message<T>));
-    message.header.crc = crc.checksum();
-    return message;
+    return CreateMessage(CLIENT_MAGIC, data, client_id);
 }
 } // namespace Request
 

src/shader_recompiler/backend/glasm/emit_glasm_context_get_set.cpp

@@ -86,7 +86,7 @@ void EmitGetAttribute(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr, Scal
     }
     switch (attr) {
     case IR::Attribute::PrimitiveId:
-        ctx.Add("MOV.S {}.x,primitive.id;", inst);
+        ctx.Add("MOV.F {}.x,primitive.id;", inst);
         break;
     case IR::Attribute::PositionX:
     case IR::Attribute::PositionY:
@@ -112,17 +112,33 @@ void EmitGetAttribute(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr, Scal
     case IR::Attribute::TessellationEvaluationPointV:
         ctx.Add("MOV.F {}.x,vertex.tesscoord.{};", inst, swizzle);
         break;
+    case IR::Attribute::InstanceId:
+        ctx.Add("MOV.F {}.x,{}.instance;", inst, ctx.attrib_name);
+        break;
+    case IR::Attribute::VertexId:
+        ctx.Add("MOV.F {}.x,{}.id;", inst, ctx.attrib_name);
+        break;
+    case IR::Attribute::FrontFace:
+        ctx.Add("CMP.F {}.x,{}.facing.x,0,-1;", inst, ctx.attrib_name);
+        break;
+    default:
+        throw NotImplementedException("Get attribute {}", attr);
+    }
+}
+
+void EmitGetAttributeU32(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr, ScalarU32) {
+    switch (attr) {
+    case IR::Attribute::PrimitiveId:
+        ctx.Add("MOV.S {}.x,primitive.id;", inst);
+        break;
     case IR::Attribute::InstanceId:
         ctx.Add("MOV.S {}.x,{}.instance;", inst, ctx.attrib_name);
         break;
     case IR::Attribute::VertexId:
         ctx.Add("MOV.S {}.x,{}.id;", inst, ctx.attrib_name);
         break;
-    case IR::Attribute::FrontFace:
-        ctx.Add("CMP.S {}.x,{}.facing.x,0,-1;", inst, ctx.attrib_name);
-        break;
     default:
-        throw NotImplementedException("Get attribute {}", attr);
+        throw NotImplementedException("Get U32 attribute {}", attr);
     }
 }
 

src/shader_recompiler/backend/glasm/emit_glasm_instructions.h

@@ -50,6 +50,7 @@ void EmitGetCbufU32(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding,
 void EmitGetCbufF32(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding, ScalarU32 offset);
 void EmitGetCbufU32x2(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding, ScalarU32 offset);
 void EmitGetAttribute(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr, ScalarU32 vertex);
+void EmitGetAttributeU32(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr, ScalarU32 vertex);
 void EmitSetAttribute(EmitContext& ctx, IR::Attribute attr, ScalarF32 value, ScalarU32 vertex);
 void EmitGetAttributeIndexed(EmitContext& ctx, IR::Inst& inst, ScalarS32 offset, ScalarU32 vertex);
 void EmitSetAttributeIndexed(EmitContext& ctx, ScalarU32 offset, ScalarF32 value, ScalarU32 vertex);

src/shader_recompiler/backend/glsl/emit_glsl.cpp

@@ -176,7 +176,7 @@ void EmitCode(EmitContext& ctx, const IR::Program& program) {
 }
 
 std::string GlslVersionSpecifier(const EmitContext& ctx) {
-    if (ctx.uses_y_direction || ctx.info.stores.Legacy() || ctx.info.loads.Legacy()) {
+    if (ctx.uses_y_direction) {
         return " compatibility";
     }
     return "";

src/shader_recompiler/backend/glsl/emit_glsl_bitwise_conversion.cpp

@@ -7,6 +7,7 @@
 #include "shader_recompiler/backend/glsl/emit_glsl_instructions.h"
 #include "shader_recompiler/backend/glsl/glsl_emit_context.h"
 #include "shader_recompiler/frontend/ir/value.h"
+#include "shader_recompiler/profile.h"
 
 namespace Shader::Backend::GLSL {
 namespace {
@@ -30,8 +31,9 @@ void EmitConditionRef(EmitContext& ctx, IR::Inst& inst, const IR::Value& value)
     inst.DestructiveAddUsage(1);
     const auto ret{ctx.var_alloc.Define(inst, GlslVarType::U1)};
     const auto input{ctx.var_alloc.Consume(value)};
+    const auto suffix{ctx.profile.has_gl_bool_ref_bug ? "?true:false" : ""};
     if (ret != input) {
-        ctx.Add("{}={};", ret, input);
+        ctx.Add("{}={}{};", ret, input, suffix);
     }
 }
 

src/shader_recompiler/backend/glsl/emit_glsl_context_get_set.cpp

@@ -98,47 +98,50 @@ void GetCbuf16(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding, const
         GetCbuf(ctx, ret, binding, offset, 16, cast, bit_offset);
     }
 }
-
-u32 TexCoordIndex(IR::Attribute attr) {
-    return (static_cast<u32>(attr) - static_cast<u32>(IR::Attribute::FixedFncTexture0S)) / 4;
-}
 } // Anonymous namespace
 
 void EmitGetCbufU8(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding,
                    const IR::Value& offset) {
-    GetCbuf8(ctx, inst, binding, offset, "ftou");
+    const auto cast{ctx.profile.has_gl_cbuf_ftou_bug ? "" : "ftou"};
+    GetCbuf8(ctx, inst, binding, offset, cast);
 }
 
 void EmitGetCbufS8(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding,
                    const IR::Value& offset) {
-    GetCbuf8(ctx, inst, binding, offset, "ftoi");
+    const auto cast{ctx.profile.has_gl_cbuf_ftou_bug ? "int" : "ftoi"};
+    GetCbuf8(ctx, inst, binding, offset, cast);
 }
 
 void EmitGetCbufU16(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding,
                     const IR::Value& offset) {
-    GetCbuf16(ctx, inst, binding, offset, "ftou");
+    const auto cast{ctx.profile.has_gl_cbuf_ftou_bug ? "" : "ftou"};
+    GetCbuf16(ctx, inst, binding, offset, cast);
 }
 
 void EmitGetCbufS16(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding,
                     const IR::Value& offset) {
-    GetCbuf16(ctx, inst, binding, offset, "ftoi");
+    const auto cast{ctx.profile.has_gl_cbuf_ftou_bug ? "int" : "ftoi"};
+    GetCbuf16(ctx, inst, binding, offset, cast);
 }
 
 void EmitGetCbufU32(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding,
                     const IR::Value& offset) {
     const auto ret{ctx.var_alloc.Define(inst, GlslVarType::U32)};
-    GetCbuf(ctx, ret, binding, offset, 32, "ftou");
+    const auto cast{ctx.profile.has_gl_cbuf_ftou_bug ? "" : "ftou"};
+    GetCbuf(ctx, ret, binding, offset, 32, cast);
 }
 
 void EmitGetCbufF32(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding,
                     const IR::Value& offset) {
     const auto ret{ctx.var_alloc.Define(inst, GlslVarType::F32)};
-    GetCbuf(ctx, ret, binding, offset, 32);
+    const auto cast{ctx.profile.has_gl_cbuf_ftou_bug ? "utof" : ""};
+    GetCbuf(ctx, ret, binding, offset, 32, cast);
 }
 
 void EmitGetCbufU32x2(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding,
                       const IR::Value& offset) {
     const auto cbuf{fmt::format("{}_cbuf{}", ctx.stage_name, binding.U32())};
+    const auto cast{ctx.profile.has_gl_cbuf_ftou_bug ? "" : "ftou"};
     if (offset.IsImmediate()) {
         static constexpr u32 cbuf_size{0x10000};
         const u32 u32_offset{offset.U32()};
@@ -149,26 +152,26 @@ void EmitGetCbufU32x2(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding
             return;
         }
         if (u32_offset % 2 == 0) {
-            ctx.AddU32x2("{}=ftou({}[{}].{}{});", inst, cbuf, u32_offset / 16,
+            ctx.AddU32x2("{}={}({}[{}].{}{});", inst, cast, cbuf, u32_offset / 16,
                          OffsetSwizzle(u32_offset), OffsetSwizzle(u32_offset + 4));
         } else {
-            ctx.AddU32x2("{}=uvec2(ftou({}[{}].{}),ftou({}[{}].{}));", inst, cbuf, u32_offset / 16,
-                         OffsetSwizzle(u32_offset), cbuf, (u32_offset + 4) / 16,
-                         OffsetSwizzle(u32_offset + 4));
+            ctx.AddU32x2("{}=uvec2({}({}[{}].{}),{}({}[{}].{}));", inst, cast, cbuf,
+                         u32_offset / 16, OffsetSwizzle(u32_offset), cast, cbuf,
+                         (u32_offset + 4) / 16, OffsetSwizzle(u32_offset + 4));
         }
         return;
     }
     const auto offset_var{ctx.var_alloc.Consume(offset)};
     if (!ctx.profile.has_gl_component_indexing_bug) {
-        ctx.AddU32x2("{}=uvec2(ftou({}[{}>>4][({}>>2)%4]),ftou({}[({}+4)>>4][(({}+4)>>2)%4]));",
-                     inst, cbuf, offset_var, offset_var, cbuf, offset_var, offset_var);
+        ctx.AddU32x2("{}=uvec2({}({}[{}>>4][({}>>2)%4]),{}({}[({}+4)>>4][(({}+4)>>2)%4]));", inst,
+                     cast, cbuf, offset_var, offset_var, cast, cbuf, offset_var, offset_var);
         return;
     }
     const auto ret{ctx.var_alloc.Define(inst, GlslVarType::U32x2)};
     const auto cbuf_offset{fmt::format("{}>>2", offset_var)};
     for (u32 swizzle = 0; swizzle < 4; ++swizzle) {
-        ctx.Add("if(({}&3)=={}){}=uvec2(ftou({}[{}>>4].{}),ftou({}[({}+4)>>4].{}));", cbuf_offset,
-                swizzle, ret, cbuf, offset_var, "xyzw"[swizzle], cbuf, offset_var,
+        ctx.Add("if(({}&3)=={}){}=uvec2({}({}[{}>>4].{}),{}({}[({}+4)>>4].{}));", cbuf_offset,
+                swizzle, ret, cast, cbuf, offset_var, "xyzw"[swizzle], cast, cbuf, offset_var,
                 "xyzw"[(swizzle + 1) % 4]);
     }
 }
@@ -190,18 +193,6 @@ void EmitGetAttribute(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr,
         ctx.AddF32("{}=in_attr{}{}.{};", inst, index, InputVertexIndex(ctx, vertex), swizzle);
         return;
     }
-    // GLSL only exposes 8 legacy texcoords
-    if (attr >= IR::Attribute::FixedFncTexture8S && attr <= IR::Attribute::FixedFncTexture9Q) {
-        LOG_WARNING(Shader_GLSL, "GLSL does not allow access to gl_TexCoord[{}]",
-                    TexCoordIndex(attr));
-        ctx.AddF32("{}=0.f;", inst);
-        return;
-    }
-    if (attr >= IR::Attribute::FixedFncTexture0S && attr <= IR::Attribute::FixedFncTexture7Q) {
-        const u32 index{TexCoordIndex(attr)};
-        ctx.AddF32("{}=gl_TexCoord[{}].{};", inst, index, swizzle);
-        return;
-    }
     switch (attr) {
     case IR::Attribute::PrimitiveId:
         ctx.AddF32("{}=itof(gl_PrimitiveID);", inst);
@@ -215,16 +206,6 @@ void EmitGetAttribute(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr,
         ctx.AddF32("{}={}{}.{};", inst, input_decorator, ctx.position_name, swizzle);
         break;
     }
-    case IR::Attribute::ColorFrontDiffuseR:
-    case IR::Attribute::ColorFrontDiffuseG:
-    case IR::Attribute::ColorFrontDiffuseB:
-    case IR::Attribute::ColorFrontDiffuseA:
-        if (ctx.stage == Stage::Fragment) {
-            ctx.AddF32("{}=gl_Color.{};", inst, swizzle);
-        } else {
-            ctx.AddF32("{}=gl_FrontColor.{};", inst, swizzle);
-        }
-        break;
     case IR::Attribute::PointSpriteS:
     case IR::Attribute::PointSpriteT:
         ctx.AddF32("{}=gl_PointCoord.{};", inst, swizzle);
@@ -247,6 +228,22 @@ void EmitGetAttribute(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr,
     }
 }
 
+void EmitGetAttributeU32(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr, std::string_view) {
+    switch (attr) {
+    case IR::Attribute::PrimitiveId:
+        ctx.AddU32("{}=uint(gl_PrimitiveID);", inst);
+        break;
+    case IR::Attribute::InstanceId:
+        ctx.AddU32("{}=uint(gl_InstanceID);", inst);
+        break;
+    case IR::Attribute::VertexId:
+        ctx.AddU32("{}=uint(gl_VertexID);", inst);
+        break;
+    default:
+        throw NotImplementedException("Get U32 attribute {}", attr);
+    }
+}
+
 void EmitSetAttribute(EmitContext& ctx, IR::Attribute attr, std::string_view value,
                       [[maybe_unused]] std::string_view vertex) {
     if (IR::IsGeneric(attr)) {
@@ -264,17 +261,6 @@ void EmitSetAttribute(EmitContext& ctx, IR::Attribute attr, std::string_view val
     }
     const u32 element{static_cast<u32>(attr) % 4};
     const char swizzle{"xyzw"[element]};
-    // GLSL only exposes 8 legacy texcoords
-    if (attr >= IR::Attribute::FixedFncTexture8S && attr <= IR::Attribute::FixedFncTexture9Q) {
-        LOG_WARNING(Shader_GLSL, "GLSL does not allow access to gl_TexCoord[{}]",
-                    TexCoordIndex(attr));
-        return;
-    }
-    if (attr >= IR::Attribute::FixedFncTexture0S && attr <= IR::Attribute::FixedFncTexture7Q) {
-        const u32 index{TexCoordIndex(attr)};
-        ctx.Add("gl_TexCoord[{}].{}={};", index, swizzle, value);
-        return;
-    }
     switch (attr) {
     case IR::Attribute::Layer:
         if (ctx.stage != Stage::Geometry &&
@@ -312,33 +298,6 @@ void EmitSetAttribute(EmitContext& ctx, IR::Attribute attr, std::string_view val
     case IR::Attribute::PositionW:
         ctx.Add("gl_Position.{}={};", swizzle, value);
         break;
-    case IR::Attribute::ColorFrontDiffuseR:
-    case IR::Attribute::ColorFrontDiffuseG:
-    case IR::Attribute::ColorFrontDiffuseB:
-    case IR::Attribute::ColorFrontDiffuseA:
-        ctx.Add("gl_FrontColor.{}={};", swizzle, value);
-        break;
-    case IR::Attribute::ColorFrontSpecularR:
-    case IR::Attribute::ColorFrontSpecularG:
-    case IR::Attribute::ColorFrontSpecularB:
-    case IR::Attribute::ColorFrontSpecularA:
-        ctx.Add("gl_FrontSecondaryColor.{}={};", swizzle, value);
-        break;
-    case IR::Attribute::ColorBackDiffuseR:
-    case IR::Attribute::ColorBackDiffuseG:
-    case IR::Attribute::ColorBackDiffuseB:
-    case IR::Attribute::ColorBackDiffuseA:
-        ctx.Add("gl_BackColor.{}={};", swizzle, value);
-        break;
-    case IR::Attribute::ColorBackSpecularR:
-    case IR::Attribute::ColorBackSpecularG:
-    case IR::Attribute::ColorBackSpecularB:
-    case IR::Attribute::ColorBackSpecularA:
-        ctx.Add("gl_BackSecondaryColor.{}={};", swizzle, value);
-        break;
-    case IR::Attribute::FogCoordinate:
-        ctx.Add("gl_FogFragCoord={};", value);
-        break;
     case IR::Attribute::ClipDistance0:
     case IR::Attribute::ClipDistance1:
     case IR::Attribute::ClipDistance2:

src/shader_recompiler/backend/glsl/emit_glsl_floating_point.cpp

@@ -125,11 +125,11 @@ void EmitFPNeg16([[maybe_unused]] EmitContext& ctx, [[maybe_unused]] IR::Inst& i
 }
 
 void EmitFPNeg32(EmitContext& ctx, IR::Inst& inst, std::string_view value) {
-    ctx.AddF32("{}=-({});", inst, value);
+    ctx.AddF32("{}=0.f-({});", inst, value);
 }
 
 void EmitFPNeg64(EmitContext& ctx, IR::Inst& inst, std::string_view value) {
-    ctx.AddF64("{}=-({});", inst, value);
+    ctx.AddF64("{}=double(0.)-({});", inst, value);
 }
 
 void EmitFPSin(EmitContext& ctx, IR::Inst& inst, std::string_view value) {

src/shader_recompiler/backend/glsl/emit_glsl_instructions.h

@@ -60,6 +60,8 @@ void EmitGetCbufU32x2(EmitContext& ctx, IR::Inst& inst, const IR::Value& binding
                       const IR::Value& offset);
 void EmitGetAttribute(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr,
                       std::string_view vertex);
+void EmitGetAttributeU32(EmitContext& ctx, IR::Inst& inst, IR::Attribute attr,
+                         std::string_view vertex);
 void EmitSetAttribute(EmitContext& ctx, IR::Attribute attr, std::string_view value,
                       std::string_view vertex);
 void EmitGetAttributeIndexed(EmitContext& ctx, IR::Inst& inst, std::string_view offset,

src/shader_recompiler/backend/glsl/emit_glsl_integer.cpp

@@ -87,11 +87,11 @@ void EmitUDiv32(EmitContext& ctx, IR::Inst& inst, std::string_view a, std::strin
 }
 
 void EmitINeg32(EmitContext& ctx, IR::Inst& inst, std::string_view value) {
-    ctx.AddU32("{}=uint(-({}));", inst, value);
+    ctx.AddU32("{}=uint(int(0)-int({}));", inst, value);
 }
 
 void EmitINeg64(EmitContext& ctx, IR::Inst& inst, std::string_view value) {
-    ctx.AddU64("{}=-({});", inst, value);
+    ctx.AddU64("{}=uint64_t(int64_t(0)-int64_t({}));", inst, value);
 }
 
 void EmitIAbs32(EmitContext& ctx, IR::Inst& inst, std::string_view value) {

src/shader_recompiler/backend/glsl/emit_glsl_special.cpp

@@ -90,7 +90,9 @@ void EmitPhiMove(EmitContext& ctx, const IR::Value& phi_value, const IR::Value&
     if (phi_reg == val_reg) {
         return;
     }
-    ctx.Add("{}={};", phi_reg, val_reg);
+    const bool needs_workaround{ctx.profile.has_gl_bool_ref_bug && phi_type == IR::Type::U1};
+    const auto suffix{needs_workaround ? "?true:false" : ""};
+    ctx.Add("{}={}{};", phi_reg, val_reg, suffix);
 }
 
 void EmitPrologue(EmitContext& ctx) {

src/shader_recompiler/backend/glsl/glsl_emit_context.cpp

@@ -211,27 +211,6 @@ std::string_view OutputPrimitive(OutputTopology topology) {
     throw InvalidArgument("Invalid output topology {}", topology);
 }
 
-void SetupLegacyOutPerVertex(EmitContext& ctx, std::string& header) {
-    if (!ctx.info.stores.Legacy()) {
-        return;
-    }
-    if (ctx.info.stores.FixedFunctionTexture()) {
-        header += "vec4 gl_TexCoord[8];";
-    }
-    if (ctx.info.stores.AnyComponent(IR::Attribute::ColorFrontDiffuseR)) {
-        header += "vec4 gl_FrontColor;";
-    }
-    if (ctx.info.stores.AnyComponent(IR::Attribute::ColorFrontSpecularR)) {
-        header += "vec4 gl_FrontSecondaryColor;";
-    }
-    if (ctx.info.stores.AnyComponent(IR::Attribute::ColorBackDiffuseR)) {
-        header += "vec4 gl_BackColor;";
-    }
-    if (ctx.info.stores.AnyComponent(IR::Attribute::ColorBackSpecularR)) {
-        header += "vec4 gl_BackSecondaryColor;";
-    }
-}
-
 void SetupOutPerVertex(EmitContext& ctx, std::string& header) {
     if (!StoresPerVertexAttributes(ctx.stage)) {
         return;
@@ -250,7 +229,6 @@ void SetupOutPerVertex(EmitContext& ctx, std::string& header) {
         ctx.profile.support_viewport_index_layer_non_geometry && ctx.stage != Stage::Geometry) {
         header += "int gl_ViewportIndex;";
     }
-    SetupLegacyOutPerVertex(ctx, header);
     header += "};";
     if (ctx.info.stores[IR::Attribute::ViewportIndex] && ctx.stage == Stage::Geometry) {
         header += "out int gl_ViewportIndex;";
@@ -282,21 +260,6 @@ void SetupInPerVertex(EmitContext& ctx, std::string& header) {
     }
     header += "}gl_in[gl_MaxPatchVertices];";
 }
-
-void SetupLegacyInPerFragment(EmitContext& ctx, std::string& header) {
-    if (!ctx.info.loads.Legacy()) {
-        return;
-    }
-    header += "in gl_PerFragment{";
-    if (ctx.info.loads.FixedFunctionTexture()) {
-        header += "vec4 gl_TexCoord[8];";
-    }
-    if (ctx.info.loads.AnyComponent(IR::Attribute::ColorFrontDiffuseR)) {
-        header += "vec4 gl_Color;";
-    }
-    header += "};";
-}
-
 } // Anonymous namespace
 
 EmitContext::EmitContext(IR::Program& program, Bindings& bindings, const Profile& profile_,
@@ -361,7 +324,6 @@ EmitContext::EmitContext(IR::Program& program, Bindings& bindings, const Profile
     }
     SetupOutPerVertex(*this, header);
     SetupInPerVertex(*this, header);
-    SetupLegacyInPerFragment(*this, header);
 
     for (size_t index = 0; index < IR::NUM_GENERICS; ++index) {
         if (!info.loads.Generic(index) || !runtime_info.previous_stage_stores.Generic(index)) {
@@ -466,9 +428,10 @@ void EmitContext::DefineConstantBuffers(Bindings& bindings) {
         return;
     }
     for (const auto& desc : info.constant_buffer_descriptors) {
-        header += fmt::format(
-            "layout(std140,binding={}) uniform {}_cbuf_{}{{vec4 {}_cbuf{}[{}];}};",
-            bindings.uniform_buffer, stage_name, desc.index, stage_name, desc.index, 4 * 1024);
+        const auto cbuf_type{profile.has_gl_cbuf_ftou_bug ? "uvec4" : "vec4"};
+        header += fmt::format("layout(std140,binding={}) uniform {}_cbuf_{}{{{} {}_cbuf{}[{}];}};",
+                              bindings.uniform_buffer, stage_name, desc.index, cbuf_type,
+                              stage_name, desc.index, 4 * 1024);
         bindings.uniform_buffer += desc.count;
     }
 }

src/shader_recompiler/backend/spirv/emit_spirv.cpp

@@ -30,11 +30,20 @@ struct FuncTraits<ReturnType_ (*)(Args...)> {
     using ArgType = std::tuple_element_t<I, std::tuple<Args...>>;
 };
 
+#ifdef _MSC_VER
+#pragma warning(push)
+#pragma warning(disable : 4702) // Ignore unreachable code warning
+#endif
+
 template <auto func, typename... Args>
 void SetDefinition(EmitContext& ctx, IR::Inst* inst, Args... args) {
     inst->SetDefinition<Id>(func(ctx, std::forward<Args>(args)...));
 }
 
+#ifdef _MSC_VER
+#pragma warning(pop)
+#endif
+
 template <typename ArgType>
 ArgType Arg(EmitContext& ctx, const IR::Value& arg) {
     if constexpr (std::is_same_v<ArgType, Id>) {

src/shader_recompiler/backend/spirv/emit_spirv_context_get_set.cpp

@@ -44,14 +44,6 @@ Id AttrPointer(EmitContext& ctx, Id pointer_type, Id vertex, Id base, Args&&...
     }
 }
 
-bool IsLegacyAttribute(IR::Attribute attribute) {
-    return (attribute >= IR::Attribute::ColorFrontDiffuseR &&
-            attribute <= IR::Attribute::ColorBackSpecularA) ||
-           attribute == IR::Attribute::FogCoordinate ||
-           (attribute >= IR::Attribute::FixedFncTexture0S &&
-            attribute <= IR::Attribute::FixedFncTexture9Q);
-}
-
 template <typename... Args>
 Id OutputAccessChain(EmitContext& ctx, Id result_type, Id base, Args&&... args) {
     if (ctx.stage == Stage::TessellationControl) {
@@ -83,17 +75,6 @@ std::optional<OutAttr> OutputAttrPointer(EmitContext& ctx, IR::Attribute attr) {
             return OutputAccessChain(ctx, ctx.output_f32, info.id, index_id);
         }
     }
-    if (IsLegacyAttribute(attr)) {
-        if (attr == IR::Attribute::FogCoordinate) {
-            return OutputAccessChain(ctx, ctx.output_f32, ctx.OutputLegacyAttribute(attr),
-                                     ctx.Const(0u));
-        } else {
-            const u32 element{static_cast<u32>(attr) % 4};
-            const Id element_id{ctx.Const(element)};
-            return OutputAccessChain(ctx, ctx.output_f32, ctx.OutputLegacyAttribute(attr),
-                                     element_id);
-        }
-    }
     switch (attr) {
     case IR::Attribute::PointSize:
         return ctx.output_point_size;
@@ -327,18 +308,6 @@ Id EmitGetAttribute(EmitContext& ctx, IR::Attribute attr, Id vertex) {
         const Id value{ctx.OpLoad(type->id, pointer)};
         return type->needs_cast ? ctx.OpBitcast(ctx.F32[1], value) : value;
     }
-    if (IsLegacyAttribute(attr)) {
-        if (attr == IR::Attribute::FogCoordinate) {
-            const Id attr_ptr{AttrPointer(ctx, ctx.input_f32, vertex,
-                                          ctx.InputLegacyAttribute(attr), ctx.Const(0u))};
-            return ctx.OpLoad(ctx.F32[1], attr_ptr);
-        } else {
-            const Id element_id{ctx.Const(element)};
-            const Id attr_ptr{AttrPointer(ctx, ctx.input_f32, vertex,
-                                          ctx.InputLegacyAttribute(attr), element_id)};
-            return ctx.OpLoad(ctx.F32[1], attr_ptr);
-        }
-    }
     switch (attr) {
     case IR::Attribute::PrimitiveId:
         return ctx.OpBitcast(ctx.F32[1], ctx.OpLoad(ctx.U32[1], ctx.primitive_id));
@@ -386,6 +355,31 @@ Id EmitGetAttribute(EmitContext& ctx, IR::Attribute attr, Id vertex) {
     }
 }
 
+Id EmitGetAttributeU32(EmitContext& ctx, IR::Attribute attr, Id) {
+    switch (attr) {
+    case IR::Attribute::PrimitiveId:
+        return ctx.OpLoad(ctx.U32[1], ctx.primitive_id);
+    case IR::Attribute::InstanceId:
+        if (ctx.profile.support_vertex_instance_id) {
+            return ctx.OpLoad(ctx.U32[1], ctx.instance_id);
+        } else {
+            const Id index{ctx.OpLoad(ctx.U32[1], ctx.instance_index)};
+            const Id base{ctx.OpLoad(ctx.U32[1], ctx.base_instance)};
+            return ctx.OpISub(ctx.U32[1], index, base);
+        }
+    case IR::Attribute::VertexId:
+        if (ctx.profile.support_vertex_instance_id) {
+            return ctx.OpLoad(ctx.U32[1], ctx.vertex_id);
+        } else {
+            const Id index{ctx.OpLoad(ctx.U32[1], ctx.vertex_index)};
+            const Id base{ctx.OpLoad(ctx.U32[1], ctx.base_vertex)};
+            return ctx.OpISub(ctx.U32[1], index, base);
+        }
+    default:
+        throw NotImplementedException("Read U32 attribute {}", attr);
+    }
+}
+
 void EmitSetAttribute(EmitContext& ctx, IR::Attribute attr, Id value, [[maybe_unused]] Id vertex) {
     const std::optional<OutAttr> output{OutputAttrPointer(ctx, attr)};
     if (!output) {

src/shader_recompiler/backend/spirv/emit_spirv_instructions.h

@@ -53,6 +53,7 @@ Id EmitGetCbufU32(EmitContext& ctx, const IR::Value& binding, const IR::Value& o
 Id EmitGetCbufF32(EmitContext& ctx, const IR::Value& binding, const IR::Value& offset);
 Id EmitGetCbufU32x2(EmitContext& ctx, const IR::Value& binding, const IR::Value& offset);
 Id EmitGetAttribute(EmitContext& ctx, IR::Attribute attr, Id vertex);
+Id EmitGetAttributeU32(EmitContext& ctx, IR::Attribute attr, Id vertex);
 void EmitSetAttribute(EmitContext& ctx, IR::Attribute attr, Id value, Id vertex);
 Id EmitGetAttributeIndexed(EmitContext& ctx, Id offset, Id vertex);
 void EmitSetAttributeIndexed(EmitContext& ctx, Id offset, Id value, Id vertex);

src/shader_recompiler/backend/spirv/spirv_emit_context.cpp

@@ -18,8 +18,6 @@
 
 namespace Shader::Backend::SPIRV {
 namespace {
-constexpr size_t NUM_FIXEDFNCTEXTURE = 10;
-
 enum class Operation {
     Increment,
     Decrement,
@@ -432,34 +430,6 @@ Id DescType(EmitContext& ctx, Id sampled_type, Id pointer_type, u32 count) {
         return pointer_type;
     }
 }
-
-size_t FindAndSetNextUnusedLocation(std::bitset<IR::NUM_GENERICS>& used_locations,
-                                    size_t& start_offset) {
-    for (size_t location = start_offset; location < used_locations.size(); ++location) {
-        if (!used_locations.test(location)) {
-            start_offset = location;
-            used_locations.set(location);
-            return location;
-        }
-    }
-    throw RuntimeError("Unable to get an unused location for legacy attribute");
-}
-
-Id DefineLegacyInput(EmitContext& ctx, std::bitset<IR::NUM_GENERICS>& used_locations,
-                     size_t& start_offset) {
-    const Id id{DefineInput(ctx, ctx.F32[4], true)};
-    const size_t location = FindAndSetNextUnusedLocation(used_locations, start_offset);
-    ctx.Decorate(id, spv::Decoration::Location, location);
-    return id;
-}
-
-Id DefineLegacyOutput(EmitContext& ctx, std::bitset<IR::NUM_GENERICS>& used_locations,
-                      size_t& start_offset, std::optional<u32> invocations) {
-    const Id id{DefineOutput(ctx, ctx.F32[4], invocations)};
-    const size_t location = FindAndSetNextUnusedLocation(used_locations, start_offset);
-    ctx.Decorate(id, spv::Decoration::Location, location);
-    return id;
-}
 } // Anonymous namespace
 
 void VectorTypes::Define(Sirit::Module& sirit_ctx, Id base_type, std::string_view name) {
@@ -543,64 +513,6 @@ Id EmitContext::BitOffset16(const IR::Value& offset) {
     return OpBitwiseAnd(U32[1], OpShiftLeftLogical(U32[1], Def(offset), Const(3u)), Const(16u));
 }
 
