Simply legacy attribute implement

Fixup channel submit IOCTL syncpoint parameters

src/common/atomic_threadsafe_queue.h

@@ -1,240 +0,0 @@
-/*
-Copyright (c) 2020 Erik Rigtorp <erik@rigtorp.se>
-
-Permission is hereby granted, free of charge, to any person obtaining a copy
-of this software and associated documentation files (the "Software"), to deal
-in the Software without restriction, including without limitation the rights
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice shall be included in all
-copies or substantial portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.
- */
-
-#pragma once
-#ifdef _MSC_VER
-#pragma warning(push)
-#pragma warning(disable : 4324)
-#endif
-
-#include <atomic>
-#include <cassert>
-#include <cstddef> // offsetof
-#include <memory>
-#include <new>
-#include <stdexcept>
-
-namespace Common {
-namespace mpmc {
-static constexpr size_t hardwareInterferenceSize = 64;
-
-template <typename T>
-using AlignedAllocator = std::allocator<T>;
-
-template <typename T>
-struct Slot {
-    ~Slot() noexcept {
-        if (turn & 1) {
-            destroy();
-        }
-    }
-
-    template <typename... Args>
-    void construct(Args&&... args) noexcept {
-        static_assert(std::is_nothrow_constructible<T, Args&&...>::value,
-                      "T must be nothrow constructible with Args&&...");
-        new (&storage) T(std::forward<Args>(args)...);
-    }
-
-    void destroy() noexcept {
-        static_assert(std::is_nothrow_destructible<T>::value, "T must be nothrow destructible");
-        reinterpret_cast<T*>(&storage)->~T();
-    }
-
-    T&& move() noexcept {
-        return reinterpret_cast<T&&>(storage);
-    }
-
-    // Align to avoid false sharing between adjacent slots
-    alignas(hardwareInterferenceSize) std::atomic<size_t> turn = {0};
-    typename std::aligned_storage<sizeof(T), alignof(T)>::type storage;
-};
-
-template <typename T, typename Allocator = AlignedAllocator<Slot<T>>>
-class Queue {
-private:
-    static_assert(std::is_nothrow_copy_assignable<T>::value ||
-                      std::is_nothrow_move_assignable<T>::value,
-                  "T must be nothrow copy or move assignable");
-
-    static_assert(std::is_nothrow_destructible<T>::value, "T must be nothrow destructible");
-
-public:
-    explicit Queue(const size_t capacity, const Allocator& allocator = Allocator())
-        : capacity_(capacity), allocator_(allocator), head_(0), tail_(0) {
-        if (capacity_ < 1) {
-            throw std::invalid_argument("capacity < 1");
-        }
-        // Allocate one extra slot to prevent false sharing on the last slot
-        slots_ = allocator_.allocate(capacity_ + 1);
-        // Allocators are not required to honor alignment for over-aligned types
-        // (see http://eel.is/c++draft/allocator.requirements#10) so we verify
-        // alignment here
-        if (reinterpret_cast<size_t>(slots_) % alignof(Slot<T>) != 0) {
-            allocator_.deallocate(slots_, capacity_ + 1);
-            throw std::bad_alloc();
-        }
-        for (size_t i = 0; i < capacity_; ++i) {
-            new (&slots_[i]) Slot<T>();
-        }
-        static_assert(alignof(Slot<T>) == hardwareInterferenceSize,
-                      "Slot must be aligned to cache line boundary to prevent false sharing");
-        static_assert(sizeof(Slot<T>) % hardwareInterferenceSize == 0,
-                      "Slot size must be a multiple of cache line size to prevent "
-                      "false sharing between adjacent slots");
-        static_assert(sizeof(Queue) % hardwareInterferenceSize == 0,
-                      "Queue size must be a multiple of cache line size to "
-                      "prevent false sharing between adjacent queues");
-        static_assert(offsetof(Queue, tail_) - offsetof(Queue, head_) ==
-                          static_cast<std::ptrdiff_t>(hardwareInterferenceSize),
-                      "head and tail must be a cache line apart to prevent false sharing");
-    }
-
-    ~Queue() noexcept {
-        for (size_t i = 0; i < capacity_; ++i) {
-            slots_[i].~Slot();
-        }
-        allocator_.deallocate(slots_, capacity_ + 1);
-    }
-
-    // non-copyable and non-movable
-    Queue(const Queue&) = delete;
-    Queue& operator=(const Queue&) = delete;
-
-    template <typename... Args>
-    void emplace(Args&&... args) noexcept {
-        static_assert(std::is_nothrow_constructible<T, Args&&...>::value,
-                      "T must be nothrow constructible with Args&&...");
-        auto const head = head_.fetch_add(1);
-        auto& slot = slots_[idx(head)];
-        while (turn(head) * 2 != slot.turn.load(std::memory_order_acquire))
-            ;
-        slot.construct(std::forward<Args>(args)...);
-        slot.turn.store(turn(head) * 2 + 1, std::memory_order_release);
-    }
-
-    template <typename... Args>
-    bool try_emplace(Args&&... args) noexcept {
-        static_assert(std::is_nothrow_constructible<T, Args&&...>::value,
-                      "T must be nothrow constructible with Args&&...");
-        auto head = head_.load(std::memory_order_acquire);
-        for (;;) {
-            auto& slot = slots_[idx(head)];
-            if (turn(head) * 2 == slot.turn.load(std::memory_order_acquire)) {
-                if (head_.compare_exchange_strong(head, head + 1)) {
-                    slot.construct(std::forward<Args>(args)...);
-                    slot.turn.store(turn(head) * 2 + 1, std::memory_order_release);
-                    return true;
-                }
-            } else {
-                auto const prevHead = head;
-                head = head_.load(std::memory_order_acquire);
-                if (head == prevHead) {
-                    return false;
-                }
-            }
-        }
-    }
-
-    void push(const T& v) noexcept {
-        static_assert(std::is_nothrow_copy_constructible<T>::value,
-                      "T must be nothrow copy constructible");
-        emplace(v);
-    }
-
-    template <typename P, typename = typename std::enable_if<
-                              std::is_nothrow_constructible<T, P&&>::value>::type>
-    void push(P&& v) noexcept {
-        emplace(std::forward<P>(v));
-    }
-
-    bool try_push(const T& v) noexcept {
-        static_assert(std::is_nothrow_copy_constructible<T>::value,
-                      "T must be nothrow copy constructible");
-        return try_emplace(v);
-    }
-
-    template <typename P, typename = typename std::enable_if<
-                              std::is_nothrow_constructible<T, P&&>::value>::type>
-    bool try_push(P&& v) noexcept {
-        return try_emplace(std::forward<P>(v));
-    }
-
-    void pop(T& v) noexcept {
-        auto const tail = tail_.fetch_add(1);
-        auto& slot = slots_[idx(tail)];
-        while (turn(tail) * 2 + 1 != slot.turn.load(std::memory_order_acquire))
-            ;
-        v = slot.move();
-        slot.destroy();
-        slot.turn.store(turn(tail) * 2 + 2, std::memory_order_release);
-    }
-
-    bool try_pop(T& v) noexcept {
-        auto tail = tail_.load(std::memory_order_acquire);
-        for (;;) {
-            auto& slot = slots_[idx(tail)];
-            if (turn(tail) * 2 + 1 == slot.turn.load(std::memory_order_acquire)) {
-                if (tail_.compare_exchange_strong(tail, tail + 1)) {
-                    v = slot.move();
-                    slot.destroy();
-                    slot.turn.store(turn(tail) * 2 + 2, std::memory_order_release);
-                    return true;
-                }
-            } else {
-                auto const prevTail = tail;
-                tail = tail_.load(std::memory_order_acquire);
-                if (tail == prevTail) {
-                    return false;
-                }
-            }
-        }
-    }
-
-private:
-    constexpr size_t idx(size_t i) const noexcept {
-        return i % capacity_;
-    }
-
-    constexpr size_t turn(size_t i) const noexcept {
-        return i / capacity_;
-    }
-
-private:
-    const size_t capacity_;
-    Slot<T>* slots_;
-    [[no_unique_address]] Allocator allocator_;
-
-    // Align to avoid false sharing between head_ and tail_
-    alignas(hardwareInterferenceSize) std::atomic<size_t> head_;
-    alignas(hardwareInterferenceSize) std::atomic<size_t> tail_;
-};
-} // namespace mpmc
-
-template <typename T, typename Allocator = mpmc::AlignedAllocator<mpmc::Slot<T>>>
-using MPMCQueue = mpmc::Queue<T, Allocator>;
-
-} // namespace Common
-
-#ifdef _MSC_VER
-#pragma warning(pop)
-#endif

