format src and tests with clang-format

The README had macOS badges and an experimental-support note but no build,
run, or system-requirement information for the platform. Document the
Rosetta 2 / MoltenVK setup, the x86-64 configure invocation, the Qt
universal-build requirement, MoltenVK installation and signing, and the
SDL_VULKAN_LIBRARY variable needed at run time.

README.md

@@ -27,8 +27,9 @@ Development is focused on compatibility and boot reliability.
 Windows is the primary platform and receives the most testing. Linux builds and runs; see
 [Building on Linux](#building-on-linux).
 
-macOS support is experimental. Compatibility with the same games on Windows and macOS has not yet
-been tested.
+macOS support is experimental. The emulator is built for x86-64 and runs on Apple Silicon under
+Rosetta 2, with Vulkan provided by MoltenVK. A small number of titles have been verified in-game
+on Apple Silicon hardware; see [Building on macOS](#building-on-macos).
 
 ## Bugs and Issues
 
@@ -114,9 +115,10 @@ the Vulkan/SPIR-V validation rules.
 
 ### System requirements
 
-- Windows 10 version 1803, or a current Linux distribution
-- A 64-bit x86 processor
-- A Vulkan 1.3-capable GPU with current drivers
+- Windows 10 version 1803, a current Linux distribution, or macOS on Apple Silicon
+- A 64-bit x86 processor (on macOS, an Apple Silicon processor with Rosetta 2)
+- A Vulkan 1.3-capable GPU with current drivers (on macOS, Vulkan is provided by the bundled
+  MoltenVK)
 
 ### Build requirements (Windows)
 
@@ -188,6 +190,45 @@ time.
 
 Note that the CMake source root is `src`, not the repository root.
 
+### Building on macOS
+
+macOS builds target x86-64 and run under Rosetta 2 on Apple Silicon, so the PS5's x86-64 game
+code executes through the same translation layer as the emulator itself. Prebuilt archives are
+attached to releases; the steps below are for building from source.
+
+Requirements:
+
+- An Apple Silicon Mac with Rosetta 2 installed (`softwareupdate --install-rosetta`)
+- Xcode (or the Command Line Tools)
+- Homebrew packages: `brew install cmake ninja glslang`
+- Qt 6 (Concurrent, Network, Widgets) with x86-64 support. The official Qt installation is
+  universal and works; Homebrew's Qt is arm64-only and will not link
+
+```bash
+git submodule update --init --recursive
+
+cmake -S src -B _Build/macos -G Ninja -DCMAKE_BUILD_TYPE=Release \
+  -DCMAKE_OSX_ARCHITECTURES=x86_64 \
+  -DCMAKE_C_COMPILER=clang -DCMAKE_CXX_COMPILER=clang++ \
+  -DCMAKE_PREFIX_PATH="$Qt6_DIR"
+
+cmake --build _Build/macos --target launcher --parallel
+cmake --install _Build/macos --prefix _Build/macos/install
+```
+
+The build re-signs `kyty_emulator` with the JIT entitlements it needs to execute translated
+guest code; no manual signing step is required.
+
+Vulkan comes from MoltenVK. Download `MoltenVK-macos.tar` from the
+[MoltenVK releases](https://github.com/KhronosGroup/MoltenVK/releases), then copy
+`MoltenVK/dynamic/dylib/macOS/libMoltenVK.dylib` next to `kyty_emulator` and ad-hoc sign it:
+
+```bash
+codesign --force --sign - _Build/macos/install/libMoltenVK.dylib
+```
+
+Release archives already include a signed `libMoltenVK.dylib`.
+
 ### Visual Studio Code
 
 A ready-made Visual Studio Code setup is included in [`.vscode`](.vscode). It configures CMake
@@ -233,6 +274,14 @@ The emulator can also be started directly with a legally obtained game directory
 ./_Build/linux/install/kyty_emulator --game "/games/ExampleGame"
 ```
 
+On macOS, point SDL at the MoltenVK library explicitly; the hardened runtime prevents it from
+being picked up from the executable's directory:
+
+```bash
+cd _Build/macos/install
+SDL_VULKAN_LIBRARY="$PWD/libMoltenVK.dylib" ./kyty_emulator --game "/games/ExampleGame"
+```
+
 Run `kyty_emulator --help` to see the available graphics, logging, validation, profiling, and
 debugging options.
 

src/CMakeLists.txt

@@ -329,6 +329,7 @@ add_kyty_full_emulator_test(shader_cfg_tests ../tests/shaderCfgTests.cpp)
 add_executable(scalar_provenance_tests EXCLUDE_FROM_ALL
 	../tests/ScalarProvenanceTests.cpp
 	graphics/host_gpu/hostMemory.cpp
+	graphics/shader/recompiler/ir/ReadLaneElimination.cpp
 	graphics/shader/recompiler/ir/ScalarProvenance.cpp
 	graphics/shader/recompiler/ir/SrtWalker.cpp
 )
@@ -441,6 +442,7 @@ if(NOT KYTY_CLANG_CL)
 endif()
 
 if(BUILD_TESTING)
+	add_test(NAME scalar_provenance COMMAND $<TARGET_FILE:scalar_provenance_tests>)
 	add_test(NAME image_page_table COMMAND $<TARGET_FILE:image_page_table_tests>)
 	add_test(NAME memory_tracker COMMAND $<TARGET_FILE:memory_tracker_tests>)
 	add_test(NAME page_manager COMMAND $<TARGET_FILE:page_manager_tests>)

src/common/hostException.cpp

@@ -175,9 +175,9 @@ static void SignalHandler(int sig, siginfo_t* si, void* uctx) {
 	}
 	g_in_exception_filter = true;
 
-	auto*       uc  = static_cast<ucontext_t*>(uctx);
-	const auto* mc  = uc->uc_mcontext;
-	const auto& ss  = mc->__ss;
+	auto*       uc = static_cast<ucontext_t*>(uctx);
+	const auto* mc = uc->uc_mcontext;
+	const auto& ss = mc->__ss;
 
 	ExceptionInfo info {};
 	info.exception_address = ss.__rip;
@@ -214,7 +214,7 @@ static void SignalHandler(int sig, siginfo_t* si, void* uctx) {
 		FailFast("host exception callback is null");
 	}
 
-	const bool resolved = handler(info);
+	const bool resolved   = handler(info);
 	g_in_exception_filter = false;
 
 	if (resolved) {
@@ -255,8 +255,8 @@ static void SignalHandler(int signal_number, siginfo_t* signal_info, void* nativ
 	info.native_context    = context;
 