-Id EmitContext::InputLegacyAttribute(IR::Attribute attribute) {
-    if (attribute >= IR::Attribute::ColorFrontDiffuseR &&
-        attribute <= IR::Attribute::ColorFrontDiffuseA) {
-        return input_front_color;
-    }
-    if (attribute >= IR::Attribute::ColorFrontSpecularR &&
-        attribute <= IR::Attribute::ColorFrontSpecularA) {
-        return input_front_secondary_color;
-    }
-    if (attribute >= IR::Attribute::ColorBackDiffuseR &&
-        attribute <= IR::Attribute::ColorBackDiffuseA) {
-        return input_back_color;
-    }
-    if (attribute >= IR::Attribute::ColorBackSpecularR &&
-        attribute <= IR::Attribute::ColorBackSpecularA) {
-        return input_back_secondary_color;
-    }
-    if (attribute == IR::Attribute::FogCoordinate) {
-        return input_fog_frag_coord;
-    }
-    if (attribute >= IR::Attribute::FixedFncTexture0S &&
-        attribute <= IR::Attribute::FixedFncTexture9Q) {
-        u32 index =
-            (static_cast<u32>(attribute) - static_cast<u32>(IR::Attribute::FixedFncTexture0S)) / 4;
-        return input_fixed_fnc_textures[index];
-    }
-    throw InvalidArgument("Attribute is not legacy attribute {}", attribute);
-}
-
-Id EmitContext::OutputLegacyAttribute(IR::Attribute attribute) {
-    if (attribute >= IR::Attribute::ColorFrontDiffuseR &&
-        attribute <= IR::Attribute::ColorFrontDiffuseA) {
-        return output_front_color;
-    }
-    if (attribute >= IR::Attribute::ColorFrontSpecularR &&
-        attribute <= IR::Attribute::ColorFrontSpecularA) {
-        return output_front_secondary_color;
-    }
-    if (attribute >= IR::Attribute::ColorBackDiffuseR &&
-        attribute <= IR::Attribute::ColorBackDiffuseA) {
-        return output_back_color;
-    }
-    if (attribute >= IR::Attribute::ColorBackSpecularR &&
-        attribute <= IR::Attribute::ColorBackSpecularA) {
-        return output_back_secondary_color;
-    }
-    if (attribute == IR::Attribute::FogCoordinate) {
-        return output_fog_frag_coord;
-    }
-    if (attribute >= IR::Attribute::FixedFncTexture0S &&
-        attribute <= IR::Attribute::FixedFncTexture9Q) {
-        u32 index =
-            (static_cast<u32>(attribute) - static_cast<u32>(IR::Attribute::FixedFncTexture0S)) / 4;
-        return output_fixed_fnc_textures[index];
-    }
-    throw InvalidArgument("Attribute is not legacy attribute {}", attribute);
-}
-
 void EmitContext::DefineCommonTypes(const Info& info) {
     void_id = TypeVoid();
 
@@ -1389,7 +1301,6 @@ void EmitContext::DefineInputs(const IR::Program& program) {
         loads[IR::Attribute::TessellationEvaluationPointV]) {
         tess_coord = DefineInput(*this, F32[3], false, spv::BuiltIn::TessCoord);
     }
-    std::bitset<IR::NUM_GENERICS> used_locations{};
     for (size_t index = 0; index < IR::NUM_GENERICS; ++index) {
         const AttributeType input_type{runtime_info.generic_input_types[index]};
         if (!runtime_info.previous_stage_stores.Generic(index)) {
@@ -1401,7 +1312,6 @@ void EmitContext::DefineInputs(const IR::Program& program) {
         if (input_type == AttributeType::Disabled) {
             continue;
         }
-        used_locations.set(index);
         const Id type{GetAttributeType(*this, input_type)};
         const Id id{DefineInput(*this, type, true)};
         Decorate(id, spv::Decoration::Location, static_cast<u32>(index));
@@ -1427,30 +1337,6 @@ void EmitContext::DefineInputs(const IR::Program& program) {
             break;
         }
     }
-    size_t previous_unused_location = 0;
-    if (loads.AnyComponent(IR::Attribute::ColorFrontDiffuseR)) {
-        input_front_color = DefineLegacyInput(*this, used_locations, previous_unused_location);
-    }
-    if (loads.AnyComponent(IR::Attribute::ColorFrontSpecularR)) {
-        input_front_secondary_color =
-            DefineLegacyInput(*this, used_locations, previous_unused_location);
-    }
-    if (loads.AnyComponent(IR::Attribute::ColorBackDiffuseR)) {
-        input_back_color = DefineLegacyInput(*this, used_locations, previous_unused_location);
-    }
-    if (loads.AnyComponent(IR::Attribute::ColorBackSpecularR)) {
-        input_back_secondary_color =
-            DefineLegacyInput(*this, used_locations, previous_unused_location);
-    }
-    if (loads.AnyComponent(IR::Attribute::FogCoordinate)) {
-        input_fog_frag_coord = DefineLegacyInput(*this, used_locations, previous_unused_location);
-    }
-    for (size_t index = 0; index < NUM_FIXEDFNCTEXTURE; ++index) {
-        if (loads.AnyComponent(IR::Attribute::FixedFncTexture0S + index * 4)) {
-            input_fixed_fnc_textures[index] =
-                DefineLegacyInput(*this, used_locations, previous_unused_location);
-        }
-    }
     if (stage == Stage::TessellationEval) {
         for (size_t index = 0; index < info.uses_patches.size(); ++index) {
             if (!info.uses_patches[index]) {
@@ -1501,38 +1387,9 @@ void EmitContext::DefineOutputs(const IR::Program& program) {
         viewport_mask = DefineOutput(*this, TypeArray(U32[1], Const(1u)), std::nullopt,
                                      spv::BuiltIn::ViewportMaskNV);
     }
-    std::bitset<IR::NUM_GENERICS> used_locations{};
     for (size_t index = 0; index < IR::NUM_GENERICS; ++index) {
         if (info.stores.Generic(index)) {
             DefineGenericOutput(*this, index, invocations);
-            used_locations.set(index);
-        }
-    }
-    size_t previous_unused_location = 0;
-    if (info.stores.AnyComponent(IR::Attribute::ColorFrontDiffuseR)) {
-        output_front_color =
-            DefineLegacyOutput(*this, used_locations, previous_unused_location, invocations);
-    }
-    if (info.stores.AnyComponent(IR::Attribute::ColorFrontSpecularR)) {
-        output_front_secondary_color =
-            DefineLegacyOutput(*this, used_locations, previous_unused_location, invocations);
-    }
-    if (info.stores.AnyComponent(IR::Attribute::ColorBackDiffuseR)) {
-        output_back_color =
-            DefineLegacyOutput(*this, used_locations, previous_unused_location, invocations);
-    }
-    if (info.stores.AnyComponent(IR::Attribute::ColorBackSpecularR)) {
-        output_back_secondary_color =
-            DefineLegacyOutput(*this, used_locations, previous_unused_location, invocations);
-    }
-    if (info.stores.AnyComponent(IR::Attribute::FogCoordinate)) {
-        output_fog_frag_coord =
-            DefineLegacyOutput(*this, used_locations, previous_unused_location, invocations);
-    }
-    for (size_t index = 0; index < NUM_FIXEDFNCTEXTURE; ++index) {
-        if (info.stores.AnyComponent(IR::Attribute::FixedFncTexture0S + index * 4)) {
-            output_fixed_fnc_textures[index] =
-                DefineLegacyOutput(*this, used_locations, previous_unused_location, invocations);
         }
     }
     switch (stage) {

src/shader_recompiler/backend/spirv/spirv_emit_context.h

@@ -113,9 +113,6 @@ public:
     [[nodiscard]] Id BitOffset8(const IR::Value& offset);
     [[nodiscard]] Id BitOffset16(const IR::Value& offset);
 
-    Id InputLegacyAttribute(IR::Attribute attribute);
-    Id OutputLegacyAttribute(IR::Attribute attribute);
-
     Id Const(u32 value) {
         return Constant(U32[1], value);
     }
@@ -281,22 +278,10 @@ public:
     Id write_global_func_u32x4{};
 
     Id input_position{};
-    Id input_front_color{};
-    Id input_front_secondary_color{};
-    Id input_back_color{};
-    Id input_back_secondary_color{};
-    Id input_fog_frag_coord{};
-    std::array<Id, 10> input_fixed_fnc_textures{};
     std::array<Id, 32> input_generics{};
 
     Id output_point_size{};
     Id output_position{};
-    Id output_front_color{};
-    Id output_front_secondary_color{};
-    Id output_back_color{};
-    Id output_back_secondary_color{};
-    Id output_fog_frag_coord{};
-    std::array<Id, 10> output_fixed_fnc_textures{};
     std::array<std::array<GenericElementInfo, 4>, 32> output_generics{};
 
     Id output_tess_level_outer{};

src/shader_recompiler/environment.h

@@ -31,6 +31,8 @@ public:
 
     [[nodiscard]] virtual std::array<u32, 3> WorkgroupSize() const = 0;
 
+    virtual void Dump(u64 hash) = 0;
+
     [[nodiscard]] const ProgramHeader& SPH() const noexcept {
         return sph;
     }

src/shader_recompiler/frontend/ir/attribute.h

@@ -224,6 +224,8 @@ enum class Attribute : u64 {
 
 constexpr size_t NUM_GENERICS = 32;
 
+constexpr size_t NUM_FIXEDFNCTEXTURE = 10;
+
 [[nodiscard]] bool IsGeneric(Attribute attribute) noexcept;
 
 [[nodiscard]] u32 GenericAttributeIndex(Attribute attribute);

src/shader_recompiler/frontend/ir/opcodes.inc

@@ -40,6 +40,7 @@ OPCODE(GetCbufU32,                                          U32,            U32,
 OPCODE(GetCbufF32,                                          F32,            U32,            U32,                                                            )
 OPCODE(GetCbufU32x2,                                        U32x2,          U32,            U32,                                                            )
 OPCODE(GetAttribute,                                        F32,            Attribute,      U32,                                                            )
+OPCODE(GetAttributeU32,                                     U32,            Attribute,      U32,                                                            )
 OPCODE(SetAttribute,                                        Void,           Attribute,      F32,            U32,                                            )
 OPCODE(GetAttributeIndexed,                                 F32,            U32,            U32,                                                            )
 OPCODE(SetAttributeIndexed,                                 Void,           U32,            F32,            U32,                                            )

src/shader_recompiler/frontend/maxwell/translate_program.cpp

@@ -5,6 +5,7 @@
 #include <algorithm>
 #include <memory>
 #include <vector>
+#include <queue>
 
 #include "common/settings.h"
 #include "shader_recompiler/exception.h"
@@ -127,6 +128,42 @@ void AddNVNStorageBuffers(IR::Program& program) {
         });
     }
 }
+
+bool IsLegacyAttribute(IR::Attribute attribute) {
+    return (attribute >= IR::Attribute::ColorFrontDiffuseR &&
+            attribute <= IR::Attribute::ColorBackSpecularA) ||
+           attribute == IR::Attribute::FogCoordinate ||
+           (attribute >= IR::Attribute::FixedFncTexture0S &&
+            attribute <= IR::Attribute::FixedFncTexture9Q);
+}
+
+std::map<IR::Attribute, IR::Attribute> GenerateLegacyToGenericMappings(
+    const VaryingState& state, std::queue<IR::Attribute> ununsed_generics) {
+    std::map<IR::Attribute, IR::Attribute> mapping;
+    for (size_t index = 0; index < 4; ++index) {
+        auto attr = IR::Attribute::ColorFrontDiffuseR + index * 4;
+        if (state.AnyComponent(attr)) {
+            for (size_t i = 0; i < 4; ++i) {
+                mapping.insert({attr + i, ununsed_generics.front() + i});
+            }
+            ununsed_generics.pop();
+        }
+    }
+    if (state[IR::Attribute::FogCoordinate]) {
+        mapping.insert({IR::Attribute::FogCoordinate, ununsed_generics.front()});
+        ununsed_generics.pop();
+    }
+    for (size_t index = 0; index < IR::NUM_FIXEDFNCTEXTURE; ++index) {
+        auto attr = IR::Attribute::FixedFncTexture0S + index * 4;
+        if (state.AnyComponent(attr)) {
+            for (size_t i = 0; i < 4; ++i) {
+                mapping.insert({attr + i, ununsed_generics.front() + i});
+            }
+            ununsed_generics.pop();
+        }
+    }
+    return mapping;
+}
 } // Anonymous namespace
 
 IR::Program TranslateProgram(ObjectPool<IR::Inst>& inst_pool, ObjectPool<IR::Block>& block_pool,
@@ -226,4 +263,62 @@ IR::Program MergeDualVertexPrograms(IR::Program& vertex_a, IR::Program& vertex_b
     return result;
 }
 
+void ConvertLegacyToGeneric(IR::Program& program, const Shader::RuntimeInfo& runtime_info) {
+    auto& stores = program.info.stores;
+    if (stores.Legacy()) {
+        std::queue<IR::Attribute> ununsed_output_generics{};
+        for (size_t index = 0; index < IR::NUM_GENERICS; ++index) {
+            if (!stores.Generic(index)) {
+                ununsed_output_generics.push(IR::Attribute::Generic0X + index * 4);
+            }
+        }
+        auto mappings = GenerateLegacyToGenericMappings(stores, ununsed_output_generics);
+        for (IR::Block* const block : program.post_order_blocks) {
+            for (IR::Inst& inst : block->Instructions()) {
+                switch (inst.GetOpcode()) {
+                case IR::Opcode::SetAttribute: {
+                    const auto attr = inst.Arg(0).Attribute();
+                    if (IsLegacyAttribute(attr)) {
+                        stores.Set(mappings[attr], true);
+                        inst.SetArg(0, Shader::IR::Value(mappings[attr]));
+                    }
+                    break;
+                }
+                default:
+                    break;
+                }
+            }
+        }
+    }
+
+    auto& loads = program.info.loads;
+    if (loads.Legacy()) {
+        std::queue<IR::Attribute> ununsed_input_generics{};
+        for (size_t index = 0; index < IR::NUM_GENERICS; ++index) {
+            const AttributeType input_type{runtime_info.generic_input_types[index]};
+            if (!runtime_info.previous_stage_stores.Generic(index) || !loads.Generic(index) ||
+                input_type == AttributeType::Disabled) {
+                ununsed_input_generics.push(IR::Attribute::Generic0X + index * 4);
+            }
+        }
+        auto mappings = GenerateLegacyToGenericMappings(loads, ununsed_input_generics);
+        for (IR::Block* const block : program.post_order_blocks) {
+            for (IR::Inst& inst : block->Instructions()) {
+                switch (inst.GetOpcode()) {
+                case IR::Opcode::GetAttribute: {
+                    const auto attr = inst.Arg(0).Attribute();
+                    if (IsLegacyAttribute(attr)) {
+                        loads.Set(mappings[attr], true);
+                        inst.SetArg(0, Shader::IR::Value(mappings[attr]));
+                    }
+                    break;
+                }
+                default:
+                    break;
+                }
+            }
+        }
+    }
+}
+
 } // namespace Shader::Maxwell

src/shader_recompiler/frontend/maxwell/translate_program.h

@@ -10,6 +10,7 @@
 #include "shader_recompiler/frontend/maxwell/control_flow.h"
 #include "shader_recompiler/host_translate_info.h"
 #include "shader_recompiler/object_pool.h"
+#include "shader_recompiler/runtime_info.h"
 
 namespace Shader::Maxwell {
 
@@ -20,4 +21,7 @@ namespace Shader::Maxwell {
 [[nodiscard]] IR::Program MergeDualVertexPrograms(IR::Program& vertex_a, IR::Program& vertex_b,
                                                   Environment& env_vertex_b);
 