src/core/arm/dynarmic/arm_dynarmic_64.cpp

@@ -263,7 +263,7 @@ void ARM_Dynarmic_64::Run() {
 }
 
 void ARM_Dynarmic_64::Step() {
-    cb->InterpreterFallback(jit->GetPC(), 1);
+    jit->Step();
 }
 
 ARM_Dynarmic_64::ARM_Dynarmic_64(System& system_, CPUInterrupts& interrupt_handlers_,

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

@@ -69,8 +69,7 @@ NvResult nvhost_nvdec_common::Submit(const std::vector<u8>& input, std::vector<u
     std::vector<Reloc> relocs(params.relocation_count);
     std::vector<u32> reloc_shifts(params.relocation_count);
     std::vector<SyncptIncr> syncpt_increments(params.syncpoint_count);
-    std::vector<SyncptIncr> wait_checks(params.syncpoint_count);
-    std::vector<Fence> fences(params.fence_count);
+    std::vector<u32> fence_thresholds(params.fence_count);
 
     // Slice input into their respective buffers
     std::size_t offset = sizeof(IoctlSubmit);
@@ -78,15 +77,13 @@ NvResult nvhost_nvdec_common::Submit(const std::vector<u8>& input, std::vector<u
     offset += SliceVectors(input, relocs, params.relocation_count, offset);
     offset += SliceVectors(input, reloc_shifts, params.relocation_count, offset);
     offset += SliceVectors(input, syncpt_increments, params.syncpoint_count, offset);
-    offset += SliceVectors(input, wait_checks, params.syncpoint_count, offset);
-    offset += SliceVectors(input, fences, params.fence_count, offset);
+    offset += SliceVectors(input, fence_thresholds, params.fence_count, offset);
 