 	if (signal_number == SIGSEGV || signal_number == SIGBUS) {
-		info.type              = ExceptionType::AccessViolation;
-		const auto error_code  = static_cast<uint64_t>(gregs[REG_ERR]);
+		info.type             = ExceptionType::AccessViolation;
+		const auto error_code = static_cast<uint64_t>(gregs[REG_ERR]);
 		if ((error_code & PAGE_FAULT_ERROR_INSTRUCTION) != 0) {
 			info.access_violation_type = AccessViolationType::Execute;
 		} else if ((error_code & PAGE_FAULT_ERROR_WRITE) != 0) {

src/common/lruCache.h

@@ -19,10 +19,10 @@ class LeastRecentlyUsedCache {
 
 public:
 	[[nodiscard]] size_t Insert(Object object, Tick tick) {
-		const auto id = Build();
+		const auto id   = Build();
 		auto&      item = m_items[id];
-		item.object = std::move(object);
-		item.tick   = tick;
+		item.object     = std::move(object);
+		item.tick       = tick;
 		Attach(item);
 		return id;
 	}
@@ -49,8 +49,7 @@ public:
 
 	template <typename Function>
 	void ForEachItemBelow(Tick tick, Function&& function) {
-		constexpr bool ReturnsBool =
-		    std::is_same_v<std::invoke_result_t<Function, Object>, bool>;
+		constexpr bool ReturnsBool = std::is_same_v<std::invoke_result_t<Function, Object>, bool>;
 		for (auto* item = m_first; item != nullptr;) {
 			if (item->tick > tick) {
 				return;
@@ -87,10 +86,10 @@ private:
 			m_last = &item;
 			return;
 		}
-		item.prev   = m_last;
+		item.prev    = m_last;
 		m_last->next = &item;
-		item.next   = nullptr;
-		m_last      = &item;
+		item.next    = nullptr;
+		m_last       = &item;
 	}
 
 	void Detach(Item& item) {

src/common/platform/sysLinuxDbg.cpp

@@ -31,10 +31,9 @@ static bool OnOwnStack() {
 	if (pthread_getattr_np(pthread_self(), &attr) != 0) {
 		return false;
 	}
-	void*  base = nullptr;
-	size_t size = 0;
-	const bool ok =
-	    pthread_attr_getstack(&attr, &base, &size) == 0 && base != nullptr && size != 0;
+	void*      base = nullptr;
+	size_t     size = 0;
+	const bool ok = pthread_attr_getstack(&attr, &base, &size) == 0 && base != nullptr && size != 0;
 	pthread_attr_destroy(&attr);
 	if (!ok) {
 		return false;

src/common/platform/sysLinuxFileIO.cpp

@@ -172,8 +172,7 @@ sys_file_t* SysFileCreate(const std::filesystem::path& file_name) {
 	return ret;
 }
 
-sys_file_t* SysFileOpenR(const std::filesystem::path& file_name,
-                         sys_file_cache_type_t cache_type) {
+sys_file_t* SysFileOpenR(const std::filesystem::path& file_name, sys_file_cache_type_t cache_type) {
 	auto* ret = new sys_file_t;
 
 	ret->type = SYS_FILE_FILE;
@@ -218,8 +217,7 @@ sys_file_t* SysFileCreate() {
 	return ret;
 }
 
-sys_file_t* SysFileOpenW(const std::filesystem::path& file_name,
-                         sys_file_cache_type_t cache_type) {
+sys_file_t* SysFileOpenW(const std::filesystem::path& file_name, sys_file_cache_type_t cache_type) {
 	auto* ret = new sys_file_t;
 
 	auto real_name     = get_internal_name(file_name);
@@ -241,7 +239,7 @@ sys_file_t* SysFileOpenW(const std::filesystem::path& file_name,
 }
 
 sys_file_t* SysFileOpenRw(const std::filesystem::path& file_name,
-                          sys_file_cache_type_t cache_type) {
+                          sys_file_cache_type_t        cache_type) {
 	auto* ret = new sys_file_t;
 
 	auto real_name     = get_internal_name(file_name);

src/common/platform/sysLinuxVirtual.cpp

@@ -136,8 +136,8 @@ static void* map_anonymous(uintptr_t addr, size_t size, int protect, int flags)
 			break;
 		}
 		const auto hint = (top - step) & ~(LOW_ARENA_GRAIN - 1);
-		void*      ptr  = mmap(reinterpret_cast<void*>(hint), size, protect,
-		                       flags | MAP_FIXED_NOREPLACE, -1, 0); // NOLINT
+		void* ptr = mmap(reinterpret_cast<void*>(hint), size, protect, flags | MAP_FIXED_NOREPLACE,
+		                 -1, 0); // NOLINT
 		if (ptr != MAP_FAILED) {
 			return ptr;
 		}
@@ -161,8 +161,8 @@ uint64_t SysVirtualAlloc(uint64_t address, uint64_t size, VirtualMemory::Mode mo
 	if (ptr != MAP_FAILED) {
 		pthread_mutex_lock(&g_virtual_mutex);
 		record_alloc(ret_addr, size);
-		uintptr_t page_start  = ret_addr >> 12u;
-		uintptr_t page_end    = (ret_addr + size - 1) >> 12u;
+		uintptr_t page_start = ret_addr >> 12u;
+		uintptr_t page_end   = (ret_addr + size - 1) >> 12u;
 		for (uintptr_t page = page_start; page <= page_end; page++) {
 			(*g_protects)[page] = protect;
 		}
@@ -194,8 +194,8 @@ uint64_t SysVirtualAllocAligned(uint64_t address, uint64_t size, VirtualMemory::
 	if (ptr != MAP_FAILED && ((ret_addr & (alignment - 1)) != 0)) {
 		munmap(ptr, size);
 
-		ptr      = map_anonymous(addr, size + alignment, protect,
-		                         MAP_PRIVATE | MAP_ANON | MAP_NORESERVE);
+		ptr =
+		    map_anonymous(addr, size + alignment, protect, MAP_PRIVATE | MAP_ANON | MAP_NORESERVE);
 		ret_addr = reinterpret_cast<uintptr_t>(ptr);
 		if (ptr != MAP_FAILED) {
 #if defined(__APPLE__)
@@ -251,8 +251,8 @@ uint64_t SysVirtualAllocAligned(uint64_t address, uint64_t size, VirtualMemory::
 
 	pthread_mutex_lock(&g_virtual_mutex);
 	record_alloc(ret_addr, size);
-	uintptr_t page_start  = ret_addr >> 12u;
-	uintptr_t page_end    = (ret_addr + size - 1) >> 12u;
+	uintptr_t page_start = ret_addr >> 12u;
+	uintptr_t page_end   = (ret_addr + size - 1) >> 12u;
 	for (uintptr_t page = page_start; page <= page_end; page++) {
 		(*g_protects)[page] = protect;
 	}
@@ -266,9 +266,9 @@ uint64_t SysVirtualAllocAligned(uint64_t address, uint64_t size, VirtualMemory::
 // the first mapped region at or above `region_addr`; if it begins before the end of the
 // requested range, the range overlaps an existing mapping.