+[[nodiscard]] void ConvertLegacyToGeneric(IR::Program& program,
+                                          const Shader::RuntimeInfo& runtime_info);
+
 } // namespace Shader::Maxwell

src/shader_recompiler/ir_opt/collect_shader_info_pass.cpp

@@ -389,6 +389,7 @@ void VisitUsages(Info& info, IR::Inst& inst) {
         info.uses_demote_to_helper_invocation = true;
         break;
     case IR::Opcode::GetAttribute:
+    case IR::Opcode::GetAttributeU32:
         info.loads.mask[static_cast<size_t>(inst.Arg(0).Attribute())] = true;
         break;
     case IR::Opcode::SetAttribute:

src/shader_recompiler/ir_opt/constant_propagation_pass.cpp

@@ -505,6 +505,29 @@ void FoldBitCast(IR::Inst& inst, IR::Opcode reverse) {
             return;
         }
     }
+    if constexpr (op == IR::Opcode::BitCastU32F32) {
+        // Workaround for new NVIDIA driver bug, where:
+        // uint attr = ftou(itof(gl_InstanceID));
+        // always returned 0.
+        // We can instead manually optimize this and work around the driver bug:
+        // uint attr = uint(gl_InstanceID);
+        if (arg_inst->GetOpcode() == IR::Opcode::GetAttribute) {
+            const IR::Attribute attr{arg_inst->Arg(0).Attribute()};
+            switch (attr) {
+            case IR::Attribute::PrimitiveId:
+            case IR::Attribute::InstanceId:
+            case IR::Attribute::VertexId:
+                break;
+            default:
+                return;
+            }
+            // Replace the bitcasts with an integer attribute get
+            inst.ReplaceOpcode(IR::Opcode::GetAttributeU32);
+            inst.SetArg(0, arg_inst->Arg(0));
+            inst.SetArg(1, arg_inst->Arg(1));
+            return;
+        }
+    }
 }
 
 void FoldInverseFunc(IR::Inst& inst, IR::Opcode reverse) {

src/shader_recompiler/profile.h

@@ -65,6 +65,10 @@ struct Profile {
     bool has_gl_component_indexing_bug{};
     /// The precise type qualifier is broken in the fragment stage of some drivers
     bool has_gl_precise_bug{};
+    /// Some drivers do not properly support floatBitsToUint when used on cbufs
+    bool has_gl_cbuf_ftou_bug{};
+    /// Some drivers poorly optimize boolean variable references
+    bool has_gl_bool_ref_bug{};
     /// Ignores SPIR-V ordered vs unordered using GLSL semantics
     bool ignore_nan_fp_comparisons{};
 

src/shader_recompiler/varying_state.h

@@ -53,7 +53,8 @@ struct VaryingState {
         return AnyComponent(IR::Attribute::ColorFrontDiffuseR) ||
                AnyComponent(IR::Attribute::ColorFrontSpecularR) ||
                AnyComponent(IR::Attribute::ColorBackDiffuseR) ||
-               AnyComponent(IR::Attribute::ColorBackSpecularR) || FixedFunctionTexture();
+               AnyComponent(IR::Attribute::ColorBackSpecularR) || FixedFunctionTexture() ||
+               mask[static_cast<size_t>(IR::Attribute::FogCoordinate)];
     }
 
     [[nodiscard]] bool FixedFunctionTexture() const noexcept {

src/tests/CMakeLists.txt

@@ -10,11 +10,12 @@ add_executable(tests
     core/network/network.cpp
     tests.cpp
     video_core/buffer_base.cpp
+    input_common/calibration_configuration_job.cpp
 )
 
 create_target_directory_groups(tests)
 
-target_link_libraries(tests PRIVATE common core)
+target_link_libraries(tests PRIVATE common core input_common)
 target_link_libraries(tests PRIVATE ${PLATFORM_LIBRARIES} catch-single-include Threads::Threads)
 
 add_test(NAME tests COMMAND tests)

src/tests/input_common/calibration_configuration_job.cpp

@@ -0,0 +1,136 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <array>
+#include <string>
+#include <thread>
+#include <boost/asio.hpp>
+#include <boost/crc.hpp>
+#include <catch2/catch.hpp>
+
+#include "input_common/drivers/udp_client.h"
+#include "input_common/helpers/udp_protocol.h"
+
+class FakeCemuhookServer {
+public:
+    FakeCemuhookServer()
+        : socket(io_service, boost::asio::ip::udp::endpoint(boost::asio::ip::udp::v4(), 0)) {}
+
+    ~FakeCemuhookServer() {
+        is_running = false;
+        boost::system::error_code error_code;
+        socket.shutdown(boost::asio::socket_base::shutdown_both, error_code);
+        socket.close();
+        if (handler.joinable()) {
+            handler.join();
+        }
+    }
+
+    u16 GetPort() {
+        return socket.local_endpoint().port();
+    }
+
+    std::string GetHost() {
+        return socket.local_endpoint().address().to_string();
+    }
+
+    void Run(const std::vector<InputCommon::CemuhookUDP::Response::TouchPad> touch_movement_path) {
+        constexpr size_t HeaderSize = sizeof(InputCommon::CemuhookUDP::Header);
+        constexpr size_t PadDataSize =
+            sizeof(InputCommon::CemuhookUDP::Message<InputCommon::CemuhookUDP::Response::PadData>);
+
+        REQUIRE(touch_movement_path.size() > 0);
+        is_running = true;
+        handler = std::thread([touch_movement_path, this]() {
+            auto current_touch_position = touch_movement_path.begin();
+            while (is_running) {
+                boost::asio::ip::udp::endpoint sender_endpoint;
+                boost::system::error_code error_code;
+                auto received_size = socket.receive_from(boost::asio::buffer(receive_buffer),
+                                                         sender_endpoint, 0, error_code);
+
+                if (received_size < HeaderSize) {
+                    continue;
+                }
+
+                InputCommon::CemuhookUDP::Header header{};
+                std::memcpy(&header, receive_buffer.data(), HeaderSize);
+                switch (header.type) {
+                case InputCommon::CemuhookUDP::Type::PadData: {
+                    InputCommon::CemuhookUDP::Response::PadData pad_data{};
+                    pad_data.touch[0] = *current_touch_position;
+                    const auto pad_message = InputCommon::CemuhookUDP::CreateMessage(
+                        InputCommon::CemuhookUDP::SERVER_MAGIC, pad_data, 0);
+                    std::memcpy(send_buffer.data(), &pad_message, PadDataSize);
+                    socket.send_to(boost::asio::buffer(send_buffer, PadDataSize), sender_endpoint,
+                                   0, error_code);
+
+                    bool can_advance =
+                        std::next(current_touch_position) != touch_movement_path.end();
+                    if (can_advance) {
+                        std::advance(current_touch_position, 1);
+                    }
+                    break;
+                }
+                case InputCommon::CemuhookUDP::Type::PortInfo:
+                case InputCommon::CemuhookUDP::Type::Version:
+                default:
+                    break;
+                }
+            }
+        });
+    }
+
+private:
+    boost::asio::io_service io_service;
+    boost::asio::ip::udp::socket socket;
+    std::array<u8, InputCommon::CemuhookUDP::MAX_PACKET_SIZE> send_buffer;
+    std::array<u8, InputCommon::CemuhookUDP::MAX_PACKET_SIZE> receive_buffer;
+    bool is_running = false;
+    std::thread handler;
+};
+
+TEST_CASE("CalibrationConfigurationJob completed", "[input_common]") {
+    Common::Event complete_event;
+    FakeCemuhookServer server;
+    server.Run({{
+                    .is_active = 1,
+                    .x = 0,
+                    .y = 0,
+                },
+                {
+                    .is_active = 1,
+                    .x = 200,
+                    .y = 200,
+                }});
+
+    InputCommon::CemuhookUDP::CalibrationConfigurationJob::Status status{};
+    u16 min_x{};
+    u16 min_y{};
+    u16 max_x{};
+    u16 max_y{};
+    InputCommon::CemuhookUDP::CalibrationConfigurationJob job(
+        server.GetHost(), server.GetPort(),
+        [&status,
+         &complete_event](InputCommon::CemuhookUDP::CalibrationConfigurationJob::Status status_) {
+            status = status_;
+            if (status ==
+                InputCommon::CemuhookUDP::CalibrationConfigurationJob::Status::Completed) {
+                complete_event.Set();
+            }
+        },
+        [&](u16 min_x_, u16 min_y_, u16 max_x_, u16 max_y_) {
+            min_x = min_x_;
+            min_y = min_y_;
+            max_x = max_x_;
+            max_y = max_y_;
+        });
+
+    complete_event.WaitUntil(std::chrono::system_clock::now() + std::chrono::seconds(10));
+    REQUIRE(status == InputCommon::CemuhookUDP::CalibrationConfigurationJob::Status::Completed);
+    REQUIRE(min_x == 0);
+    REQUIRE(min_y == 0);
+    REQUIRE(max_x == 200);
+    REQUIRE(max_y == 200);
+}

src/video_core/command_classes/codecs/codec.cpp

@@ -32,7 +32,7 @@ constexpr std::array PREFERRED_GPU_DECODERS = {
 #ifdef _WIN32
     AV_HWDEVICE_TYPE_D3D11VA,
     AV_HWDEVICE_TYPE_DXVA2,
-#elif defined(__linux__)
+#elif defined(__unix__)
     AV_HWDEVICE_TYPE_VAAPI,
     AV_HWDEVICE_TYPE_VDPAU,
 #endif

src/video_core/renderer_opengl/gl_device.cpp

@@ -182,17 +182,13 @@ Device::Device() {
         shader_backend = Settings::ShaderBackend::GLSL;
     }
 
-    if (shader_backend == Settings::ShaderBackend::GLSL && is_nvidia &&
-        !Settings::values.renderer_debug) {
+    if (shader_backend == Settings::ShaderBackend::GLSL && is_nvidia) {
         const std::string_view driver_version = version.substr(13);
         const int version_major =
             std::atoi(driver_version.substr(0, driver_version.find(".")).data());
-
         if (version_major >= 495) {
-            LOG_WARNING(Render_OpenGL, "NVIDIA drivers 495 and later causes significant problems "
-                                       "with yuzu. Forcing GLASM as a mitigation.");
-            shader_backend = Settings::ShaderBackend::GLASM;
-            use_assembly_shaders = true;
+            has_cbuf_ftou_bug = true;
+            has_bool_ref_bug = true;
         }
     }
 

src/video_core/renderer_opengl/gl_device.h

@@ -152,6 +152,14 @@ public:
         return need_fastmath_off;
     }
 
+    bool HasCbufFtouBug() const {
+        return has_cbuf_ftou_bug;
+    }
+
+    bool HasBoolRefBug() const {
+        return has_bool_ref_bug;
+    }
+
     Settings::ShaderBackend GetShaderBackend() const {
         return shader_backend;
     }
@@ -200,6 +208,8 @@ private:
     bool has_sparse_texture_2{};
     bool warp_size_potentially_larger_than_guest{};
     bool need_fastmath_off{};
+    bool has_cbuf_ftou_bug{};
+    bool has_bool_ref_bug{};
 
     std::string vendor_name;
 };

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -42,6 +42,7 @@ namespace {
 using Shader::Backend::GLASM::EmitGLASM;
 using Shader::Backend::GLSL::EmitGLSL;
 using Shader::Backend::SPIRV::EmitSPIRV;
+using Shader::Maxwell::ConvertLegacyToGeneric;
 using Shader::Maxwell::MergeDualVertexPrograms;
 using Shader::Maxwell::TranslateProgram;
 using VideoCommon::ComputeEnvironment;
@@ -213,6 +214,8 @@ ShaderCache::ShaderCache(RasterizerOpenGL& rasterizer_, Core::Frontend::EmuWindo
           .has_broken_fp16_float_controls = false,
           .has_gl_component_indexing_bug = device.HasComponentIndexingBug(),
           .has_gl_precise_bug = device.HasPreciseBug(),
+          .has_gl_cbuf_ftou_bug = device.HasCbufFtouBug(),
+          .has_gl_bool_ref_bug = device.HasBoolRefBug(),
           .ignore_nan_fp_comparisons = true,
           .gl_max_compute_smem_size = device.GetMaxComputeSharedMemorySize(),
       },
@@ -422,6 +425,11 @@ std::unique_ptr<GraphicsPipeline> ShaderCache::CreateGraphicsPipeline(
 
         const u32 cfg_offset{static_cast<u32>(env.StartAddress() + sizeof(Shader::ProgramHeader))};
         Shader::Maxwell::Flow::CFG cfg(env, pools.flow_block, cfg_offset, index == 0);
+
+        if (Settings::values.dump_shaders) {
+            env.Dump(key.unique_hashes[index]);
+        }
+
         if (!uses_vertex_a || index != 1) {
             // Normal path
             programs[index] = TranslateProgram(pools.inst, pools.block, env, cfg, host_info);
@@ -462,12 +470,14 @@ std::unique_ptr<GraphicsPipeline> ShaderCache::CreateGraphicsPipeline(
             MakeRuntimeInfo(key, program, previous_program, glasm_use_storage_buffers, use_glasm)};
         switch (device.GetShaderBackend()) {
         case Settings::ShaderBackend::GLSL:
+            ConvertLegacyToGeneric(program, runtime_info);
             sources[stage_index] = EmitGLSL(profile, runtime_info, program, binding);
             break;
         case Settings::ShaderBackend::GLASM:
             sources[stage_index] = EmitGLASM(profile, runtime_info, program, binding);
             break;
         case Settings::ShaderBackend::SPIRV:
+            ConvertLegacyToGeneric(program, runtime_info);
             sources_spirv[stage_index] = EmitSPIRV(profile, runtime_info, program, binding);
             break;
         }
@@ -506,8 +516,12 @@ std::unique_ptr<ComputePipeline> ShaderCache::CreateComputePipeline(
     LOG_INFO(Render_OpenGL, "0x{:016x}", key.Hash());
 
     Shader::Maxwell::Flow::CFG cfg{env, pools.flow_block, env.StartAddress()};
-    auto program{TranslateProgram(pools.inst, pools.block, env, cfg, host_info)};
 
+    if (Settings::values.dump_shaders) {
+        env.Dump(key.Hash());
+    }
+
+    auto program{TranslateProgram(pools.inst, pools.block, env, cfg, host_info)};
     const u32 num_storage_buffers{Shader::NumDescriptors(program.info.storage_buffers_descriptors)};
     Shader::RuntimeInfo info;
     info.glasm_use_storage_buffers = num_storage_buffers <= device.GetMaxGLASMStorageBufferBlocks();

src/video_core/renderer_opengl/gl_texture_cache.cpp

@@ -1047,7 +1047,7 @@ bool Image::ScaleDown(bool ignore) {
 }
 
 ImageView::ImageView(TextureCacheRuntime& runtime, const VideoCommon::ImageViewInfo& info,
-                     ImageId image_id_, Image& image)
+                     ImageId image_id_, Image& image, const SlotVector<Image>&)
     : VideoCommon::ImageViewBase{info, image.info, image_id_}, views{runtime.null_image_views} {
     const Device& device = runtime.device;
     if (True(image.flags & ImageFlagBits::Converted)) {

src/video_core/renderer_opengl/gl_texture_cache.h

@@ -36,6 +36,7 @@ using VideoCommon::ImageViewType;
 using VideoCommon::NUM_RT;
 using VideoCommon::Region2D;
 using VideoCommon::RenderTargets;
+using VideoCommon::SlotVector;
 
 struct ImageBufferMap {
     ~ImageBufferMap();
@@ -92,7 +93,7 @@ public:
 
     void ReinterpretImage(Image& dst, Image& src, std::span<const VideoCommon::ImageCopy> copies);
 
-    void ConvertImage(Framebuffer* dst, ImageView& dst_view, ImageView& src_view, bool rescaled) {
+    void ConvertImage(Framebuffer* dst, ImageView& dst_view, ImageView& src_view) {
         UNIMPLEMENTED();
     }
 
@@ -234,7 +235,8 @@ class ImageView : public VideoCommon::ImageViewBase {
     friend Image;
 
 public:
-    explicit ImageView(TextureCacheRuntime&, const VideoCommon::ImageViewInfo&, ImageId, Image&);
+    explicit ImageView(TextureCacheRuntime&, const VideoCommon::ImageViewInfo&, ImageId, Image&,
+                       const SlotVector<Image>&);
     explicit ImageView(TextureCacheRuntime&, const VideoCommon::ImageInfo&,
                        const VideoCommon::ImageViewInfo&, GPUVAddr);
     explicit ImageView(TextureCacheRuntime&, const VideoCommon::ImageInfo& info,