     auto& gpu = system.GPU();
     if (gpu.UseNvdec()) {
         for (std::size_t i = 0; i < syncpt_increments.size(); i++) {
             const SyncptIncr& syncpt_incr = syncpt_increments[i];
-            fences[i].id = syncpt_incr.id;
-            fences[i].value =
+            fence_thresholds[i] =
                 syncpoint_manager.IncreaseSyncpoint(syncpt_incr.id, syncpt_incr.increments);
         }
     }
@@ -98,11 +95,6 @@ NvResult nvhost_nvdec_common::Submit(const std::vector<u8>& input, std::vector<u
                                   cmdlist.size() * sizeof(u32));
         gpu.PushCommandBuffer(cmdlist);
     }
-    if (gpu.UseNvdec()) {
-        fences[0].value = syncpoint_manager.IncreaseSyncpoint(fences[0].id, 1);
-        Tegra::ChCommandHeaderList cmdlist{{(4 << 28) | fences[0].id}};
-        gpu.PushCommandBuffer(cmdlist);
-    }
     std::memcpy(output.data(), &params, sizeof(IoctlSubmit));
     // Some games expect command_buffers to be written back
     offset = sizeof(IoctlSubmit);
@@ -110,8 +102,7 @@ NvResult nvhost_nvdec_common::Submit(const std::vector<u8>& input, std::vector<u
     offset += WriteVectors(output, relocs, offset);
     offset += WriteVectors(output, reloc_shifts, offset);
     offset += WriteVectors(output, syncpt_increments, offset);
-    offset += WriteVectors(output, wait_checks, offset);
-    offset += WriteVectors(output, fences, offset);
+    offset += WriteVectors(output, fence_thresholds, offset);
 
     return NvResult::Success;
 }

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

@@ -56,19 +56,16 @@ protected:
         s32 target{};
         s32 target_offset{};
     };
-    static_assert(sizeof(Reloc) == 0x10, "CommandBuffer has incorrect size");
+    static_assert(sizeof(Reloc) == 0x10, "Reloc has incorrect size");
 
     struct SyncptIncr {
         u32 id{};
         u32 increments{};
+        u32 unk0{};
+        u32 unk1{};
+        u32 unk2{};
     };
-    static_assert(sizeof(SyncptIncr) == 0x8, "CommandBuffer has incorrect size");
-
-    struct Fence {
-        u32 id{};
-        u32 value{};
-    };
-    static_assert(sizeof(Fence) == 0x8, "CommandBuffer has incorrect size");
+    static_assert(sizeof(SyncptIncr) == 0x14, "SyncptIncr has incorrect size");
 
     struct IoctlGetSyncpoint {
         // Input

src/input_common/gcadapter/gc_adapter.cpp

@@ -170,7 +170,7 @@ void Adapter::UpdateYuzuSettings(std::size_t port) {
 
     if (pads[port].buttons != 0) {
         pad_status.button = pads[port].last_button;
-        pad_queue.push(pad_status);
+        pad_queue.Push(pad_status);
     }
 
     // Accounting for a threshold here to ensure an intentional press
@@ -181,7 +181,7 @@ void Adapter::UpdateYuzuSettings(std::size_t port) {
             pad_status.axis = static_cast<PadAxes>(i);
             pad_status.axis_value = value;
             pad_status.axis_threshold = axis_threshold;
-            pad_queue.push(pad_status);
+            pad_queue.Push(pad_status);
         }
     }
 }
@@ -478,18 +478,20 @@ bool Adapter::DeviceConnected(std::size_t port) const {
 }
 
 void Adapter::BeginConfiguration() {
+    pad_queue.Clear();
     configuring = true;
 }
 
 void Adapter::EndConfiguration() {
+    pad_queue.Clear();
     configuring = false;
 }
 
-Common::MPMCQueue<GCPadStatus>& Adapter::GetPadQueue() {
+Common::SPSCQueue<GCPadStatus>& Adapter::GetPadQueue() {
     return pad_queue;
 }
 
-const Common::MPMCQueue<GCPadStatus>& Adapter::GetPadQueue() const {
+const Common::SPSCQueue<GCPadStatus>& Adapter::GetPadQueue() const {
     return pad_queue;
 }
 

src/input_common/gcadapter/gc_adapter.h

@@ -8,8 +8,8 @@
 #include <mutex>
 #include <thread>
 #include <unordered_map>
-#include "common/atomic_threadsafe_queue.h"
 #include "common/common_types.h"
+#include "common/threadsafe_queue.h"
 #include "input_common/main.h"
 
 struct libusb_context;
@@ -85,8 +85,8 @@ public:
     void BeginConfiguration();
     void EndConfiguration();
 
-    Common::MPMCQueue<GCPadStatus>& GetPadQueue();
-    const Common::MPMCQueue<GCPadStatus>& GetPadQueue() const;
+    Common::SPSCQueue<GCPadStatus>& GetPadQueue();
+    const Common::SPSCQueue<GCPadStatus>& GetPadQueue() const;
 
     GCController& GetPadState(std::size_t port);
     const GCController& GetPadState(std::size_t port) const;
@@ -145,7 +145,7 @@ private:
 
     libusb_device_handle* usb_adapter_handle = nullptr;
     std::array<GCController, 4> pads;
-    Common::MPMCQueue<GCPadStatus> pad_queue{1024};
+    Common::SPSCQueue<GCPadStatus> pad_queue;
 
     std::thread adapter_input_thread;
     std::thread adapter_scan_thread;

src/input_common/gcadapter/gc_poller.cpp

@@ -7,7 +7,7 @@
 #include <mutex>
 #include <utility>
 #include "common/assert.h"
-#include "common/atomic_threadsafe_queue.h"
+#include "common/threadsafe_queue.h"
 #include "input_common/gcadapter/gc_adapter.h"
 #include "input_common/gcadapter/gc_poller.h"
 