 static bool is_mapped(void* ptr, size_t length) {
-	auto              query_addr  = reinterpret_cast<mach_vm_address_t>(ptr);
-	mach_vm_address_t region_addr = query_addr;
-	mach_vm_size_t    region_size = 0;
+	auto                           query_addr  = reinterpret_cast<mach_vm_address_t>(ptr);
+	mach_vm_address_t              region_addr = query_addr;
+	mach_vm_size_t                 region_size = 0;
 	vm_region_basic_info_data_64_t info {};
 	mach_msg_type_number_t         count       = VM_REGION_BASIC_INFO_COUNT_64;
 	mach_port_t                    object_name = MACH_PORT_NULL;
@@ -337,8 +337,8 @@ bool SysVirtualAllocFixed(uint64_t address, uint64_t size, VirtualMemory::Mode m
 	if (ptr != MAP_FAILED) {
 		pthread_mutex_lock(&g_virtual_mutex);
 		record_alloc(ret_addr, size);
-		uintptr_t page_start  = ret_addr >> 12u;
-		uintptr_t page_end    = (ret_addr + size - 1) >> 12u;
+		uintptr_t page_start = ret_addr >> 12u;
+		uintptr_t page_end   = (ret_addr + size - 1) >> 12u;
 		for (uintptr_t page = page_start; page <= page_end; page++) {
 			(*g_protects)[page] = protect;
 		}

src/common/threads.cpp

@@ -5,9 +5,9 @@
 
 #include <algorithm>
 #include <atomic>
+#include <cerrno>
 #include <chrono>             // IWYU pragma: keep
 #include <condition_variable> // IWYU pragma: keep
-#include <cerrno>
 #include <mutex>
 #include <vector>
 

src/common/uniqueFunction.h

@@ -11,7 +11,7 @@ template <typename Result, typename... Args>
 class UniqueFunction {
 	class CallableBase {
 	public:
-		virtual ~CallableBase() = default;
+		virtual ~CallableBase()               = default;
 		virtual Result Invoke(Args&&... args) = 0;
 	};
 
@@ -20,9 +20,7 @@ class UniqueFunction {
 	public:
 		explicit Callable(Function function): m_function(std::move(function)) {}
 
-		Result Invoke(Args&&... args) override {
-			return m_function(std::forward<Args>(args)...);
-		}
+		Result Invoke(Args&&... args) override { return m_function(std::forward<Args>(args)...); }
 
 	private:
 		Function m_function;

src/graphics/guest_gpu/command_processor/commandProcessor.h

@@ -158,7 +158,7 @@ private:
 	void                 CheckBuffer() const { GetScheduler().CheckActive(); }
 	GpuResourceManager&  GetGpuResources() const { return m_renderer.GetGpuResources(); }
 
-	RenderContext&    m_renderer;
+	RenderContext&   m_renderer;
 	HW::Context      m_ctx;
 	HW::UserConfig   m_ucfg;
 	HW::Shader       m_sh_ctx;
@@ -170,9 +170,9 @@ private:
 	uint64_t         m_dispatch_indirect_args_base_addr = 0;
 	uint32_t         m_num_instances                    = 1;
 
-	uint32_t m_de_count          = 0;
-	uint32_t m_ce_count          = 0;
-	bool     m_ce_complete       = false;
+	uint32_t m_de_count        = 0;
+	uint32_t m_ce_count        = 0;
+	bool     m_ce_complete     = false;
 	bool     m_readback_active = false;
 
 	uint32_t m_const_ram[0x3000] = {0};

src/graphics/guest_gpu/gpu_defs.h

@@ -374,6 +374,8 @@ enum class BufferFormat : uint32_t {
 	k32_32_32_32UInt    = 75,
 	k32_32_32_32SInt    = 76,
 	k32_32_32_32Float   = 77,
+	k8Srgb              = 128,
+	k8_8Srgb            = 129,
 	k8_8_8_8Srgb        = 130,
 	k9_9_9_5Float       = 132,
 	k5_6_5UNorm         = 133,

src/graphics/guest_gpu/gpu_format.cpp

@@ -57,6 +57,8 @@ constexpr FormatInfo kFormatInfo[] = {
     {GpuEnumValue(BufferFormat::k32_32_32_32UInt), 16, 0, 16, true, true},
     {GpuEnumValue(BufferFormat::k32_32_32_32SInt), 16, 0, 16, false, false},
     {GpuEnumValue(BufferFormat::k32_32_32_32Float), 16, 0, 16, true, false},
+    {GpuEnumValue(BufferFormat::k8Srgb), 1, 0, 0, true, false},
+    {GpuEnumValue(BufferFormat::k8_8Srgb), 2, 0, 0, true, false},
     {GpuEnumValue(BufferFormat::k8_8_8_8Srgb), 4, 0, 4, true, false},
     {GpuEnumValue(BufferFormat::k9_9_9_5Float), 4, 0, 0, true, false},
     {GpuEnumValue(BufferFormat::k5_6_5UNorm), 2, 0, 2, true, false},

src/graphics/guest_gpu/graphicsRun.cpp

@@ -962,9 +962,8 @@ void CommandProcessor::DrawIndexOffset(uint32_t index_offset, uint32_t index_cou
 	auto* index_addr = reinterpret_cast<const void*>(
 	    m_index_base_addr + static_cast<uint64_t>(index_offset) * index_size);
 
-	m_renderer.GetRenderExecutor().DrawIndex(m_submit_id, CurrentBuffer(),
-	                                             m_index_type_and_size, index_count, index_addr,
-	                                             flags, 1, m_num_instances);
+	m_renderer.GetRenderExecutor().DrawIndex(m_submit_id, CurrentBuffer(), m_index_type_and_size,
+	                                         index_count, index_addr, flags, 1, m_num_instances);
 }
 
 void CommandProcessor::DrawIndirect(uint32_t data_offset, uint32_t draw_initiator, bool indexed) {
@@ -1190,8 +1189,8 @@ void CommandProcessor::DispatchDirect(uint32_t thread_group_x, uint32_t thread_g
 			}
 		}
 
-		m_renderer.GetRenderExecutor().DispatchDirect(
-		    m_submit_id, CurrentBuffer(), thread_group_x, thread_group_y, thread_group_z, mode);
+		m_renderer.GetRenderExecutor().DispatchDirect(m_submit_id, CurrentBuffer(), thread_group_x,
+		                                              thread_group_y, thread_group_z, mode);
 	}
 
 	constexpr uint32_t DispatchInitiatorUseThreadDimensions = 1u << 5u;
@@ -1237,16 +1236,16 @@ void CommandProcessor::DrawIndexAuto(uint32_t index_count, uint32_t flags,
                                      uint32_t first_vertex, uint32_t first_instance) {
 	CheckBuffer();
 
-	m_renderer.GetRenderExecutor().DrawAuto(
-	    m_submit_id, CurrentBuffer(), index_count, flags, render_target_slice_offset,