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -437,39 +437,29 @@ void RendererOpenGL::DrawScreen(const Layout::FramebufferLayout& layout) {
 
         glBindTextureUnit(0, fxaa_texture.handle);
     }
-
-    // Set projection matrix
     const std::array ortho_matrix =
         MakeOrthographicMatrix(static_cast<float>(layout.width), static_cast<float>(layout.height));
 
-    GLuint fragment_handle;
-    const auto filter = Settings::values.scaling_filter.GetValue();
-    switch (filter) {
-    case Settings::ScalingFilter::NearestNeighbor:
-        fragment_handle = present_bilinear_fragment.handle;
-        break;
-    case Settings::ScalingFilter::Bilinear:
-        fragment_handle = present_bilinear_fragment.handle;
-        break;
-    case Settings::ScalingFilter::Bicubic:
-        fragment_handle = present_bicubic_fragment.handle;
-        break;
-    case Settings::ScalingFilter::Gaussian:
-        fragment_handle = present_gaussian_fragment.handle;
-        break;
-    case Settings::ScalingFilter::ScaleForce:
-        fragment_handle = present_scaleforce_fragment.handle;
-        break;
-    case Settings::ScalingFilter::Fsr:
-        LOG_WARNING(
-            Render_OpenGL,
-            "FidelityFX FSR Super Sampling is not supported in OpenGL, changing to ScaleForce");
-        fragment_handle = present_scaleforce_fragment.handle;
-        break;
-    default:
-        fragment_handle = present_bilinear_fragment.handle;
-        break;
-    }
+    const auto fragment_handle = [this]() {
+        switch (Settings::values.scaling_filter.GetValue()) {
+        case Settings::ScalingFilter::NearestNeighbor:
+        case Settings::ScalingFilter::Bilinear:
+            return present_bilinear_fragment.handle;
+        case Settings::ScalingFilter::Bicubic:
+            return present_bicubic_fragment.handle;
+        case Settings::ScalingFilter::Gaussian:
+            return present_gaussian_fragment.handle;
+        case Settings::ScalingFilter::ScaleForce:
+            return present_scaleforce_fragment.handle;
+        case Settings::ScalingFilter::Fsr:
+            LOG_WARNING(
+                Render_OpenGL,
+                "FidelityFX Super Resolution is not supported in OpenGL, changing to ScaleForce");
+            return present_scaleforce_fragment.handle;
+        default:
+            return present_bilinear_fragment.handle;
+        }
+    }();
     program_manager.BindPresentPrograms(present_vertex.handle, fragment_handle);
     glProgramUniformMatrix3x2fv(present_vertex.handle, ModelViewMatrixLocation, 1, GL_FALSE,
                                 ortho_matrix.data());

src/video_core/renderer_vulkan/blit_image.cpp

@@ -4,6 +4,7 @@
 
 #include <algorithm>
 
+#include "common/settings.h"
 #include "video_core/host_shaders/convert_abgr8_to_d24s8_frag_spv.h"
 #include "video_core/host_shaders/convert_d24s8_to_abgr8_frag_spv.h"
 #include "video_core/host_shaders/convert_depth_to_float_frag_spv.h"
@@ -335,6 +336,17 @@ void BindBlitState(vk::CommandBuffer cmdbuf, VkPipelineLayout layout, const Regi
     cmdbuf.SetScissor(0, scissor);
     cmdbuf.PushConstants(layout, VK_SHADER_STAGE_VERTEX_BIT, push_constants);
 }
+
+VkExtent2D GetConversionExtent(const ImageView& src_image_view) {
+    const auto& resolution = Settings::values.resolution_info;
+    const bool is_rescaled = src_image_view.IsRescaled();
+    u32 width = src_image_view.size.width;
+    u32 height = src_image_view.size.height;
+    return VkExtent2D{
+        .width = is_rescaled ? resolution.ScaleUp(width) : width,
+        .height = is_rescaled ? resolution.ScaleUp(height) : height,
+    };
+}
 } // Anonymous namespace
 
 BlitImageHelper::BlitImageHelper(const Device& device_, VKScheduler& scheduler_,
@@ -425,108 +437,52 @@ void BlitImageHelper::BlitDepthStencil(const Framebuffer* dst_framebuffer,
 }
 
 void BlitImageHelper::ConvertD32ToR32(const Framebuffer* dst_framebuffer,
-                                      const ImageView& src_image_view, u32 up_scale,
-                                      u32 down_shift) {
+                                      const ImageView& src_image_view) {
     ConvertDepthToColorPipeline(convert_d32_to_r32_pipeline, dst_framebuffer->RenderPass());
-    Convert(*convert_d32_to_r32_pipeline, dst_framebuffer, src_image_view, up_scale, down_shift);
+    Convert(*convert_d32_to_r32_pipeline, dst_framebuffer, src_image_view);
 }
 
 void BlitImageHelper::ConvertR32ToD32(const Framebuffer* dst_framebuffer,
-                                      const ImageView& src_image_view, u32 up_scale,
-                                      u32 down_shift) {
+                                      const ImageView& src_image_view) {
     ConvertColorToDepthPipeline(convert_r32_to_d32_pipeline, dst_framebuffer->RenderPass());
-    Convert(*convert_r32_to_d32_pipeline, dst_framebuffer, src_image_view, up_scale, down_shift);
+    Convert(*convert_r32_to_d32_pipeline, dst_framebuffer, src_image_view);
 }
 
 void BlitImageHelper::ConvertD16ToR16(const Framebuffer* dst_framebuffer,
-                                      const ImageView& src_image_view, u32 up_scale,
-                                      u32 down_shift) {
+                                      const ImageView& src_image_view) {
     ConvertDepthToColorPipeline(convert_d16_to_r16_pipeline, dst_framebuffer->RenderPass());
-    Convert(*convert_d16_to_r16_pipeline, dst_framebuffer, src_image_view, up_scale, down_shift);
+    Convert(*convert_d16_to_r16_pipeline, dst_framebuffer, src_image_view);
 }
 
 void BlitImageHelper::ConvertR16ToD16(const Framebuffer* dst_framebuffer,
-                                      const ImageView& src_image_view, u32 up_scale,
-                                      u32 down_shift) {
+                                      const ImageView& src_image_view) {
     ConvertColorToDepthPipeline(convert_r16_to_d16_pipeline, dst_framebuffer->RenderPass());
-    Convert(*convert_r16_to_d16_pipeline, dst_framebuffer, src_image_view, up_scale, down_shift);
+    Convert(*convert_r16_to_d16_pipeline, dst_framebuffer, src_image_view);
 }
 
 void BlitImageHelper::ConvertABGR8ToD24S8(const Framebuffer* dst_framebuffer,
-                                          ImageView& src_image_view, u32 up_scale, u32 down_shift) {
+                                          const ImageView& src_image_view) {
     ConvertPipelineDepthTargetEx(convert_abgr8_to_d24s8_pipeline, dst_framebuffer->RenderPass(),
-                                 convert_abgr8_to_d24s8_frag, true);
-    ConvertColor(*convert_abgr8_to_d24s8_pipeline, dst_framebuffer, src_image_view, up_scale,
-                 down_shift);
+                                 convert_abgr8_to_d24s8_frag);
+    Convert(*convert_abgr8_to_d24s8_pipeline, dst_framebuffer, src_image_view);
 }
 
 void BlitImageHelper::ConvertD24S8ToABGR8(const Framebuffer* dst_framebuffer,
-                                          ImageView& src_image_view, u32 up_scale, u32 down_shift) {
+                                          ImageView& src_image_view) {
     ConvertPipelineColorTargetEx(convert_d24s8_to_abgr8_pipeline, dst_framebuffer->RenderPass(),
-                                 convert_d24s8_to_abgr8_frag, false);
-    ConvertDepthStencil(*convert_d24s8_to_abgr8_pipeline, dst_framebuffer, src_image_view, up_scale,
-                        down_shift);
+                                 convert_d24s8_to_abgr8_frag);
+    ConvertDepthStencil(*convert_d24s8_to_abgr8_pipeline, dst_framebuffer, src_image_view);
 }
 
 void BlitImageHelper::Convert(VkPipeline pipeline, const Framebuffer* dst_framebuffer,
-                              const ImageView& src_image_view, u32 up_scale, u32 down_shift) {
+                              const ImageView& src_image_view) {
     const VkPipelineLayout layout = *one_texture_pipeline_layout;
     const VkImageView src_view = src_image_view.Handle(Shader::TextureType::Color2D);
     const VkSampler sampler = *nearest_sampler;
-    const VkExtent2D extent{
-        .width = std::max((src_image_view.size.width * up_scale) >> down_shift, 1U),
-        .height = std::max((src_image_view.size.height * up_scale) >> down_shift, 1U),
-    };
-    scheduler.RequestRenderpass(dst_framebuffer);
-    scheduler.Record([pipeline, layout, sampler, src_view, extent, up_scale, down_shift,
-                      this](vk::CommandBuffer cmdbuf) {
-        const VkOffset2D offset{
-            .x = 0,
-            .y = 0,
-        };
-        const VkViewport viewport{
-            .x = 0.0f,
-            .y = 0.0f,
-            .width = static_cast<float>(extent.width),
-            .height = static_cast<float>(extent.height),
-            .minDepth = 0.0f,
-            .maxDepth = 0.0f,
-        };
-        const VkRect2D scissor{
-            .offset = offset,
-            .extent = extent,
-        };
-        const PushConstants push_constants{
-            .tex_scale = {viewport.width, viewport.height},
-            .tex_offset = {0.0f, 0.0f},
-        };
-        const VkDescriptorSet descriptor_set = one_texture_descriptor_allocator.Commit();
-        UpdateOneTextureDescriptorSet(device, descriptor_set, sampler, src_view);
+    const VkExtent2D extent = GetConversionExtent(src_image_view);
 
-        // TODO: Barriers
-        cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, pipeline);
-        cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, layout, 0, descriptor_set,
-                                  nullptr);
-        cmdbuf.SetViewport(0, viewport);
-        cmdbuf.SetScissor(0, scissor);
-        cmdbuf.PushConstants(layout, VK_SHADER_STAGE_VERTEX_BIT, push_constants);
-        cmdbuf.Draw(3, 1, 0, 0);
-    });
-    scheduler.InvalidateState();
-}
-
-void BlitImageHelper::ConvertColor(VkPipeline pipeline, const Framebuffer* dst_framebuffer,
-                                   ImageView& src_image_view, u32 up_scale, u32 down_shift) {
-    const VkPipelineLayout layout = *one_texture_pipeline_layout;
-    const VkImageView src_view = src_image_view.ColorView();
-    const VkSampler sampler = *nearest_sampler;
-    const VkExtent2D extent{
-        .width = std::max((src_image_view.size.width * up_scale) >> down_shift, 1U),
-        .height = std::max((src_image_view.size.height * up_scale) >> down_shift, 1U),
-    };
     scheduler.RequestRenderpass(dst_framebuffer);
-    scheduler.Record([pipeline, layout, sampler, src_view, extent, up_scale, down_shift,
-                      this](vk::CommandBuffer cmdbuf) {
+    scheduler.Record([pipeline, layout, sampler, src_view, extent, this](vk::CommandBuffer cmdbuf) {
         const VkOffset2D offset{
             .x = 0,
             .y = 0,
@@ -563,18 +519,16 @@ void BlitImageHelper::ConvertColor(VkPipeline pipeline, const Framebuffer* dst_f
 }
 
 void BlitImageHelper::ConvertDepthStencil(VkPipeline pipeline, const Framebuffer* dst_framebuffer,
-                                          ImageView& src_image_view, u32 up_scale, u32 down_shift) {
+                                          ImageView& src_image_view) {
     const VkPipelineLayout layout = *two_textures_pipeline_layout;
     const VkImageView src_depth_view = src_image_view.DepthView();
     const VkImageView src_stencil_view = src_image_view.StencilView();
     const VkSampler sampler = *nearest_sampler;
-    const VkExtent2D extent{
-        .width = std::max((src_image_view.size.width * up_scale) >> down_shift, 1U),
-        .height = std::max((src_image_view.size.height * up_scale) >> down_shift, 1U),
-    };
+    const VkExtent2D extent = GetConversionExtent(src_image_view);
+
     scheduler.RequestRenderpass(dst_framebuffer);
-    scheduler.Record([pipeline, layout, sampler, src_depth_view, src_stencil_view, extent, up_scale,
-                      down_shift, this](vk::CommandBuffer cmdbuf) {
+    scheduler.Record([pipeline, layout, sampler, src_depth_view, src_stencil_view, extent,
+                      this](vk::CommandBuffer cmdbuf) {
         const VkOffset2D offset{
             .x = 0,
             .y = 0,
@@ -695,11 +649,14 @@ VkPipeline BlitImageHelper::FindOrEmplaceDepthStencilPipeline(const BlitImagePip
     return *blit_depth_stencil_pipelines.back();
 }
 
-void BlitImageHelper::ConvertDepthToColorPipeline(vk::Pipeline& pipeline, VkRenderPass renderpass) {
+void BlitImageHelper::ConvertPipeline(vk::Pipeline& pipeline, VkRenderPass renderpass,
+                                      bool is_target_depth) {
     if (pipeline) {
         return;
     }
-    const std::array stages = MakeStages(*full_screen_vert, *convert_depth_to_float_frag);
+    VkShaderModule frag_shader =
+        is_target_depth ? *convert_float_to_depth_frag : *convert_depth_to_float_frag;
+    const std::array stages = MakeStages(*full_screen_vert, frag_shader);
     pipeline = device.GetLogical().CreateGraphicsPipeline({
         .sType = VK_STRUCTURE_TYPE_GRAPHICS_PIPELINE_CREATE_INFO,
         .pNext = nullptr,
@@ -712,8 +669,9 @@ void BlitImageHelper::ConvertDepthToColorPipeline(vk::Pipeline& pipeline, VkRend
         .pViewportState = &PIPELINE_VIEWPORT_STATE_CREATE_INFO,
         .pRasterizationState = &PIPELINE_RASTERIZATION_STATE_CREATE_INFO,
         .pMultisampleState = &PIPELINE_MULTISAMPLE_STATE_CREATE_INFO,
-        .pDepthStencilState = nullptr,
-        .pColorBlendState = &PIPELINE_COLOR_BLEND_STATE_GENERIC_CREATE_INFO,
+        .pDepthStencilState = is_target_depth ? &PIPELINE_DEPTH_STENCIL_STATE_CREATE_INFO : nullptr,
+        .pColorBlendState = is_target_depth ? &PIPELINE_COLOR_BLEND_STATE_EMPTY_CREATE_INFO
+                                            : &PIPELINE_COLOR_BLEND_STATE_GENERIC_CREATE_INFO,
         .pDynamicState = &PIPELINE_DYNAMIC_STATE_CREATE_INFO,
         .layout = *one_texture_pipeline_layout,
         .renderPass = renderpass,
@@ -723,37 +681,17 @@ void BlitImageHelper::ConvertDepthToColorPipeline(vk::Pipeline& pipeline, VkRend
     });
 }
 