@@ -103,7 +103,7 @@ Common::ParamPackage GCButtonFactory::GetNextInput() const {
     Common::ParamPackage params;
     GCAdapter::GCPadStatus pad;
     auto& queue = adapter->GetPadQueue();
-    while (queue.try_pop(pad)) {
+    while (queue.Pop(pad)) {
         // This while loop will break on the earliest detected button
         params.Set("engine", "gcpad");
         params.Set("port", static_cast<s32>(pad.port));
@@ -263,7 +263,7 @@ Common::ParamPackage GCAnalogFactory::GetNextInput() {
     GCAdapter::GCPadStatus pad;
     Common::ParamPackage params;
     auto& queue = adapter->GetPadQueue();
-    while (queue.try_pop(pad)) {
+    while (queue.Pop(pad)) {
         if (pad.button != GCAdapter::PadButton::Undefined) {
             params.Set("engine", "gcpad");
             params.Set("port", static_cast<s32>(pad.port));

src/input_common/mouse/mouse_input.cpp

@@ -52,7 +52,7 @@ void Mouse::UpdateYuzuSettings() {
         return;
     }
 
-    mouse_queue.push(MouseStatus{
+    mouse_queue.Push(MouseStatus{
         .button = last_button,
     });
 }
@@ -153,6 +153,7 @@ void Mouse::ReleaseAllButtons() {
 void Mouse::BeginConfiguration() {
     buttons = 0;
     last_button = MouseButton::Undefined;
+    mouse_queue.Clear();
     configuring = true;
 }
 
@@ -164,6 +165,7 @@ void Mouse::EndConfiguration() {
         info.data.axis = {0, 0};
     }
     last_button = MouseButton::Undefined;
+    mouse_queue.Clear();
     configuring = false;
 }
 
@@ -203,11 +205,11 @@ bool Mouse::UnlockButton(std::size_t button_) {
     return button_state;
 }
 
-Common::MPMCQueue<MouseStatus>& Mouse::GetMouseQueue() {
+Common::SPSCQueue<MouseStatus>& Mouse::GetMouseQueue() {
     return mouse_queue;
 }
 
-const Common::MPMCQueue<MouseStatus>& Mouse::GetMouseQueue() const {
+const Common::SPSCQueue<MouseStatus>& Mouse::GetMouseQueue() const {
     return mouse_queue;
 }
 

src/input_common/mouse/mouse_input.h

@@ -9,8 +9,8 @@
 #include <stop_token>
 #include <thread>
 
-#include "common/atomic_threadsafe_queue.h"
 #include "common/common_types.h"
+#include "common/threadsafe_queue.h"
 #include "common/vector_math.h"
 #include "core/frontend/input.h"
 #include "input_common/motion_input.h"
@@ -79,8 +79,8 @@ public:
     [[nodiscard]] bool ToggleButton(std::size_t button_);
     [[nodiscard]] bool UnlockButton(std::size_t button_);
 
-    [[nodiscard]] Common::MPMCQueue<MouseStatus>& GetMouseQueue();
-    [[nodiscard]] const Common::MPMCQueue<MouseStatus>& GetMouseQueue() const;
+    [[nodiscard]] Common::SPSCQueue<MouseStatus>& GetMouseQueue();
+    [[nodiscard]] const Common::SPSCQueue<MouseStatus>& GetMouseQueue() const;
 
     [[nodiscard]] MouseData& GetMouseState(std::size_t button);
     [[nodiscard]] const MouseData& GetMouseState(std::size_t button) const;
@@ -109,7 +109,7 @@ private:
     std::jthread update_thread;
     MouseButton last_button{MouseButton::Undefined};
     std::array<MouseInfo, 7> mouse_info;
-    Common::MPMCQueue<MouseStatus> mouse_queue{1024};
+    Common::SPSCQueue<MouseStatus> mouse_queue;
     bool configuring{false};
     int mouse_panning_timout{};
 };

src/input_common/mouse/mouse_poller.cpp

@@ -52,7 +52,7 @@ Common::ParamPackage MouseButtonFactory::GetNextInput() const {
     MouseInput::MouseStatus pad;
     Common::ParamPackage params;
     auto& queue = mouse_input->GetMouseQueue();
-    while (queue.try_pop(pad)) {
+    while (queue.Pop(pad)) {
         // This while loop will break on the earliest detected button
         if (pad.button != MouseInput::MouseButton::Undefined) {
             params.Set("engine", "mouse");
@@ -184,7 +184,7 @@ Common::ParamPackage MouseAnalogFactory::GetNextInput() const {
     MouseInput::MouseStatus pad;
     Common::ParamPackage params;
     auto& queue = mouse_input->GetMouseQueue();
-    while (queue.try_pop(pad)) {
+    while (queue.Pop(pad)) {
         // This while loop will break on the earliest detected button
         if (pad.button != MouseInput::MouseButton::Undefined) {
             params.Set("engine", "mouse");
@@ -227,7 +227,7 @@ Common::ParamPackage MouseMotionFactory::GetNextInput() const {
     MouseInput::MouseStatus pad;
     Common::ParamPackage params;
     auto& queue = mouse_input->GetMouseQueue();
-    while (queue.try_pop(pad)) {
+    while (queue.Pop(pad)) {
         // This while loop will break on the earliest detected button
         if (pad.button != MouseInput::MouseButton::Undefined) {
             params.Set("engine", "mouse");
@@ -275,7 +275,7 @@ Common::ParamPackage MouseTouchFactory::GetNextInput() const {
     MouseInput::MouseStatus pad;
     Common::ParamPackage params;
     auto& queue = mouse_input->GetMouseQueue();
-    while (queue.try_pop(pad)) {
+    while (queue.Pop(pad)) {
         // This while loop will break on the earliest detected button
         if (pad.button != MouseInput::MouseButton::Undefined) {
             params.Set("engine", "mouse");

src/input_common/sdl/sdl_impl.cpp

@@ -46,7 +46,7 @@ static int SDLEventWatcher(void* user_data, SDL_Event* event) {
 