-	    instance_count, first_vertex, first_instance);
+	m_renderer.GetRenderExecutor().DrawAuto(m_submit_id, CurrentBuffer(), index_count, flags,
+	                                        render_target_slice_offset, instance_count,
+	                                        first_vertex, first_instance);
 }
 
 void CommandProcessor::WaitFlipDone(uint32_t video_out_handle, uint32_t display_buffer_index) {
 	BufferFlush();
 
 	m_renderer.GetVideoOut().WaitFlipDone(static_cast<int>(video_out_handle),
-	                                              static_cast<int>(display_buffer_index));
+	                                      static_cast<int>(display_buffer_index));
 }
 
 template <typename T>
@@ -1317,8 +1316,8 @@ void CommandProcessor::WriteAtEndOfPipe(uint32_t cache_policy, uint32_t event_wr
 				if (eop_event_type == 0x2f && cache_action == 0x00 && event_index == 0x06) {
 					auto* dst = static_cast<uint32_t*>(dst_gpu_addr);
 					SynchronizeGpu();
-					Sync::ReadGds(m_renderer.GetBufferCache().GetGdsBuffer(), dst,
-					              value & 0xffffu, value >> 16u);
+					Sync::ReadGds(m_renderer.GetBufferCache().GetGdsBuffer(), dst, value & 0xffffu,
+					              value >> 16u);
 					Sync::WriteAtEndOfPipeGds32(m_submit_id, CurrentBuffer(), dst, value & 0xffffu,
 					                            value >> 16u);
 					return;
@@ -1486,8 +1485,7 @@ void CommandProcessor::EmitGlobalBarrier() {
 	barrier.srcStageMask  = vk::PipelineStageFlagBits2::eAllCommands;
 	barrier.srcAccessMask = vk::AccessFlagBits2::eMemoryWrite;
 	barrier.dstStageMask  = vk::PipelineStageFlagBits2::eAllCommands;
-	barrier.dstAccessMask =
-	    vk::AccessFlagBits2::eMemoryRead | vk::AccessFlagBits2::eMemoryWrite;
+	barrier.dstAccessMask = vk::AccessFlagBits2::eMemoryRead | vk::AccessFlagBits2::eMemoryWrite;
 
 	vk::DependencyInfo dependency {};
 	dependency.memoryBarrierCount = 1;

src/graphics/guest_gpu/tile.h

@@ -65,41 +65,41 @@ struct TileVolumeLayout {
 };
 
 bool TileGetBlockLayout(TileBlockFamily family, uint32_t bytes_per_element,
-	                    TileBlockLayout& layout);
+                        TileBlockLayout& layout);
 bool TileGetBlockOffset(const TileBlockLayout& layout, uint32_t x, uint32_t y, uint32_t z,
-	                    uint32_t& byte_offset);
+                        uint32_t& byte_offset);
 bool TileGetBlockXor(const TileBlockLayout& layout, uint32_t block_x, uint32_t block_y,
-	                 uint32_t& byte_offset);
+                     uint32_t& byte_offset);
 bool TileGetBlockXor(const TileBlockLayout& layout, uint32_t block_x, uint32_t block_y,
-	                 uint32_t block_z, uint32_t& byte_offset);
+                     uint32_t block_z, uint32_t& byte_offset);
 bool TileIsStandard256BTextureSupported(uint32_t format);
 bool TileIsStandard4KBTextureSupported(uint32_t format);
 bool TileIsStandard64KBTextureSupported(uint32_t format);
 bool TileGetTextureVolumeLayout(uint32_t format, uint32_t width, uint32_t height, uint32_t depth,
-	                            uint32_t levels, uint32_t tile, TileVolumeLayout& layout);
+                                uint32_t levels, uint32_t tile, TileVolumeLayout& layout);
 
 bool     TileGetHtileSize(uint32_t width, uint32_t height, TileSizeAlign& htile_size);
 bool     TileGetDepthSize(uint32_t width, uint32_t height, uint32_t pitch, uint32_t z_format,
-	                      uint32_t stencil_format, bool htile, TileSizeAlign& stencil_size,
-	                      TileSizeAlign& htile_size, TileSizeAlign& depth_size,
+                          uint32_t stencil_format, bool htile, TileSizeAlign& stencil_size,
+                          TileSizeAlign& htile_size, TileSizeAlign& depth_size,
                           uint32_t num_fragments_log2 = 0);
 uint32_t TileGetRenderTargetPitch(uint32_t width, uint32_t bytes_per_element,
                                   uint32_t num_fragments_log2 = 0);
 uint32_t TileGetDepthPitch(uint32_t width, uint32_t bytes_per_element,
                            uint32_t num_fragments_log2 = 0);
 bool     TileGetRenderTargetSize(uint32_t width, uint32_t height, uint32_t pitch,
-	                             uint32_t bytes_per_element, TileSizeAlign& total_size,
+                                 uint32_t bytes_per_element, TileSizeAlign& total_size,
                                  uint32_t num_fragments_log2 = 0);
 bool     TileGetRenderTargetMipLayout(uint32_t width, uint32_t height, uint32_t pitch,
-	                                  uint32_t bytes_per_element, uint32_t levels,
-	                                  TileSizeAlign& total_size, TileSizeOffset* level_sizes,
+                                      uint32_t bytes_per_element, uint32_t levels,
+                                      TileSizeAlign& total_size, TileSizeOffset* level_sizes,
                                       TilePaddedSize* padded_size);
 void     TileGetTextureSize(uint32_t format, uint32_t width, uint32_t height, uint32_t pitch,
                             uint32_t levels, uint32_t tile, TileSizeAlign* total_size,
                             TileSizeOffset* level_sizes, TilePaddedSize* padded_size);
 void TileGetTextureTotalSize(uint32_t format, uint32_t width, uint32_t height, uint32_t depth,
                              uint32_t pitch, uint32_t levels, uint32_t tile, bool volume_texture,
-	                         TileSizeAlign& total_size);
+                             TileSizeAlign& total_size);
 uint32_t TileGetTexturePitch(uint32_t format, uint32_t width, uint32_t levels, uint32_t tile);
 