+void BlitImageHelper::ConvertDepthToColorPipeline(vk::Pipeline& pipeline, VkRenderPass renderpass) {
+    ConvertPipeline(pipeline, renderpass, false);
+}
+
 void BlitImageHelper::ConvertColorToDepthPipeline(vk::Pipeline& pipeline, VkRenderPass renderpass) {
-    if (pipeline) {
-        return;
-    }
-    const std::array stages = MakeStages(*full_screen_vert, *convert_float_to_depth_frag);
-    pipeline = device.GetLogical().CreateGraphicsPipeline({
-        .sType = VK_STRUCTURE_TYPE_GRAPHICS_PIPELINE_CREATE_INFO,
-        .pNext = nullptr,
-        .flags = 0,
-        .stageCount = static_cast<u32>(stages.size()),
-        .pStages = stages.data(),
-        .pVertexInputState = &PIPELINE_VERTEX_INPUT_STATE_CREATE_INFO,
-        .pInputAssemblyState = &PIPELINE_INPUT_ASSEMBLY_STATE_CREATE_INFO,
-        .pTessellationState = nullptr,
-        .pViewportState = &PIPELINE_VIEWPORT_STATE_CREATE_INFO,
-        .pRasterizationState = &PIPELINE_RASTERIZATION_STATE_CREATE_INFO,
-        .pMultisampleState = &PIPELINE_MULTISAMPLE_STATE_CREATE_INFO,
-        .pDepthStencilState = &PIPELINE_DEPTH_STENCIL_STATE_CREATE_INFO,
-        .pColorBlendState = &PIPELINE_COLOR_BLEND_STATE_EMPTY_CREATE_INFO,
-        .pDynamicState = &PIPELINE_DYNAMIC_STATE_CREATE_INFO,
-        .layout = *one_texture_pipeline_layout,
-        .renderPass = renderpass,
-        .subpass = 0,
-        .basePipelineHandle = VK_NULL_HANDLE,
-        .basePipelineIndex = 0,
-    });
+    ConvertPipeline(pipeline, renderpass, true);
 }
 
 void BlitImageHelper::ConvertPipelineEx(vk::Pipeline& pipeline, VkRenderPass renderpass,
-                                        vk::ShaderModule& module, bool is_target_depth,
-                                        bool single_texture) {
+                                        vk::ShaderModule& module, bool single_texture,
+                                        bool is_target_depth) {
     if (pipeline) {
         return;
     }
@@ -782,13 +720,13 @@ void BlitImageHelper::ConvertPipelineEx(vk::Pipeline& pipeline, VkRenderPass ren
 }
 
 void BlitImageHelper::ConvertPipelineColorTargetEx(vk::Pipeline& pipeline, VkRenderPass renderpass,
-                                                   vk::ShaderModule& module, bool single_texture) {
-    ConvertPipelineEx(pipeline, renderpass, module, false, single_texture);
+                                                   vk::ShaderModule& module) {
+    ConvertPipelineEx(pipeline, renderpass, module, false, false);
 }
 
 void BlitImageHelper::ConvertPipelineDepthTargetEx(vk::Pipeline& pipeline, VkRenderPass renderpass,
-                                                   vk::ShaderModule& module, bool single_texture) {
-    ConvertPipelineEx(pipeline, renderpass, module, true, single_texture);
+                                                   vk::ShaderModule& module) {
+    ConvertPipelineEx(pipeline, renderpass, module, true, true);
 }
 
 } // namespace Vulkan

src/video_core/renderer_vulkan/blit_image.h

@@ -44,50 +44,43 @@ public:
                           const Region2D& src_region, Tegra::Engines::Fermi2D::Filter filter,
                           Tegra::Engines::Fermi2D::Operation operation);
 
-    void ConvertD32ToR32(const Framebuffer* dst_framebuffer, const ImageView& src_image_view,
-                         u32 up_scale, u32 down_shift);
+    void ConvertD32ToR32(const Framebuffer* dst_framebuffer, const ImageView& src_image_view);
 
-    void ConvertR32ToD32(const Framebuffer* dst_framebuffer, const ImageView& src_image_view,
-                         u32 up_scale, u32 down_shift);
+    void ConvertR32ToD32(const Framebuffer* dst_framebuffer, const ImageView& src_image_view);
 
-    void ConvertD16ToR16(const Framebuffer* dst_framebuffer, const ImageView& src_image_view,
-                         u32 up_scale, u32 down_shift);
+    void ConvertD16ToR16(const Framebuffer* dst_framebuffer, const ImageView& src_image_view);
 
-    void ConvertR16ToD16(const Framebuffer* dst_framebuffer, const ImageView& src_image_view,
-                         u32 up_scale, u32 down_shift);
+    void ConvertR16ToD16(const Framebuffer* dst_framebuffer, const ImageView& src_image_view);
 
-    void ConvertABGR8ToD24S8(const Framebuffer* dst_framebuffer, ImageView& src_image_view,
-                             u32 up_scale, u32 down_shift);
+    void ConvertABGR8ToD24S8(const Framebuffer* dst_framebuffer, const ImageView& src_image_view);
 
-    void ConvertD24S8ToABGR8(const Framebuffer* dst_framebuffer, ImageView& src_image_view,
-                             u32 up_scale, u32 down_shift);
+    void ConvertD24S8ToABGR8(const Framebuffer* dst_framebuffer, ImageView& src_image_view);
 
 private:
     void Convert(VkPipeline pipeline, const Framebuffer* dst_framebuffer,
-                 const ImageView& src_image_view, u32 up_scale, u32 down_shift);
-
-    void ConvertColor(VkPipeline pipeline, const Framebuffer* dst_framebuffer,
-                      ImageView& src_image_view, u32 up_scale, u32 down_shift);
+                 const ImageView& src_image_view);
 
     void ConvertDepthStencil(VkPipeline pipeline, const Framebuffer* dst_framebuffer,
-                             ImageView& src_image_view, u32 up_scale, u32 down_shift);
+                             ImageView& src_image_view);
 
     [[nodiscard]] VkPipeline FindOrEmplaceColorPipeline(const BlitImagePipelineKey& key);
 
     [[nodiscard]] VkPipeline FindOrEmplaceDepthStencilPipeline(const BlitImagePipelineKey& key);
 
+    void ConvertPipeline(vk::Pipeline& pipeline, VkRenderPass renderpass, bool is_target_depth);
+
     void ConvertDepthToColorPipeline(vk::Pipeline& pipeline, VkRenderPass renderpass);
 
     void ConvertColorToDepthPipeline(vk::Pipeline& pipeline, VkRenderPass renderpass);
 
     void ConvertPipelineEx(vk::Pipeline& pipeline, VkRenderPass renderpass,
-                           vk::ShaderModule& module, bool is_target_depth, bool single_texture);
+                           vk::ShaderModule& module, bool single_texture, bool is_target_depth);
 
     void ConvertPipelineColorTargetEx(vk::Pipeline& pipeline, VkRenderPass renderpass,
-                                      vk::ShaderModule& module, bool single_texture);
+                                      vk::ShaderModule& module);
 
     void ConvertPipelineDepthTargetEx(vk::Pipeline& pipeline, VkRenderPass renderpass,
-                                      vk::ShaderModule& module, bool single_texture);
+                                      vk::ShaderModule& module);
 
     const Device& device;
     VKScheduler& scheduler;

src/video_core/renderer_vulkan/vk_blit_screen.cpp

@@ -391,28 +391,23 @@ VkSemaphore VKBlitScreen::Draw(const Tegra::FramebufferConfig& framebuffer,
                 .offset = {0, 0},
                 .extent = size,
             };
-            const auto filter = Settings::values.scaling_filter.GetValue();
             cmdbuf.BeginRenderPass(renderpass_bi, VK_SUBPASS_CONTENTS_INLINE);
-            switch (filter) {
-            case Settings::ScalingFilter::NearestNeighbor:
-                cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, *bilinear_pipeline);
-                break;
-            case Settings::ScalingFilter::Bilinear:
-                cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, *bilinear_pipeline);
-                break;
-            case Settings::ScalingFilter::Bicubic:
-                cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, *bicubic_pipeline);
-                break;
-            case Settings::ScalingFilter::Gaussian:
-                cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, *gaussian_pipeline);
-                break;
-            case Settings::ScalingFilter::ScaleForce:
-                cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, *scaleforce_pipeline);
-                break;
-            default:
-                cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, *bilinear_pipeline);
-                break;
-            }
+            auto graphics_pipeline = [this]() {
+                switch (Settings::values.scaling_filter.GetValue()) {
+                case Settings::ScalingFilter::NearestNeighbor:
+                case Settings::ScalingFilter::Bilinear:
+                    return *bilinear_pipeline;
+                case Settings::ScalingFilter::Bicubic:
+                    return *bicubic_pipeline;
+                case Settings::ScalingFilter::Gaussian:
+                    return *gaussian_pipeline;
+                case Settings::ScalingFilter::ScaleForce:
+                    return *scaleforce_pipeline;
+                default:
+                    return *bilinear_pipeline;
+                }
+            }();
+            cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, graphics_pipeline);
             cmdbuf.SetViewport(0, viewport);
             cmdbuf.SetScissor(0, scissor);
 

src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp

@@ -605,7 +605,11 @@ void GraphicsPipeline::MakePipeline(VkRenderPass render_pass) {
         .flags = 0,
         .topology = input_assembly_topology,
         .primitiveRestartEnable = key.state.primitive_restart_enable != 0 &&
-                                  SupportsPrimitiveRestart(input_assembly_topology),
+                                  ((input_assembly_topology != VK_PRIMITIVE_TOPOLOGY_PATCH_LIST &&
+                                    device.IsTopologyListPrimitiveRestartSupported()) ||
+                                   SupportsPrimitiveRestart(input_assembly_topology) ||
+                                   (input_assembly_topology == VK_PRIMITIVE_TOPOLOGY_PATCH_LIST &&
+                                    device.IsPatchListPrimitiveRestartSupported())),
     };
     const VkPipelineTessellationStateCreateInfo tessellation_ci{
         .sType = VK_STRUCTURE_TYPE_PIPELINE_TESSELLATION_STATE_CREATE_INFO,
@@ -613,7 +617,6 @@ void GraphicsPipeline::MakePipeline(VkRenderPass render_pass) {
         .flags = 0,
         .patchControlPoints = key.state.patch_control_points_minus_one.Value() + 1,
     };
-
     std::array<VkViewportSwizzleNV, Maxwell::NumViewports> swizzles;
     std::ranges::transform(key.state.viewport_swizzles, swizzles.begin(), UnpackViewportSwizzle);
     const VkPipelineViewportSwizzleStateCreateInfoNV swizzle_ci{
@@ -748,8 +751,8 @@ void GraphicsPipeline::MakePipeline(VkRenderPass render_pass) {
         .sType = VK_STRUCTURE_TYPE_PIPELINE_COLOR_BLEND_STATE_CREATE_INFO,
         .pNext = nullptr,
         .flags = 0,
-        .logicOpEnable = VK_FALSE,
-        .logicOp = VK_LOGIC_OP_COPY,
+        .logicOpEnable = key.state.logic_op_enable != 0,
+        .logicOp = static_cast<VkLogicOp>(key.state.logic_op.Value()),
         .attachmentCount = static_cast<u32>(cb_attachments.size()),
         .pAttachments = cb_attachments.data(),
         .blendConstants = {},

src/video_core/renderer_vulkan/vk_pipeline_cache.cpp

@@ -48,6 +48,7 @@ MICROPROFILE_DECLARE(Vulkan_PipelineCache);
 
 namespace {
 using Shader::Backend::SPIRV::EmitSPIRV;
+using Shader::Maxwell::ConvertLegacyToGeneric;
 using Shader::Maxwell::MergeDualVertexPrograms;
 using Shader::Maxwell::TranslateProgram;
 using VideoCommon::ComputeEnvironment;
@@ -516,6 +517,9 @@ std::unique_ptr<GraphicsPipeline> PipelineCache::CreateGraphicsPipeline(
 
         const u32 cfg_offset{static_cast<u32>(env.StartAddress() + sizeof(Shader::ProgramHeader))};
         Shader::Maxwell::Flow::CFG cfg(env, pools.flow_block, cfg_offset, index == 0);
+        if (Settings::values.dump_shaders) {
+            env.Dump(key.unique_hashes[index]);
+        }
         if (!uses_vertex_a || index != 1) {
             // Normal path
             programs[index] = TranslateProgram(pools.inst, pools.block, env, cfg, host_info);
@@ -543,6 +547,7 @@ std::unique_ptr<GraphicsPipeline> PipelineCache::CreateGraphicsPipeline(
         infos[stage_index] = &program.info;
 
         const auto runtime_info{MakeRuntimeInfo(programs, key, program, previous_stage)};
+        ConvertLegacyToGeneric(program, runtime_info);
         const std::vector<u32> code{EmitSPIRV(profile, runtime_info, program, binding)};
         device.SaveShader(code);
         modules[stage_index] = BuildShader(device, code);
@@ -611,6 +616,12 @@ std::unique_ptr<ComputePipeline> PipelineCache::CreateComputePipeline(
     LOG_INFO(Render_Vulkan, "0x{:016x}", key.Hash());
 
     Shader::Maxwell::Flow::CFG cfg{env, pools.flow_block, env.StartAddress()};
+
+    // Dump it before error.
+    if (Settings::values.dump_shaders) {
+        env.Dump(key.Hash());
+    }
+
     auto program{TranslateProgram(pools.inst, pools.block, env, cfg, host_info)};
     const std::vector<u32> code{EmitSPIRV(profile, program)};
     device.SaveShader(code);

src/video_core/renderer_vulkan/vk_texture_cache.cpp

@@ -1057,37 +1057,37 @@ void TextureCacheRuntime::BlitImage(Framebuffer* dst_framebuffer, ImageView& dst
     });
 }
 
-void TextureCacheRuntime::ConvertImage(Framebuffer* dst, ImageView& dst_view, ImageView& src_view,
-                                       bool rescaled) {
-    const u32 up_scale = rescaled ? resolution.up_scale : 1;
-    const u32 down_shift = rescaled ? resolution.down_shift : 0;
+void TextureCacheRuntime::ConvertImage(Framebuffer* dst, ImageView& dst_view, ImageView& src_view) {
     switch (dst_view.format) {
     case PixelFormat::R16_UNORM:
         if (src_view.format == PixelFormat::D16_UNORM) {
-            return blit_image_helper.ConvertD16ToR16(dst, src_view, up_scale, down_shift);
+            return blit_image_helper.ConvertD16ToR16(dst, src_view);
         }
         break;
     case PixelFormat::A8B8G8R8_UNORM:
         if (src_view.format == PixelFormat::S8_UINT_D24_UNORM) {
-            return blit_image_helper.ConvertD24S8ToABGR8(dst, src_view, up_scale, down_shift);
+            return blit_image_helper.ConvertD24S8ToABGR8(dst, src_view);
         }
         break;
     case PixelFormat::R32_FLOAT:
         if (src_view.format == PixelFormat::D32_FLOAT) {
-            return blit_image_helper.ConvertD32ToR32(dst, src_view, up_scale, down_shift);
+            return blit_image_helper.ConvertD32ToR32(dst, src_view);
         }
         break;
     case PixelFormat::D16_UNORM:
         if (src_view.format == PixelFormat::R16_UNORM) {
-            return blit_image_helper.ConvertR16ToD16(dst, src_view, up_scale, down_shift);
+            return blit_image_helper.ConvertR16ToD16(dst, src_view);
         }
         break;
     case PixelFormat::S8_UINT_D24_UNORM:
-        return blit_image_helper.ConvertABGR8ToD24S8(dst, src_view, up_scale, down_shift);
+        if (src_view.format == PixelFormat::A8B8G8R8_UNORM ||
+            src_view.format == PixelFormat::B8G8R8A8_UNORM) {
+            return blit_image_helper.ConvertABGR8ToD24S8(dst, src_view);
+        }
         break;
     case PixelFormat::D32_FLOAT:
         if (src_view.format == PixelFormat::R32_FLOAT) {
-            return blit_image_helper.ConvertR32ToD32(dst, src_view, up_scale, down_shift);
+            return blit_image_helper.ConvertR32ToD32(dst, src_view);
         }
         break;
     default:
@@ -1329,6 +1329,10 @@ void Image::DownloadMemory(const StagingBufferRef& map, std::span<const BufferIm
     }
 }
 