     // Don't handle the event if we are configuring
     if (sdl_state->polling) {
-        sdl_state->event_queue.push(*event);
+        sdl_state->event_queue.Push(*event);
     } else {
         sdl_state->HandleGameControllerEvent(*event);
     }
@@ -1460,6 +1460,7 @@ public:
     explicit SDLPoller(SDLState& state_) : state(state_) {}
 
     void Start([[maybe_unused]] const std::string& device_id) override {
+        state.event_queue.Clear();
         state.polling = true;
     }
 
@@ -1477,7 +1478,7 @@ public:
 
     Common::ParamPackage GetNextInput() override {
         SDL_Event event;
-        while (state.event_queue.try_pop(event)) {
+        while (state.event_queue.Pop(event)) {
             const auto package = FromEvent(event);
             if (package) {
                 return *package;
@@ -1549,7 +1550,7 @@ public:
 
     Common::ParamPackage GetNextInput() override {
         SDL_Event event;
-        while (state.event_queue.try_pop(event)) {
+        while (state.event_queue.Pop(event)) {
             const auto package = FromEvent(event);
             if (package) {
                 return *package;
@@ -1591,7 +1592,7 @@ public:
 
     Common::ParamPackage GetNextInput() override {
         SDL_Event event;
-        while (state.event_queue.try_pop(event)) {
+        while (state.event_queue.Pop(event)) {
             if (event.type != SDL_JOYAXISMOTION) {
                 // Check for a button press
                 auto button_press = button_poller.FromEvent(event);

src/input_common/sdl/sdl_impl.h

@@ -12,8 +12,8 @@
 
 #include <SDL.h>
 
-#include "common/atomic_threadsafe_queue.h"
 #include "common/common_types.h"
+#include "common/threadsafe_queue.h"
 #include "input_common/sdl/sdl.h"
 
 union SDL_Event;
@@ -59,7 +59,7 @@ public:
 
     /// Used by the Pollers during config
     std::atomic<bool> polling = false;
-    Common::MPMCQueue<SDL_Event> event_queue{1024};
+    Common::SPSCQueue<SDL_Event> event_queue;
 
     std::vector<Common::ParamPackage> GetInputDevices() override;
 

src/input_common/udp/client.cpp

@@ -338,7 +338,7 @@ void Client::UpdateYuzuSettings(std::size_t client, std::size_t pad_index,
                   gyro[0], gyro[1], gyro[2], acc[0], acc[1], acc[2]);
     }
     UDPPadStatus pad{
-        .host = clients[client].host.c_str(),
+        .host = clients[client].host,
         .port = clients[client].port,
         .pad_index = pad_index,
     };
@@ -346,12 +346,12 @@ void Client::UpdateYuzuSettings(std::size_t client, std::size_t pad_index,
         if (gyro[i] > 5.0f || gyro[i] < -5.0f) {
             pad.motion = static_cast<PadMotion>(i);
             pad.motion_value = gyro[i];
-            pad_queue.push(pad);
+            pad_queue.Push(pad);
         }
         if (acc[i] > 1.75f || acc[i] < -1.75f) {
             pad.motion = static_cast<PadMotion>(i + 3);
             pad.motion_value = acc[i];
-            pad_queue.push(pad);
+            pad_queue.Push(pad);
         }
     }
 }
@@ -401,10 +401,12 @@ void Client::UpdateTouchInput(Response::TouchPad& touch_pad, std::size_t client,
 }
 
 void Client::BeginConfiguration() {
+    pad_queue.Clear();
     configuring = true;
 }
 
 void Client::EndConfiguration() {
+    pad_queue.Clear();
     configuring = false;
 }
 
@@ -432,11 +434,11 @@ const Input::TouchStatus& Client::GetTouchState() const {
     return touch_status;
 }
 
-Common::MPMCQueue<UDPPadStatus>& Client::GetPadQueue() {
+Common::SPSCQueue<UDPPadStatus>& Client::GetPadQueue() {
     return pad_queue;
 }
 
-const Common::MPMCQueue<UDPPadStatus>& Client::GetPadQueue() const {
+const Common::SPSCQueue<UDPPadStatus>& Client::GetPadQueue() const {
     return pad_queue;
 }
 

src/input_common/udp/client.h

@@ -11,10 +11,10 @@
 #include <string>
 #include <thread>
 #include <tuple>
-#include "common/atomic_threadsafe_queue.h"
 #include "common/common_types.h"
 #include "common/param_package.h"
 #include "common/thread.h"
+#include "common/threadsafe_queue.h"
 #include "common/vector_math.h"
 #include "core/frontend/input.h"
 #include "input_common/motion_input.h"
@@ -46,7 +46,7 @@ enum class PadTouch {
 };
 
 struct UDPPadStatus {
-    const char* host{"127.0.0.1"};
+    std::string host{"127.0.0.1"};
     u16 port{26760};
     std::size_t pad_index{};
     PadMotion motion{PadMotion::Undefined};
@@ -85,8 +85,8 @@ public:
     bool DeviceConnected(std::size_t pad) const;
     void ReloadSockets();
 