 } // namespace Libs::Graphics

src/graphics/host_gpu/graphicContext.h

@@ -60,19 +60,19 @@ struct VulkanImage {
 	VulkanImage() = default;
 	KYTY_CLASS_NO_COPY(VulkanImage);
 
-	vk::Format             format      = vk::Format::eUndefined;
-	vk::ImageType          image_type  = vk::ImageType::e2D;
-	vk::Extent3D           extent      = {1, 1, 1};
-	uint32_t               guest_pitch = 0;
-	uint32_t               layers      = 1;
-	uint32_t               mip_levels  = 1;
-	uint32_t               samples     = 1;
-	vk::ImageUsageFlags    usage       = {};
-	vk::ImageCreateFlags   flags       = {};
-	vk::Image              image       = nullptr;
-	VulkanImageState       state;
+	vk::Format                    format      = vk::Format::eUndefined;
+	vk::ImageType                 image_type  = vk::ImageType::e2D;
+	vk::Extent3D                  extent      = {1, 1, 1};
+	uint32_t                      guest_pitch = 0;
+	uint32_t                      layers      = 1;
+	uint32_t                      mip_levels  = 1;
+	uint32_t                      samples     = 1;
+	vk::ImageUsageFlags           usage       = {};
+	vk::ImageCreateFlags          flags       = {};
+	vk::Image                     image       = nullptr;
+	VulkanImageState              state;
 	std::vector<VulkanImageState> subresource_states;
-	Graphics::VulkanMemory memory;
+	Graphics::VulkanMemory        memory;
 };
 
 struct VulkanBuffer {

src/graphics/host_gpu/hostMemory.cpp

@@ -30,7 +30,7 @@ bool IsAccessible(DWORD protect, HostMemoryAccess access) {
 } // namespace
 
 bool HostMemoryQueryRange(uint64_t addr, uint64_t requested_size, HostMemoryAccess access,
-	                      uint64_t& accessible_size) {
+                          uint64_t& accessible_size) {
 	accessible_size = 0;
 	if (addr == 0 || requested_size == 0) {
 		return false;

src/graphics/host_gpu/hostMemory.h

@@ -8,7 +8,7 @@ namespace Libs::Graphics {
 enum class HostMemoryAccess { Read, Mapped };
 
 bool HostMemoryQueryRange(uint64_t addr, uint64_t requested_size, HostMemoryAccess access,
-	                      uint64_t& accessible_size);
+                          uint64_t& accessible_size);
 bool HostMemoryQueryReadable(uint64_t addr, uint64_t requested_size, uint64_t& readable_size);
 bool HostMemoryIsReadable(uint64_t addr);
 bool HostMemoryRangeIsReadable(uint64_t addr, uint64_t size);

src/graphics/host_gpu/renderer/cache/streamBuffer.cpp

@@ -135,8 +135,8 @@ void Buffer::Write(uint64_t offset, const void* source, uint64_t size) {
 void Buffer::Flush(uint64_t offset, uint64_t size) {
 	EXIT_IF(m_mapped.empty() || offset > m_size || size > m_size - offset);
 	if (!m_is_coherent && size != 0) {
-		const auto result = vmaFlushAllocation(m_graphics->allocator, m_buffer->memory.allocation,
-		                                      offset, size);
+		const auto result =
+		    vmaFlushAllocation(m_graphics->allocator, m_buffer->memory.allocation, offset, size);
 		EXIT_NOT_IMPLEMENTED(static_cast<vk::Result>(result) != vk::Result::eSuccess);
 	}
 }
@@ -144,8 +144,8 @@ void Buffer::Flush(uint64_t offset, uint64_t size) {
 vk::BufferMemoryBarrier Buffer::Barrier(uint64_t offset, uint64_t size, vk::AccessFlags source,
                                         vk::AccessFlags destination) const {
 	if (Handle() == nullptr || size == 0 || offset > m_size || size > m_size - offset) {
-		EXIT("Buffer: invalid DMA barrier, handle=%p offset=0x%016" PRIx64
-		     " size=0x%016" PRIx64 " capacity=0x%016" PRIx64 "\n",
+		EXIT("Buffer: invalid DMA barrier, handle=%p offset=0x%016" PRIx64 " size=0x%016" PRIx64
+		     " capacity=0x%016" PRIx64 "\n",
 		     static_cast<const void*>(Handle()), offset, size, m_size);
 	}
 	vk::BufferMemoryBarrier barrier {};
@@ -175,10 +175,9 @@ void Buffer::CopyFrom(CommandBuffer& command, const Buffer& source, uint64_t sou
 	command.EndRendering();
 	const vk::BufferMemoryBarrier before[] = {
 	    source.Barrier(source_offset, size, source_before, vk::AccessFlagBits::eTransferRead),
-	    Barrier(destination_offset, size, destination_before,
-	            vk::AccessFlagBits::eTransferWrite),
+	    Barrier(destination_offset, size, destination_before, vk::AccessFlagBits::eTransferWrite),
 	};
-	const auto host_access = vk::AccessFlagBits::eHostRead | vk::AccessFlagBits::eHostWrite;
+	const auto host_access  = vk::AccessFlagBits::eHostRead | vk::AccessFlagBits::eHostWrite;
 	auto       before_stage = vk::PipelineStageFlags {vk::PipelineStageFlagBits::eAllCommands};
 	if (static_cast<bool>((source_before | destination_before) & host_access)) {
 		before_stage |= vk::PipelineStageFlagBits::eHost;
@@ -214,9 +213,8 @@ void Buffer::Fill(uint64_t offset, uint64_t size, uint32_t value) {
 	                       vk::PipelineStageFlagBits::eTransfer, vk::DependencyFlagBits::eByRegion,
 	                       0, nullptr, 1, &before, 0, nullptr);
 	native.fillBuffer(Handle(), offset, size, value);
-	const auto after =
-	    Barrier(offset, size, vk::AccessFlagBits::eTransferWrite,
-	            vk::AccessFlagBits::eMemoryRead | vk::AccessFlagBits::eMemoryWrite);
+	const auto after = Barrier(offset, size, vk::AccessFlagBits::eTransferWrite,
+	                           vk::AccessFlagBits::eMemoryRead | vk::AccessFlagBits::eMemoryWrite);
 	native.pipelineBarrier(vk::PipelineStageFlagBits::eTransfer,
 	                       vk::PipelineStageFlagBits::eAllCommands,
 	                       vk::DependencyFlagBits::eByRegion, 0, nullptr, 1, &after, 0, nullptr);