+bool Image::IsRescaled() const noexcept {
+    return True(flags & ImageFlagBits::Rescaled);
+}
+
 bool Image::ScaleUp(bool ignore) {
     if (True(flags & ImageFlagBits::Rescaled)) {
         return false;
@@ -1340,7 +1344,6 @@ bool Image::ScaleUp(bool ignore) {
         return false;
     }
     has_scaled = true;
-    const auto& device = runtime->device;
     if (!scaled_image) {
         const bool is_2d = info.type == ImageType::e2D;
         const u32 scaled_width = resolution.ScaleUp(info.size.width);
@@ -1348,7 +1351,7 @@ bool Image::ScaleUp(bool ignore) {
         auto scaled_info = info;
         scaled_info.size.width = scaled_width;
         scaled_info.size.height = scaled_height;
-        scaled_image = MakeImage(device, scaled_info);
+        scaled_image = MakeImage(runtime->device, scaled_info);
         auto& allocator = runtime->memory_allocator;
         scaled_commit = MemoryCommit(allocator.Commit(scaled_image, MemoryUsage::DeviceLocal));
         ignore = false;
@@ -1357,18 +1360,13 @@ bool Image::ScaleUp(bool ignore) {
     if (ignore) {
         return true;
     }
-
     if (aspect_mask == 0) {
         aspect_mask = ImageAspectMask(info.format);
     }
-    static constexpr auto OPTIMAL_FORMAT = FormatType::Optimal;
-    const PixelFormat format = StorageFormat(info.format);
-    const auto vk_format = MaxwellToVK::SurfaceFormat(device, OPTIMAL_FORMAT, false, format).format;
-    const auto blit_usage = VK_FORMAT_FEATURE_BLIT_SRC_BIT | VK_FORMAT_FEATURE_BLIT_DST_BIT;
-    if (device.IsFormatSupported(vk_format, blit_usage, OPTIMAL_FORMAT)) {
-        BlitScale(*scheduler, *original_image, *scaled_image, info, aspect_mask, resolution);
-    } else {
+    if (NeedsScaleHelper()) {
         return BlitScaleHelper(true);
+    } else {
+        BlitScale(*scheduler, *original_image, *scaled_image, info, aspect_mask, resolution);
     }
     return true;
 }
@@ -1390,15 +1388,10 @@ bool Image::ScaleDown(bool ignore) {
     if (aspect_mask == 0) {
         aspect_mask = ImageAspectMask(info.format);
     }
-    static constexpr auto OPTIMAL_FORMAT = FormatType::Optimal;
-    const PixelFormat format = StorageFormat(info.format);
-    const auto& device = runtime->device;
-    const auto vk_format = MaxwellToVK::SurfaceFormat(device, OPTIMAL_FORMAT, false, format).format;
-    const auto blit_usage = VK_FORMAT_FEATURE_BLIT_SRC_BIT | VK_FORMAT_FEATURE_BLIT_DST_BIT;
-    if (device.IsFormatSupported(vk_format, blit_usage, OPTIMAL_FORMAT)) {
-        BlitScale(*scheduler, *scaled_image, *original_image, info, aspect_mask, resolution, false);
-    } else {
+    if (NeedsScaleHelper()) {
         return BlitScaleHelper(false);
+    } else {
+        BlitScale(*scheduler, *scaled_image, *original_image, info, aspect_mask, resolution, false);
     }
     return true;
 }
@@ -1466,10 +1459,24 @@ bool Image::BlitScaleHelper(bool scale_up) {
     return true;
 }
 
+bool Image::NeedsScaleHelper() const {
+    const auto& device = runtime->device;
+    const bool needs_msaa_helper = info.num_samples > 1 && device.CantBlitMSAA();
+    if (needs_msaa_helper) {
+        return true;
+    }
+    static constexpr auto OPTIMAL_FORMAT = FormatType::Optimal;
+    const PixelFormat format = StorageFormat(info.format);
+    const auto vk_format = MaxwellToVK::SurfaceFormat(device, OPTIMAL_FORMAT, false, format).format;
+    const auto blit_usage = VK_FORMAT_FEATURE_BLIT_SRC_BIT | VK_FORMAT_FEATURE_BLIT_DST_BIT;
+    const bool needs_blit_helper = !device.IsFormatSupported(vk_format, blit_usage, OPTIMAL_FORMAT);
+    return needs_blit_helper;
+}
+
 ImageView::ImageView(TextureCacheRuntime& runtime, const VideoCommon::ImageViewInfo& info,
                      ImageId image_id_, Image& image)
     : VideoCommon::ImageViewBase{info, image.info, image_id_}, device{&runtime.device},
-      image_handle{image.Handle()}, samples{ConvertSampleCount(image.info.num_samples)} {
+      image_handle{image.Handle()}, samples(ConvertSampleCount(image.info.num_samples)) {
     using Shader::TextureType;
 
     const VkImageAspectFlags aspect_mask = ImageViewAspectMask(info);
@@ -1552,6 +1559,12 @@ ImageView::ImageView(TextureCacheRuntime& runtime, const VideoCommon::ImageViewI
     }
 }
 
+ImageView::ImageView(TextureCacheRuntime& runtime, const VideoCommon::ImageViewInfo& info,
+                     ImageId image_id_, Image& image, const SlotVector<Image>& slot_imgs)
+    : ImageView{runtime, info, image_id_, image} {
+    slot_images = &slot_imgs;
+}
+
 ImageView::ImageView(TextureCacheRuntime&, const VideoCommon::ImageInfo& info,
                      const VideoCommon::ImageViewInfo& view_info, GPUVAddr gpu_addr_)
     : VideoCommon::ImageViewBase{info, view_info}, gpu_addr{gpu_addr_},
@@ -1607,6 +1620,15 @@ VkImageView ImageView::StorageView(Shader::TextureType texture_type,
     return *view;
 }
 
+bool ImageView::IsRescaled() const noexcept {
+    if (!slot_images) {
+        return false;
+    }
+    const auto& slots = *slot_images;
+    const auto& src_image = slots[image_id];
+    return src_image.IsRescaled();
+}
+
 vk::ImageView ImageView::MakeView(VkFormat vk_format, VkImageAspectFlags aspect_mask) {
     return device->GetLogical().CreateImageView({
         .sType = VK_STRUCTURE_TYPE_IMAGE_VIEW_CREATE_INFO,

src/video_core/renderer_vulkan/vk_texture_cache.h

@@ -23,6 +23,7 @@ using VideoCommon::ImageId;
 using VideoCommon::NUM_RT;
 using VideoCommon::Region2D;
 using VideoCommon::RenderTargets;
+using VideoCommon::SlotVector;
 using VideoCore::Surface::PixelFormat;
 
 class ASTCDecoderPass;
@@ -65,7 +66,7 @@ public:
 
     void ReinterpretImage(Image& dst, Image& src, std::span<const VideoCommon::ImageCopy> copies);
 
-    void ConvertImage(Framebuffer* dst, ImageView& dst_view, ImageView& src_view, bool rescaled);
+    void ConvertImage(Framebuffer* dst, ImageView& dst_view, ImageView& src_view);
 
     bool CanAccelerateImageUpload(Image&) const noexcept {
         return false;
@@ -139,6 +140,8 @@ public:
         return std::exchange(initialized, true);
     }
 
+    bool IsRescaled() const noexcept;
+
     bool ScaleUp(bool ignore = false);
 
     bool ScaleDown(bool ignore = false);
@@ -146,6 +149,8 @@ public:
 private:
     bool BlitScaleHelper(bool scale_up);
 
+    bool NeedsScaleHelper() const;
+
     VKScheduler* scheduler{};
     TextureCacheRuntime* runtime{};
 
@@ -168,6 +173,8 @@ private:
 class ImageView : public VideoCommon::ImageViewBase {
 public:
     explicit ImageView(TextureCacheRuntime&, const VideoCommon::ImageViewInfo&, ImageId, Image&);
+    explicit ImageView(TextureCacheRuntime&, const VideoCommon::ImageViewInfo&, ImageId, Image&,
+                       const SlotVector<Image>&);
     explicit ImageView(TextureCacheRuntime&, const VideoCommon::ImageInfo&,
                        const VideoCommon::ImageViewInfo&, GPUVAddr);
     explicit ImageView(TextureCacheRuntime&, const VideoCommon::NullImageViewParams&);
@@ -189,6 +196,8 @@ public:
     [[nodiscard]] VkImageView StorageView(Shader::TextureType texture_type,
                                           Shader::ImageFormat image_format);
 
+    [[nodiscard]] bool IsRescaled() const noexcept;
+
     [[nodiscard]] VkImageView Handle(Shader::TextureType texture_type) const noexcept {
         return *image_views[static_cast<size_t>(texture_type)];
     }
@@ -222,6 +231,8 @@ private:
     [[nodiscard]] vk::ImageView MakeView(VkFormat vk_format, VkImageAspectFlags aspect_mask);
 
     const Device* device = nullptr;
+    const SlotVector<Image>* slot_images = nullptr;
+
     std::array<vk::ImageView, Shader::NUM_TEXTURE_TYPES> image_views;
     std::unique_ptr<StorageViews> storage_views;
     vk::ImageView depth_view;

src/video_core/shader_environment.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <algorithm>
+#include <bit>
 #include <filesystem>
 #include <fstream>
 #include <memory>
@@ -14,6 +15,7 @@
 #include "common/common_types.h"
 #include "common/div_ceil.h"
 #include "common/fs/fs.h"
+#include "common/fs/path_util.h"
 #include "common/logging/log.h"
 #include "shader_recompiler/environment.h"
 #include "video_core/engines/kepler_compute.h"
@@ -57,6 +59,47 @@ static Shader::TextureType ConvertType(const Tegra::Texture::TICEntry& entry) {
     }
 }
 
+static std::string_view StageToPrefix(Shader::Stage stage) {
+    switch (stage) {
+    case Shader::Stage::VertexB:
+        return "VB";
+    case Shader::Stage::TessellationControl:
+        return "TC";
+    case Shader::Stage::TessellationEval:
+        return "TE";
+    case Shader::Stage::Geometry:
+        return "GS";
+    case Shader::Stage::Fragment:
+        return "FS";
+    case Shader::Stage::Compute:
+        return "CS";
+    case Shader::Stage::VertexA:
+        return "VA";
+    default:
+        return "UK";
+    }
+}
+
+static void DumpImpl(u64 hash, const u64* code, u32 read_highest, u32 read_lowest,
+                     u32 initial_offset, Shader::Stage stage) {
+    const auto shader_dir{Common::FS::GetYuzuPath(Common::FS::YuzuPath::DumpDir)};
+    const auto base_dir{shader_dir / "shaders"};
+    if (!Common::FS::CreateDir(shader_dir) || !Common::FS::CreateDir(base_dir)) {
+        LOG_ERROR(Common_Filesystem, "Failed to create shader dump directories");
+        return;
+    }
+    const auto prefix = StageToPrefix(stage);
+    const auto name{base_dir / fmt::format("{}{:016x}.ash", prefix, hash)};
+    const size_t real_size = read_highest - read_lowest + initial_offset;
+    const size_t padding_needed = ((32 - (real_size % 32)) % 32);
+    std::fstream shader_file(name, std::ios::out | std::ios::binary);
+    const size_t jump_index = initial_offset / sizeof(u64);
+    shader_file.write(reinterpret_cast<const char*>(code + jump_index), real_size);
+    for (size_t i = 0; i < padding_needed; i++) {
+        shader_file.put(0);
+    }
+}
+
 GenericEnvironment::GenericEnvironment(Tegra::MemoryManager& gpu_memory_, GPUVAddr program_base_,
                                        u32 start_address_)
     : gpu_memory{&gpu_memory_}, program_base{program_base_} {
@@ -128,6 +171,10 @@ u64 GenericEnvironment::CalculateHash() const {
     return Common::CityHash64(data.get(), size);
 }
 
+void GenericEnvironment::Dump(u64 hash) {
+    DumpImpl(hash, code.data(), read_highest, read_lowest, initial_offset, stage);
+}
+
 void GenericEnvironment::Serialize(std::ofstream& file) const {
     const u64 code_size{static_cast<u64>(CachedSize())};
     const u64 num_texture_types{static_cast<u64>(texture_types.size())};
@@ -207,6 +254,7 @@ GraphicsEnvironment::GraphicsEnvironment(Tegra::Engines::Maxwell3D& maxwell3d_,
                                          u32 start_address_)
     : GenericEnvironment{gpu_memory_, program_base_, start_address_}, maxwell3d{&maxwell3d_} {
     gpu_memory->ReadBlock(program_base + start_address, &sph, sizeof(sph));
+    initial_offset = sizeof(sph);
     gp_passthrough_mask = maxwell3d->regs.gp_passthrough_mask;
     switch (program) {
     case Maxwell::ShaderProgram::VertexA:
@@ -323,14 +371,20 @@ void FileEnvironment::Deserialize(std::ifstream& file) {
     if (stage == Shader::Stage::Compute) {
         file.read(reinterpret_cast<char*>(&workgroup_size), sizeof(workgroup_size))
             .read(reinterpret_cast<char*>(&shared_memory_size), sizeof(shared_memory_size));
+        initial_offset = 0;
     } else {
         file.read(reinterpret_cast<char*>(&sph), sizeof(sph));
+        initial_offset = sizeof(sph);
         if (stage == Shader::Stage::Geometry) {
             file.read(reinterpret_cast<char*>(&gp_passthrough_mask), sizeof(gp_passthrough_mask));
         }
     }
 }
 
+void FileEnvironment::Dump(u64 [[maybe_unused]] hash) {
+    DumpImpl(hash, code.get(), read_highest, read_lowest, initial_offset, stage);
+}
+
 u64 FileEnvironment::ReadInstruction(u32 address) {
     if (address < read_lowest || address > read_highest) {
         throw Shader::LogicError("Out of bounds address {}", address);

src/video_core/shader_environment.h

@@ -57,6 +57,8 @@ public:
 
     [[nodiscard]] u64 CalculateHash() const;
 
+    void Dump(u64 hash) override;
+
     void Serialize(std::ofstream& file) const;
 
 protected:
@@ -82,6 +84,7 @@ protected:
 
     u32 cached_lowest = std::numeric_limits<u32>::max();
     u32 cached_highest = 0;
+    u32 initial_offset = 0;
 
     bool has_unbound_instructions = false;
 };
@@ -149,6 +152,8 @@ public:
 
     [[nodiscard]] std::array<u32, 3> WorkgroupSize() const override;
 
+    void Dump(u64 hash) override;
+
 private:
     std::unique_ptr<u64[]> code;
     std::unordered_map<u32, Shader::TextureType> texture_types;
@@ -159,6 +164,7 @@ private:
     u32 texture_bound{};
     u32 read_lowest{};
     u32 read_highest{};
+    u32 initial_offset{};
 };
 
 void SerializePipeline(std::span<const char> key, std::span<const GenericEnvironment* const> envs,

src/video_core/texture_cache/texture_cache.h

@@ -1137,8 +1137,13 @@ typename TextureCache<P>::BlitImages TextureCache<P>::GetBlitImages(
     } while (has_deleted_images);
     const ImageBase& src_image = slot_images[src_id];
     const ImageBase& dst_image = slot_images[dst_id];
+    const bool native_bgr = runtime.HasNativeBgr();
     if (GetFormatType(dst_info.format) != GetFormatType(dst_image.info.format) ||
-        GetFormatType(src_info.format) != GetFormatType(src_image.info.format)) {
+        GetFormatType(src_info.format) != GetFormatType(src_image.info.format) ||
+        !VideoCore::Surface::IsViewCompatible(dst_info.format, dst_image.info.format, false,
+                                              native_bgr) ||
+        !VideoCore::Surface::IsViewCompatible(src_info.format, src_image.info.format, false,
+                                              native_bgr)) {
         // Make sure the images match the expected format.
         do {
             has_deleted_images = false;
@@ -1392,7 +1397,8 @@ ImageViewId TextureCache<P>::FindOrEmplaceImageView(ImageId image_id, const Imag
     if (const ImageViewId image_view_id = image.FindView(info); image_view_id) {
         return image_view_id;
     }
-    const ImageViewId image_view_id = slot_image_views.insert(runtime, info, image_id, image);
+    const ImageViewId image_view_id =
+        slot_image_views.insert(runtime, info, image_id, image, slot_images);
     image.InsertView(info, image_view_id);
     return image_view_id;
 }
@@ -1850,9 +1856,20 @@ void TextureCache<P>::CopyImage(ImageId dst_id, ImageId src_id, std::vector<Imag
             .height = std::min(dst_view.size.height, src_view.size.height),
             .depth = std::min(dst_view.size.depth, src_view.size.depth),
         };
-        UNIMPLEMENTED_IF(copy.extent != expected_size);
+        const Extent3D scaled_extent = [is_rescaled, expected_size]() {
+            if (!is_rescaled) {
+                return expected_size;
+            }
+            const auto& resolution = Settings::values.resolution_info;
+            return Extent3D{
+                .width = resolution.ScaleUp(expected_size.width),
+                .height = resolution.ScaleUp(expected_size.height),
+                .depth = expected_size.depth,
+            };
+        }();
+        UNIMPLEMENTED_IF(copy.extent != scaled_extent);
 