-    Common::MPMCQueue<UDPPadStatus>& GetPadQueue();
-    const Common::MPMCQueue<UDPPadStatus>& GetPadQueue() const;
+    Common::SPSCQueue<UDPPadStatus>& GetPadQueue();
+    const Common::SPSCQueue<UDPPadStatus>& GetPadQueue() const;
 
     DeviceStatus& GetPadState(const std::string& host, u16 port, std::size_t pad);
     const DeviceStatus& GetPadState(const std::string& host, u16 port, std::size_t pad) const;
@@ -146,7 +146,7 @@ private:
     static constexpr std::size_t MAX_TOUCH_FINGERS = MAX_UDP_CLIENTS * 2;
     std::array<PadData, MAX_UDP_CLIENTS * PADS_PER_CLIENT> pads{};
     std::array<ClientConnection, MAX_UDP_CLIENTS> clients{};
-    Common::MPMCQueue<UDPPadStatus> pad_queue{1024};
+    Common::SPSCQueue<UDPPadStatus> pad_queue{};
     Input::TouchStatus touch_status{};
     std::array<std::size_t, MAX_TOUCH_FINGERS> finger_id{};
 };

src/input_common/udp/udp.cpp

@@ -59,7 +59,7 @@ Common::ParamPackage UDPMotionFactory::GetNextInput() {
     Common::ParamPackage params;
     CemuhookUDP::UDPPadStatus pad;
     auto& queue = client->GetPadQueue();
-    while (queue.try_pop(pad)) {
+    while (queue.Pop(pad)) {
         if (pad.motion == CemuhookUDP::PadMotion::Undefined || std::abs(pad.motion_value) < 1) {
             continue;
         }

src/shader_recompiler/backend/spirv/emit_context.cpp

@@ -430,15 +430,33 @@ Id DescType(EmitContext& ctx, Id sampled_type, Id pointer_type, u32 count) {
     }
 }
 
-size_t FindNextUnusedLocation(const std::bitset<IR::NUM_GENERICS>& used_locations,
-                              size_t start_offset) {
+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) {
@@ -520,6 +538,64 @@ 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();
 
@@ -1279,22 +1355,26 @@ void EmitContext::DefineInputs(const IR::Program& program) {
     }
     size_t previous_unused_location = 0;
     if (loads.AnyComponent(IR::Attribute::ColorFrontDiffuseR)) {
-        const size_t location = FindNextUnusedLocation(used_locations, previous_unused_location);
-        previous_unused_location = location;
-        used_locations.set(location);
-        const Id id{DefineInput(*this, F32[4], true)};
-        Decorate(id, spv::Decoration::Location, location);
-        input_front_color = id;
+        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)) {
-            const size_t location =
-                FindNextUnusedLocation(used_locations, previous_unused_location);
-            previous_unused_location = location;
-            used_locations.set(location);
-            const Id id{DefineInput(*this, F32[4], true)};
-            Decorate(id, spv::Decoration::Location, location);
-            input_fixed_fnc_textures[index] = id;
+            input_fixed_fnc_textures[index] =
+                DefineLegacyInput(*this, used_locations, previous_unused_location);
         }
     }
     if (stage == Stage::TessellationEval) {
@@ -1356,22 +1436,29 @@ void EmitContext::DefineOutputs(const IR::Program& program) {
     }
     size_t previous_unused_location = 0;
     if (info.stores.AnyComponent(IR::Attribute::ColorFrontDiffuseR)) {
-        const size_t location = FindNextUnusedLocation(used_locations, previous_unused_location);
-        previous_unused_location = location;
-        used_locations.set(location);
-        const Id id{DefineOutput(*this, F32[4], invocations)};
-        Decorate(id, spv::Decoration::Location, static_cast<u32>(location));
-        output_front_color = id;
+        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)) {
-            const size_t location =
-                FindNextUnusedLocation(used_locations, previous_unused_location);
-            previous_unused_location = location;
-            used_locations.set(location);
-            const Id id{DefineOutput(*this, F32[4], invocations)};
-            Decorate(id, spv::Decoration::Location, location);
-            output_fixed_fnc_textures[index] = id;
+            output_fixed_fnc_textures[index] =
+                DefineLegacyOutput(*this, used_locations, previous_unused_location, invocations);
         }
     }
     switch (stage) {

src/shader_recompiler/backend/spirv/emit_context.h

@@ -113,6 +113,9 @@ 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);
     }
@@ -269,12 +272,20 @@ public:
 
     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{};
 

src/shader_recompiler/backend/spirv/emit_spirv_context_get_set.cpp

@@ -43,23 +43,12 @@ Id AttrPointer(EmitContext& ctx, Id pointer_type, Id vertex, Id base, Args&&...
     }
 }
 