@@ -250,8 +248,8 @@ std::pair<uint8_t*, uint64_t> StreamBuffer::Map(uint64_t size, uint64_t alignmen
 	if (Mapped().empty()) {
 		return {nullptr, 0};
 	}
-	uint64_t mapped_size = size;
-	const auto atom = Graphics().physical_device_properties.limits.nonCoherentAtomSize;
+	uint64_t   mapped_size = size;
+	const auto atom        = Graphics().physical_device_properties.limits.nonCoherentAtomSize;
 	if (!NormalizeReservation(IsCoherent(), atom, mapped_size, alignment)) {
 		return {nullptr, 0};
 	}

src/graphics/host_gpu/renderer/cache/streamBuffer.h

@@ -54,16 +54,15 @@ public:
 	[[nodiscard]] bool IsInBounds(uint64_t address, uint64_t size) const noexcept;
 	void               Write(uint64_t offset, const void* source, uint64_t size);
 	void               Flush(uint64_t offset, uint64_t size);
-	void               CopyFrom(
-	                  CommandBuffer& command, const Buffer& source, uint64_t source_offset,
-	                  uint64_t destination_offset, uint64_t size,
-	                  vk::AccessFlags source_before = vk::AccessFlagBits::eMemoryWrite,
-	                  vk::AccessFlags destination_before =
-	                      vk::AccessFlagBits::eMemoryRead | vk::AccessFlagBits::eMemoryWrite,
-	                  vk::AccessFlags source_after =
-	                      vk::AccessFlagBits::eMemoryRead | vk::AccessFlagBits::eMemoryWrite,
-	                  vk::AccessFlags destination_after =
-	                      vk::AccessFlagBits::eMemoryRead | vk::AccessFlagBits::eMemoryWrite);
+	void CopyFrom(CommandBuffer& command, const Buffer& source, uint64_t source_offset,
+	              uint64_t destination_offset, uint64_t size,
+	              vk::AccessFlags source_before      = vk::AccessFlagBits::eMemoryWrite,
+	              vk::AccessFlags destination_before = vk::AccessFlagBits::eMemoryRead |
+	                                                   vk::AccessFlagBits::eMemoryWrite,
+	              vk::AccessFlags source_after       = vk::AccessFlagBits::eMemoryRead |
+	                                                   vk::AccessFlagBits::eMemoryWrite,
+	              vk::AccessFlags destination_after  = vk::AccessFlagBits::eMemoryRead |
+	                                                   vk::AccessFlagBits::eMemoryWrite);
 	void Fill(uint64_t offset, uint64_t size, uint32_t value);
 
 protected:
@@ -107,13 +106,13 @@ private:
 		uint64_t upper_bound = 0;
 	};
 
-	void               ReserveWatches(std::vector<Watch>& watches, size_t grow_size);
+	void                      ReserveWatches(std::vector<Watch>& watches, size_t grow_size);
 	[[nodiscard]] static bool NormalizeReservation(bool coherent, uint64_t atom, uint64_t& size,
 	                                               uint64_t& alignment);
-	[[nodiscard]] bool WaitPendingOperations(const std::vector<Watch>& watches,
-	                                         std::optional<size_t>     invalidation_mark,
-	                                         uint64_t requested_upper_bound, bool allow_wait,
-	                                         size_t& wait_cursor, uint64_t& wait_bound);
+	[[nodiscard]] bool        WaitPendingOperations(const std::vector<Watch>& watches,
+	                                                std::optional<size_t>     invalidation_mark,
+	                                                uint64_t requested_upper_bound, bool allow_wait,
+	                                                size_t& wait_cursor, uint64_t& wait_bound);
 
 	uint64_t              m_offset      = 0;
 	uint64_t              m_mapped_size = 0;

src/graphics/host_gpu/renderer/colorRenderTarget.cpp

@@ -7,8 +7,8 @@
 #include "graphics/guest_gpu/hardwareContext.h"
 #include "graphics/guest_gpu/tile.h"
 #include "graphics/host_gpu/graphicContext.h"
-#include "graphics/host_gpu/renderer/image/textureCommon.h"
 #include "graphics/host_gpu/renderer/debug.h"
+#include "graphics/host_gpu/renderer/image/textureCommon.h"
 #include "graphics/host_gpu/renderer/pipeline/descriptorCache.h"
 #include "graphics/host_gpu/renderer/render.h"
 #include "graphics/host_gpu/renderer/renderContext.h"
@@ -23,10 +23,10 @@ static std::atomic<uint32_t> g_render_color_log_count = 0;
 
 // NOLINTNEXTLINE(readability-function-cognitive-complexity)
 void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandBuffer& buffer,
-                                          RenderColorInfo& r,
-                                          uint32_t render_target_slice_offset,
-                                          uint32_t render_target_slot, bool ignore_target_mask,
-                                          bool exact_format) {
+                                              RenderColorInfo& r,
+                                              uint32_t         render_target_slice_offset,
+                                              uint32_t render_target_slot, bool ignore_target_mask,
+                                              bool exact_format) {
 	KYTY_PROFILER_FUNCTION();
 	const auto& hw = buffer.GetRegisters();
 
@@ -79,10 +79,8 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 	const auto view = ResolveTargetViewInfo(
 	    rt.view.base_array_slice_index, rt.view.last_array_slice_index, render_target_slice_offset);
 	switch (view.type) {
-		case TargetViewType::Image2D: break;
-		case TargetViewType::Image2DArray:
-			EXIT("layered render-target views are unsupported: base=%u count=%u\n", view.base_layer,
-			     view.layer_count);
+		case TargetViewType::Image2D:
+		case TargetViewType::Image2DArray: break;
 		case TargetViewType::Unsupported:
 			EXIT("invalid render-target view: base=%u last=%u draw_offset=%u\n",
 			     rt.view.base_array_slice_index, rt.view.last_array_slice_index,
@@ -241,12 +239,12 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 	}
 
 	TextureCache::ImageDesc desc {};
-	desc.type              = TextureCache::BindingType::RenderTarget;
-	desc.info.data         = {rt.base.addr, backing_size};
-	desc.info.pixel_format = target_format.format;
-	desc.info.guest_format = ImageOps::RenderTargetTransferFormat(bytes_per_element);
-	desc.info.type   = Prospero::ImageType::kColor2D;
-	desc.info.extent = {width, height, 1};
+	desc.type                 = TextureCache::BindingType::RenderTarget;
+	desc.info.data            = {rt.base.addr, backing_size};
+	desc.info.pixel_format    = target_format.format;
+	desc.info.guest_format    = ImageOps::RenderTargetTransferFormat(bytes_per_element);
+	desc.info.type            = Prospero::ImageType::kColor2D;
+	desc.info.extent          = {width, height, 1};
 	desc.info.resources       = {levels, view.image_layers};
 	desc.info.pitch           = pitch;
 	desc.info.bytes_per_block = bytes_per_element;
@@ -275,20 +273,20 @@ void RenderExecutor::ResolveRenderColorTarget(uint64_t submit_id, RenderCommandB
 	desc.view_info.base_layer  = view.base_layer;
 	desc.view_info.layer_count = view.layer_count;
 	desc.view_info.usage       = vk::ImageUsageFlagBits::eColorAttachment;
-	auto& texture_cache = m_context.GetTextureCache();
-	r.desc              = std::move(desc);
-	r.image_id          = texture_cache.FindImage(r.desc, exact_format);
-	r.type              = RenderColorType::RenderTexture;
-	r.base_addr         = rt.base.addr;
-	r.image_view        = nullptr;
-	r.format            = r.desc.view_info.format;
-	r.extent            = view_extent;
-	r.base_mip_level    = rt.view.current_mip_level;
-	r.buffer_size       = backing_size;
-	r.samples           = samples;
-	r.export_mapping    = target_format.export_mapping;
-	r.color_clear_enable = false;
-	r.color_clear_value = {};
+	auto& texture_cache        = m_context.GetTextureCache();
+	r.desc                     = std::move(desc);
+	r.image_id                 = texture_cache.FindImage(r.desc, exact_format);
+	r.type                     = RenderColorType::RenderTexture;
+	r.base_addr                = rt.base.addr;
+	r.image_view               = nullptr;
+	r.format                   = r.desc.view_info.format;
+	r.extent                   = view_extent;
+	r.base_mip_level           = rt.view.current_mip_level;
+	r.buffer_size              = backing_size;
+	r.samples                  = samples;
+	r.export_mapping           = target_format.export_mapping;
+	r.color_clear_enable       = false;
+	r.color_clear_value        = {};
 	BindRenderTarget(r.image_id);
 }
 

src/graphics/host_gpu/renderer/colorRenderTarget.h

@@ -2,8 +2,8 @@
 #define EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_COLORRENDERTARGET_H_
 
 #include "graphics/guest_gpu/gpu_defs.h"
-#include "graphics/host_gpu/renderer/renderTarget.h"
 #include "graphics/host_gpu/renderer/cache/textureCache.h"
+#include "graphics/host_gpu/renderer/renderTarget.h"
 #include "graphics/host_gpu/vulkanCommon.h"
 
 #include <cstdint>

src/graphics/host_gpu/renderer/commandScheduler.cpp

@@ -44,13 +44,13 @@ CommandSlot* CommandScheduler::CommandPool::CreateSlot() {
 	allocate.commandPool        = m_pool;
 	allocate.level              = vk::CommandBufferLevel::ePrimary;
 	allocate.commandBufferCount = 1;
-	vk::CommandBuffer buffer = nullptr;
+	vk::CommandBuffer buffer    = nullptr;
 	EXIT_IF(graphics.device.allocateCommandBuffers(&allocate, &buffer) != vk::Result::eSuccess);
 
 	vk::FenceCreateInfo fence_create {};
 	fence_create.sType = vk::StructureType::eFenceCreateInfo;
 	fence_create.flags = vk::FenceCreateFlagBits::eSignaled;
-	vk::Fence fence = nullptr;
+	vk::Fence fence    = nullptr;
 	if (graphics.device.createFence(&fence_create, nullptr, &fence) != vk::Result::eSuccess) {
 		graphics.device.freeCommandBuffers(m_pool, 1, &buffer);
 		EXIT("failed to create command-buffer fence\n");
@@ -70,9 +70,9 @@ CommandSlot* CommandScheduler::CommandPool::Allocate(GraphicContext& graphics) {
 		Create(graphics);
 	}
 	EXIT_IF(m_graphics != &graphics);
-	auto found = std::ranges::find_if(m_slots, [](const auto& slot) { return !slot.busy; });
-	auto* slot = found != m_slots.end() ? &*found : CreateSlot();
-	slot->busy = true;
+	auto  found = std::ranges::find_if(m_slots, [](const auto& slot) { return !slot.busy; });
+	auto* slot  = found != m_slots.end() ? &*found : CreateSlot();
+	slot->busy  = true;
 	slot->Reset();
 	return slot;
 }
@@ -331,8 +331,7 @@ void CommandScheduler::WaitPriorityOperations(uint64_t tick) {
 	EXIT_IF(g_deferred_callback_scheduler == this);
 	std::unique_lock lock(m_operation_mutex);
 	m_operation_available.wait(lock, [this, tick] {
-		const bool active_before_or_at =
-		    m_priority_active && m_priority_active_tick <= tick;
+		const bool active_before_or_at = m_priority_active && m_priority_active_tick <= tick;
 		const bool queued_before_or_at =
 		    !m_priority_operations.empty() && m_priority_operations.front().tick <= tick;
 		return !active_before_or_at && !queued_before_or_at;

src/graphics/host_gpu/renderer/commandScheduler.h

@@ -47,21 +47,21 @@ public:
 	void           FinishCurrent();
 	// Deferred callbacks can observe an externally owned drain, but cannot initiate shutdown:
 	// the priority runner cannot join itself.