-        runtime.ConvertImage(dst_framebuffer, dst_view, src_view, is_rescaled);
+        runtime.ConvertImage(dst_framebuffer, dst_view, src_view);
     }
 }
 

src/video_core/texture_cache/util.cpp

@@ -364,14 +364,14 @@ template <u32 GOB_EXTENT>
 
 [[nodiscard]] std::optional<SubresourceExtent> ResolveOverlapRightAddress2D(
     const ImageInfo& new_info, GPUVAddr gpu_addr, const ImageBase& overlap, bool strict_size) {
-    const u32 layer_stride = new_info.layer_stride;
-    const s32 new_size = layer_stride * new_info.resources.layers;
-    const s32 diff = static_cast<s32>(overlap.gpu_addr - gpu_addr);
+    const u64 layer_stride = new_info.layer_stride;
+    const u64 new_size = layer_stride * new_info.resources.layers;
+    const u64 diff = overlap.gpu_addr - gpu_addr;
     if (diff > new_size) {
         return std::nullopt;
     }
-    const s32 base_layer = diff / layer_stride;
-    const s32 mip_offset = diff % layer_stride;
+    const s32 base_layer = static_cast<s32>(diff / layer_stride);
+    const s32 mip_offset = static_cast<s32>(diff % layer_stride);
     const std::array offsets = CalculateMipLevelOffsets(new_info);
     const auto end = offsets.begin() + new_info.resources.levels;
     const auto it = std::find(offsets.begin(), end, static_cast<u32>(mip_offset));

src/video_core/vulkan_common/vulkan_device.cpp

@@ -271,7 +271,7 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
             .tessellationShader = true,
             .sampleRateShading = true,
             .dualSrcBlend = true,
-            .logicOp = false,
+            .logicOp = true,
             .multiDrawIndirect = false,
             .drawIndirectFirstInstance = false,
             .depthClamp = true,
@@ -433,6 +433,19 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
         LOG_INFO(Render_Vulkan, "Device doesn't support uint8 indexes");
     }
 
+    VkPhysicalDevicePrimitiveTopologyListRestartFeaturesEXT primitive_topology_list_restart;
+    if (is_topology_list_restart_supported || is_patch_list_restart_supported) {
+        primitive_topology_list_restart = {
+            .sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PRIMITIVE_TOPOLOGY_LIST_RESTART_FEATURES_EXT,
+            .pNext = nullptr,
+            .primitiveTopologyListRestart = is_topology_list_restart_supported,
+            .primitiveTopologyPatchListRestart = is_patch_list_restart_supported,
+        };
+        SetNext(next, primitive_topology_list_restart);
+    } else {
+        LOG_INFO(Render_Vulkan, "Device doesn't support list topology primitive restart");
+    }
+
     VkPhysicalDeviceTransformFeedbackFeaturesEXT transform_feedback;
     if (ext_transform_feedback) {
         transform_feedback = {
@@ -625,15 +638,20 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
         }
     }
 
-    if (ext_vertex_input_dynamic_state && driver_id == VK_DRIVER_ID_INTEL_PROPRIETARY_WINDOWS) {
+    const bool is_intel_windows = driver_id == VK_DRIVER_ID_INTEL_PROPRIETARY_WINDOWS;
+    if (ext_vertex_input_dynamic_state && is_intel_windows) {
         LOG_WARNING(Render_Vulkan, "Blacklisting Intel for VK_EXT_vertex_input_dynamic_state");
         ext_vertex_input_dynamic_state = false;
     }
-    if (is_float16_supported && driver_id == VK_DRIVER_ID_INTEL_PROPRIETARY_WINDOWS) {
+    if (is_float16_supported && is_intel_windows) {
         // Intel's compiler crashes when using fp16 on Astral Chain, disable it for the time being.
         LOG_WARNING(Render_Vulkan, "Blacklisting Intel proprietary from float16 math");
         is_float16_supported = false;
     }
+    if (is_intel_windows) {
+        LOG_WARNING(Render_Vulkan, "Intel proprietary drivers do not support MSAA image blits");
+        cant_blit_msaa = true;
+    }
 
     supports_d24_depth =
         IsFormatSupported(VK_FORMAT_D24_UNORM_S8_UINT,
@@ -891,6 +909,7 @@ std::vector<const char*> Device::LoadExtensions(bool requires_surface) {
     bool has_ext_provoking_vertex{};
     bool has_ext_vertex_input_dynamic_state{};
     bool has_ext_line_rasterization{};
+    bool has_ext_primitive_topology_list_restart{};
     for (const std::string& extension : supported_extensions) {
         const auto test = [&](std::optional<std::reference_wrapper<bool>> status, const char* name,
                               bool push) {
@@ -915,6 +934,8 @@ std::vector<const char*> Device::LoadExtensions(bool requires_surface) {
         test(has_khr_shader_float16_int8, VK_KHR_SHADER_FLOAT16_INT8_EXTENSION_NAME, false);
         test(ext_depth_range_unrestricted, VK_EXT_DEPTH_RANGE_UNRESTRICTED_EXTENSION_NAME, true);
         test(ext_index_type_uint8, VK_EXT_INDEX_TYPE_UINT8_EXTENSION_NAME, true);
+        test(has_ext_primitive_topology_list_restart,
+             VK_EXT_PRIMITIVE_TOPOLOGY_LIST_RESTART_EXTENSION_NAME, true);
         test(ext_sampler_filter_minmax, VK_EXT_SAMPLER_FILTER_MINMAX_EXTENSION_NAME, true);
         test(ext_shader_viewport_index_layer, VK_EXT_SHADER_VIEWPORT_INDEX_LAYER_EXTENSION_NAME,
              true);
@@ -1113,6 +1134,19 @@ std::vector<const char*> Device::LoadExtensions(bool requires_surface) {
             khr_pipeline_executable_properties = true;
         }
     }
+    if (has_ext_primitive_topology_list_restart) {
+        VkPhysicalDevicePrimitiveTopologyListRestartFeaturesEXT primitive_topology_list_restart{};
+        primitive_topology_list_restart.sType =
+            VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_PRIMITIVE_TOPOLOGY_LIST_RESTART_FEATURES_EXT;
+        primitive_topology_list_restart.pNext = nullptr;
+        features.pNext = &primitive_topology_list_restart;
+        physical.GetFeatures2KHR(features);
+
+        is_topology_list_restart_supported =
+            primitive_topology_list_restart.primitiveTopologyListRestart;
+        is_patch_list_restart_supported =
+            primitive_topology_list_restart.primitiveTopologyPatchListRestart;
+    }
     if (has_khr_image_format_list && has_khr_swapchain_mutable_format) {
         extensions.push_back(VK_KHR_IMAGE_FORMAT_LIST_EXTENSION_NAME);
         extensions.push_back(VK_KHR_SWAPCHAIN_MUTABLE_FORMAT_EXTENSION_NAME);

src/video_core/vulkan_common/vulkan_device.h

@@ -238,6 +238,16 @@ public:
         return khr_workgroup_memory_explicit_layout;
     }
 
+    /// Returns true if the device supports VK_EXT_primitive_topology_list_restart.
+    bool IsTopologyListPrimitiveRestartSupported() const {
+        return is_topology_list_restart_supported;
+    }
+
+    /// Returns true if the device supports VK_EXT_primitive_topology_list_restart.
+    bool IsPatchListPrimitiveRestartSupported() const {
+        return is_patch_list_restart_supported;
+    }
+
     /// Returns true if the device supports VK_EXT_index_type_uint8.
     bool IsExtIndexTypeUint8Supported() const {
         return ext_index_type_uint8;
@@ -340,6 +350,10 @@ public:
         return supports_d24_depth;
     }
 
+    bool CantBlitMSAA() const {
+        return cant_blit_msaa;
+    }
+
 private:
     /// Checks if the physical device is suitable.
     void CheckSuitability(bool requires_swapchain) const;
@@ -401,6 +415,9 @@ private:
     bool is_shader_int16_supported{};           ///< Support for int16.
     bool is_shader_storage_image_multisample{}; ///< Support for image operations on MSAA images.
     bool is_blit_depth_stencil_supported{};     ///< Support for blitting from and to depth stencil.
+    bool is_topology_list_restart_supported{};  ///< Support for primitive restart with list
+                                                ///< topologies.
+    bool is_patch_list_restart_supported{};     ///< Support for primitive restart with list patch.
     bool nv_viewport_swizzle{};                 ///< Support for VK_NV_viewport_swizzle.
     bool nv_viewport_array2{};                  ///< Support for VK_NV_viewport_array2.
     bool nv_geometry_shader_passthrough{};      ///< Support for VK_NV_geometry_shader_passthrough.
@@ -430,6 +447,7 @@ private:
     bool has_renderdoc{};                   ///< Has RenderDoc attached
     bool has_nsight_graphics{};             ///< Has Nsight Graphics attached
     bool supports_d24_depth{};              ///< Supports D24 depth buffers.
+    bool cant_blit_msaa{};                  ///< Does not support MSAA<->MSAA blitting.
 
     // Telemetry parameters
     std::string vendor_name;                       ///< Device's driver name.

src/yuzu/CMakeLists.txt

@@ -251,6 +251,9 @@ target_include_directories(yuzu PRIVATE ../../externals/Vulkan-Headers/include)
 if (NOT WIN32)
     target_include_directories(yuzu PRIVATE ${Qt5Gui_PRIVATE_INCLUDE_DIRS})
 endif()
+if (${CMAKE_SYSTEM_NAME} STREQUAL "Linux")
+    target_link_libraries(yuzu PRIVATE Qt5::DBus)
+endif()
 
 target_compile_definitions(yuzu PRIVATE
     # Use QStringBuilder for string concatenation to reduce

src/yuzu/applets/qt_controller.cpp

@@ -12,7 +12,6 @@
 #include "core/hid/emulated_controller.h"
 #include "core/hid/hid_core.h"
 #include "core/hid/hid_types.h"
-#include "core/hle/lock.h"
 #include "core/hle/service/hid/controllers/npad.h"
 #include "core/hle/service/hid/hid.h"
 #include "core/hle/service/sm/sm.h"
@@ -34,7 +33,7 @@ void UpdateController(Core::HID::EmulatedController* controller,
     }
     controller->SetNpadStyleIndex(controller_type);
     if (connected) {
-        controller->Connect();
+        controller->Connect(true);
     }
 }
 
@@ -664,7 +663,5 @@ void QtControllerSelector::ReconfigureControllers(
 }
 
 void QtControllerSelector::MainWindowReconfigureFinished() {
-    // Acquire the HLE mutex
-    std::lock_guard lock(HLE::g_hle_lock);
     callback();
 }

src/yuzu/applets/qt_error.cpp

@@ -3,7 +3,6 @@
 // Refer to the license.txt file included.
 
 #include <QDateTime>
-#include "core/hle/lock.h"
 #include "yuzu/applets/qt_error.h"
 #include "yuzu/main.h"
 
@@ -57,7 +56,5 @@ void QtErrorDisplay::ShowCustomErrorText(ResultCode error, std::string dialog_te
 }
 
 void QtErrorDisplay::MainWindowFinishedError() {
-    // Acquire the HLE mutex
-    std::lock_guard lock{HLE::g_hle_lock};
     callback();
 }

src/yuzu/applets/qt_profile_select.cpp

@@ -14,7 +14,6 @@
 #include "common/fs/path_util.h"
 #include "common/string_util.h"
 #include "core/constants.h"
-#include "core/hle/lock.h"
 #include "yuzu/applets/qt_profile_select.h"
 #include "yuzu/main.h"
 #include "yuzu/util/controller_navigation.h"
@@ -170,7 +169,5 @@ void QtProfileSelector::SelectProfile(
 }
 
 void QtProfileSelector::MainWindowFinishedSelection(std::optional<Common::UUID> uuid) {
-    // Acquire the HLE mutex
-    std::lock_guard lock{HLE::g_hle_lock};
     callback(uuid);
 }

src/yuzu/configuration/config.cpp

@@ -776,6 +776,7 @@ void Config::ReadUIGamelistValues() {
     ReadBasicSetting(UISettings::values.row_1_text_id);
     ReadBasicSetting(UISettings::values.row_2_text_id);
     ReadBasicSetting(UISettings::values.cache_game_list);
+    ReadBasicSetting(UISettings::values.favorites_expanded);
     const int favorites_size = qt_config->beginReadArray(QStringLiteral("favorites"));
     for (int i = 0; i < favorites_size; i++) {
         qt_config->setArrayIndex(i);
@@ -1300,6 +1301,7 @@ void Config::SaveUIGamelistValues() {
     WriteBasicSetting(UISettings::values.row_1_text_id);
     WriteBasicSetting(UISettings::values.row_2_text_id);
     WriteBasicSetting(UISettings::values.cache_game_list);
+    WriteBasicSetting(UISettings::values.favorites_expanded);
     qt_config->beginWriteArray(QStringLiteral("favorites"));
     for (int i = 0; i < UISettings::values.favorited_ids.size(); i++) {
         qt_config->setArrayIndex(i);

src/yuzu/configuration/configure_debug.cpp

@@ -51,6 +51,8 @@ void ConfigureDebug::SetConfiguration() {
     ui->enable_cpu_debugging->setChecked(Settings::values.cpu_debug_mode.GetValue());
     ui->enable_nsight_aftermath->setEnabled(runtime_lock);
     ui->enable_nsight_aftermath->setChecked(Settings::values.enable_nsight_aftermath.GetValue());
+    ui->dump_shaders->setEnabled(runtime_lock);
+    ui->dump_shaders->setChecked(Settings::values.dump_shaders.GetValue());
     ui->disable_macro_jit->setEnabled(runtime_lock);
     ui->disable_macro_jit->setChecked(Settings::values.disable_macro_jit.GetValue());
     ui->disable_loop_safety_checks->setEnabled(runtime_lock);
@@ -73,6 +75,7 @@ void ConfigureDebug::ApplyConfiguration() {
     Settings::values.renderer_shader_feedback = ui->enable_shader_feedback->isChecked();
     Settings::values.cpu_debug_mode = ui->enable_cpu_debugging->isChecked();
     Settings::values.enable_nsight_aftermath = ui->enable_nsight_aftermath->isChecked();
+    Settings::values.dump_shaders = ui->dump_shaders->isChecked();
     Settings::values.disable_shader_loop_safety_checks =
         ui->disable_loop_safety_checks->isChecked();
     Settings::values.disable_macro_jit = ui->disable_macro_jit->isChecked();

src/yuzu/configuration/configure_debug.ui

@@ -105,6 +105,19 @@
         </property>
        </widget>
       </item>
+      <item row="2" column="1">
+       <widget class="QCheckBox" name="dump_shaders">
+        <property name="enabled">
+         <bool>true</bool>
+        </property>
+        <property name="toolTip">
+         <string>When checked, it will dump all the original assembler shaders from the disk shader cache or game as found</string>
+        </property>
+        <property name="text">
+         <string>Dump Game Shaders</string>
+        </property>
+       </widget>
+      </item>
       <item row="0" column="1">
        <widget class="QCheckBox" name="disable_macro_jit">
         <property name="enabled">