-bool IsFixedFncTexture(IR::Attribute attribute) {
-    return attribute >= IR::Attribute::FixedFncTexture0S &&
-           attribute <= IR::Attribute::FixedFncTexture9Q;
-}
-
-u32 FixedFncTextureAttributeIndex(IR::Attribute attribute) {
-    if (!IsFixedFncTexture(attribute)) {
-        throw InvalidArgument("Attribute {} is not a FixedFncTexture", attribute);
-    }
-    return (static_cast<u32>(attribute) - static_cast<u32>(IR::Attribute::FixedFncTexture0S)) / 4u;
-}
-
-u32 FixedFncTextureAttributeElement(IR::Attribute attribute) {
-    if (!IsFixedFncTexture(attribute)) {
-        throw InvalidArgument("Attribute {} is not a FixedFncTexture", attribute);
-    }
-    return static_cast<u32>(attribute) % 4u;
+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>
@@ -93,12 +82,16 @@ std::optional<OutAttr> OutputAttrPointer(EmitContext& ctx, IR::Attribute attr) {
             return OutputAccessChain(ctx, ctx.output_f32, info.id, index_id);
         }
     }
-    if (IsFixedFncTexture(attr)) {
-        const u32 index{FixedFncTextureAttributeIndex(attr)};
-        const u32 element{FixedFncTextureAttributeElement(attr)};
-        const Id element_id{ctx.Const(element)};
-        return OutputAccessChain(ctx, ctx.output_f32, ctx.output_fixed_fnc_textures[index],
-                                 element_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:
@@ -111,14 +104,6 @@ std::optional<OutAttr> OutputAttrPointer(EmitContext& ctx, IR::Attribute attr) {
         const Id element_id{ctx.Const(element)};
         return OutputAccessChain(ctx, ctx.output_f32, ctx.output_position, element_id);
     }
-    case IR::Attribute::ColorFrontDiffuseR:
-    case IR::Attribute::ColorFrontDiffuseG:
-    case IR::Attribute::ColorFrontDiffuseB:
-    case IR::Attribute::ColorFrontDiffuseA: {
-        const u32 element{static_cast<u32>(attr) % 4};
-        const Id element_id{ctx.Const(element)};
-        return OutputAccessChain(ctx, ctx.output_f32, ctx.output_front_color, element_id);
-    }
     case IR::Attribute::ClipDistance0:
     case IR::Attribute::ClipDistance1:
     case IR::Attribute::ClipDistance2:
@@ -341,11 +326,17 @@ 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 (IsFixedFncTexture(attr)) {
-        const u32 index{FixedFncTextureAttributeIndex(attr)};
-        const Id attr_id{ctx.input_fixed_fnc_textures[index]};
-        const Id attr_ptr{AttrPointer(ctx, ctx.input_f32, vertex, attr_id, ctx.Const(element))};
-        return ctx.OpLoad(ctx.F32[1], attr_ptr);
+    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:
@@ -356,13 +347,6 @@ Id EmitGetAttribute(EmitContext& ctx, IR::Attribute attr, Id vertex) {
     case IR::Attribute::PositionW:
         return ctx.OpLoad(ctx.F32[1], AttrPointer(ctx, ctx.input_f32, vertex, ctx.input_position,
                                                   ctx.Const(element)));
-    case IR::Attribute::ColorFrontDiffuseR:
-    case IR::Attribute::ColorFrontDiffuseG:
-    case IR::Attribute::ColorFrontDiffuseB:
-    case IR::Attribute::ColorFrontDiffuseA: {
-        return ctx.OpLoad(ctx.F32[1], AttrPointer(ctx, ctx.input_f32, vertex, ctx.input_front_color,
-                                                  ctx.Const(element)));
-    }
     case IR::Attribute::InstanceId:
         if (ctx.profile.support_vertex_instance_id) {
             return ctx.OpBitcast(ctx.F32[1], ctx.OpLoad(ctx.U32[1], ctx.instance_id));

src/shader_recompiler/ir_opt/texture_pass.cpp

@@ -492,7 +492,7 @@ void TexturePass(Environment& env, IR::Program& program) {
             const auto insert_point{IR::Block::InstructionList::s_iterator_to(*inst)};
             IR::IREmitter ir{*texture_inst.block, insert_point};
             const IR::U32 shift{ir.Imm32(std::countr_zero(DESCRIPTOR_SIZE))};
-            inst->SetArg(0, ir.SMin(ir.ShiftRightArithmetic(cbuf.dynamic_offset, shift),
+            inst->SetArg(0, ir.UMin(ir.ShiftRightArithmetic(cbuf.dynamic_offset, shift),
                                     ir.Imm32(DESCRIPTOR_SIZE - 1)));
         } else {
             inst->SetArg(0, IR::Value{});

src/video_core/dirty_flags.h

@@ -38,6 +38,9 @@ enum : u8 {
 
     Shaders,
 
+    // Special entries
+    DepthBiasGlobal,
+
     LastCommonEntry,
 };
 

src/video_core/gpu_thread.cpp

@@ -32,10 +32,7 @@ static void RunThread(std::stop_token stop_token, Core::System& system,
     VideoCore::RasterizerInterface* const rasterizer = renderer.ReadRasterizer();
 
     while (!stop_token.stop_requested()) {
-        CommandDataContainer next;
-        if (!state.queue.try_pop(next)) {
-            continue;
-        }
+        CommandDataContainer next = state.queue.PopWait(stop_token);
         if (stop_token.stop_requested()) {
             break;
         }
@@ -122,7 +119,7 @@ u64 ThreadManager::PushCommand(CommandData&& command_data, bool block) {
 
     std::unique_lock lk(state.write_lock);
     const u64 fence{++state.last_fence};
-    state.queue.push(CommandDataContainer(std::move(command_data), fence, block));
+    state.queue.Push(CommandDataContainer(std::move(command_data), fence, block));
 
     if (block) {
         state.cv.wait(lk, thread.get_stop_token(), [this, fence] {

src/video_core/gpu_thread.h

@@ -11,7 +11,7 @@
 #include <thread>
 #include <variant>
 
-#include "common/atomic_threadsafe_queue.h"
+#include "common/threadsafe_queue.h"
 #include "video_core/framebuffer_config.h"
 
 namespace Tegra {
@@ -97,9 +97,9 @@ struct CommandDataContainer {
 
 /// Struct used to synchronize the GPU thread
 struct SynchState final {
-    using CommandQueue = Common::MPMCQueue<CommandDataContainer>;
+    using CommandQueue = Common::SPSCQueue<CommandDataContainer, true>;
     std::mutex write_lock;
-    CommandQueue queue{100000};
+    CommandQueue queue;
     u64 last_fence{};
     std::atomic<u64> signaled_fence{};
     std::condition_variable_any cv;

src/video_core/renderer_vulkan/vk_rasterizer.cpp

@@ -627,9 +627,21 @@ void RasterizerVulkan::UpdateDepthBias(Tegra::Engines::Maxwell3D::Regs& regs) {
     if (!state_tracker.TouchDepthBias()) {
         return;
     }
-    scheduler.Record([constant = regs.polygon_offset_units, clamp = regs.polygon_offset_clamp,
+    float units = regs.polygon_offset_units / 2.0f;
+    const bool is_d24 = regs.zeta.format == Tegra::DepthFormat::S8_UINT_Z24_UNORM ||
+                        regs.zeta.format == Tegra::DepthFormat::D24X8_UNORM ||
+                        regs.zeta.format == Tegra::DepthFormat::D24S8_UNORM ||
+                        regs.zeta.format == Tegra::DepthFormat::D24C8_UNORM;
+    if (is_d24 && !device.SupportsD24DepthBuffer()) {
+        // the base formulas can be obtained from here:
+        //   https://docs.microsoft.com/en-us/windows/win32/direct3d11/d3d10-graphics-programming-guide-output-merger-stage-depth-bias
+        const double rescale_factor =
+            static_cast<double>(1ULL << (32 - 24)) / (static_cast<double>(0x1.ep+127));
+        units = static_cast<float>(static_cast<double>(units) * rescale_factor);
+    }
+    scheduler.Record([constant = units, clamp = regs.polygon_offset_clamp,
                       factor = regs.polygon_offset_factor](vk::CommandBuffer cmdbuf) {
-        cmdbuf.SetDepthBias(constant, clamp, factor / 2.0f);
+        cmdbuf.SetDepthBias(constant, clamp, factor);
     });
 }
 

src/video_core/renderer_vulkan/vk_state_tracker.cpp

@@ -54,6 +54,7 @@ void SetupDirtyViewports(Tables& tables) {
     FillBlock(tables[0], OFF(viewport_transform), NUM(viewport_transform), Viewports);
     FillBlock(tables[0], OFF(viewports), NUM(viewports), Viewports);
     tables[0][OFF(viewport_transform_enabled)] = Viewports;
+    tables[1][OFF(screen_y_control)] = Viewports;
 }
 
 void SetupDirtyScissors(Tables& tables) {

src/video_core/renderer_vulkan/vk_state_tracker.h

@@ -79,7 +79,8 @@ public:
     }
 
     bool TouchDepthBias() {
-        return Exchange(Dirty::DepthBias, false);
+        return Exchange(Dirty::DepthBias, false) ||
+               Exchange(VideoCommon::Dirty::DepthBiasGlobal, false);
     }
 
     bool TouchBlendConstants() {

src/video_core/texture_cache/texture_cache.h

@@ -221,6 +221,7 @@ void TextureCache<P>::UpdateRenderTargets(bool is_clear) {
         BindRenderTarget(&render_targets.depth_buffer_id, FindDepthBuffer(is_clear));
     }
     const ImageViewId depth_buffer_id = render_targets.depth_buffer_id;
+
     PrepareImageView(depth_buffer_id, true, is_clear && IsFullClear(depth_buffer_id));
 
     for (size_t index = 0; index < NUM_RT; ++index) {
@@ -230,6 +231,8 @@ void TextureCache<P>::UpdateRenderTargets(bool is_clear) {
         maxwell3d.regs.render_area.width,
         maxwell3d.regs.render_area.height,
     };
+
+    flags[Dirty::DepthBiasGlobal] = true;
 }
 
 template <class P>

src/video_core/vulkan_common/vulkan_device.cpp

@@ -623,6 +623,10 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
         is_float16_supported = false;
     }
 
+    supports_d24_depth =
+        IsFormatSupported(VK_FORMAT_D24_UNORM_S8_UINT,
+                          VK_FORMAT_FEATURE_DEPTH_STENCIL_ATTACHMENT_BIT, FormatType::Optimal);
+
     graphics_queue = logical.GetQueue(graphics_family);
     present_queue = logical.GetQueue(present_family);
 }

src/video_core/vulkan_common/vulkan_device.h

@@ -332,6 +332,10 @@ public:
         return sets_per_pool;
     }
 
+    bool SupportsD24DepthBuffer() const {
+        return supports_d24_depth;
+    }
+
 private:
     /// Checks if the physical device is suitable.
     void CheckSuitability(bool requires_swapchain) const;
@@ -425,6 +429,7 @@ private:
     bool has_broken_cube_compatibility{};   ///< Has broken cube compatiblity bit
     bool has_renderdoc{};                   ///< Has RenderDoc attached
     bool has_nsight_graphics{};             ///< Has Nsight Graphics attached
+    bool supports_d24_depth{};              ///< Supports D24 depth buffers.
 
     // Telemetry parameters
     std::string vendor_name;                       ///< Device's driver name.