-	void Shutdown();
-	void Wait(uint64_t tick);
-	void PopPendingOperations();
-	void DrainPriorityOperations();
-	void WaitPriorityOperations(uint64_t tick);
-	void DeferOperation(Common::UniqueFunction<void>&& operation);
-	void DeferPriorityOperation(Common::UniqueFunction<void>&& operation);
+	void                      Shutdown();
+	void                      Wait(uint64_t tick);
+	void                      PopPendingOperations();
+	void                      DrainPriorityOperations();
+	void                      WaitPriorityOperations(uint64_t tick);
+	void                      DeferOperation(Common::UniqueFunction<void>&& operation);
+	void                      DeferPriorityOperation(Common::UniqueFunction<void>&& operation);
 	[[nodiscard]] static bool InDeferredOperation() noexcept;
 
-	[[nodiscard]] bool     Active() const noexcept { return m_current >= 0; }
-	void                   CheckActive() const;
-	RenderCommandBuffer&   Current() const;
-	[[nodiscard]] uint64_t CurrentTick() const noexcept { return m_master.CurrentTick(); }
-	[[nodiscard]] bool     IsFree(uint64_t tick);
-	[[nodiscard]] RenderContext& Context() const noexcept { return m_context; }
+	[[nodiscard]] bool            Active() const noexcept { return m_current >= 0; }
+	void                          CheckActive() const;
+	RenderCommandBuffer&          Current() const;
+	[[nodiscard]] uint64_t        CurrentTick() const noexcept { return m_master.CurrentTick(); }
+	[[nodiscard]] bool            IsFree(uint64_t tick);
+	[[nodiscard]] RenderContext&  Context() const noexcept { return m_context; }
 	[[nodiscard]] GraphicContext& Graphics() const noexcept { return m_graphics; }
 
 private:
@@ -91,11 +91,11 @@ private:
 		uint64_t                     tick = 0;
 	};
 
-	void           BindCurrent() const;
-	CommandBuffer& SubmitCurrent(SubmitInfo& submit);
-	void           BeginNext();
-	void           PriorityOperationsThread(std::stop_token stop);
-	void           RunOperation(Common::UniqueFunction<void>&& operation);
+	void                       BindCurrent() const;
+	CommandBuffer&             SubmitCurrent(SubmitInfo& submit);
+	void                       BeginNext();
+	void                       PriorityOperationsThread(std::stop_token stop);
+	void                       RunOperation(Common::UniqueFunction<void>&& operation);
 	[[nodiscard]] CommandSlot* AllocateCommandBuffer();
 	[[nodiscard]] uint64_t     NextSubmitSequence() noexcept;
 
@@ -109,14 +109,14 @@ private:
 	std::mutex                                                    m_operation_mutex;
 	std::condition_variable                                       m_operation_available;
 	std::jthread                                                  m_priority_thread;
-	bool                                                          m_priority_active = false;
+	bool                                                          m_priority_active      = false;
 	uint64_t                                                      m_priority_active_tick = 0;
-	OperationState  m_operation_state = OperationState::Open;
-	int             m_current         = -1;
-	bool            m_recording       = false;
-	HW::Context*    m_registers       = nullptr;
-	HW::UserConfig* m_user_config     = nullptr;
-	HW::Shader*     m_shaders         = nullptr;
+	OperationState        m_operation_state = OperationState::Open;
+	int                   m_current         = -1;
+	bool                  m_recording       = false;
+	HW::Context*          m_registers       = nullptr;
+	HW::UserConfig*       m_user_config     = nullptr;
+	HW::Shader*           m_shaders         = nullptr;
 	std::atomic<uint64_t> m_submit_sequence = 0;
 
 	friend class CommandBuffer;

src/graphics/host_gpu/renderer/context.cpp

@@ -8,8 +8,8 @@
 #include "graphics/host_gpu/renderer/colorRenderTarget.h"
 #include "graphics/host_gpu/renderer/debug.h"
 #include "graphics/host_gpu/renderer/depthRenderTarget.h"
-#include "graphics/host_gpu/renderer/pipeline/descriptorCache.h"
 #include "graphics/host_gpu/renderer/image/imageView.h"
+#include "graphics/host_gpu/renderer/pipeline/descriptorCache.h"
 #include "graphics/host_gpu/renderer/render.h"
 #include "graphics/host_gpu/renderer/renderContext.h"
 #include "graphics/host_gpu/vma.h"
@@ -270,30 +270,30 @@ void CommandBuffer::BeginRendering(const RenderState& state) const {
 		colors[i].sType        = vk::StructureType::eRenderingAttachmentInfo;
 		colors[i].imageView    = attachment.image_view;
 		colors[i].imageLayout  = attachment.image_layout;
-		colors[i].loadOp = attachment.is_clear ? vk::AttachmentLoadOp::eClear
-		                                      : vk::AttachmentLoadOp::eLoad;
-		colors[i].storeOp                  = vk::AttachmentStoreOp::eStore;
-		colors[i].clearValue.color.uint32  = attachment.clear_value;
+		colors[i].loadOp =
+		    attachment.is_clear ? vk::AttachmentLoadOp::eClear : vk::AttachmentLoadOp::eLoad;
+		colors[i].storeOp                 = vk::AttachmentStoreOp::eStore;
+		colors[i].clearValue.color.uint32 = attachment.clear_value;
 	}
 
-	const auto& depth_stencil = state.depth_stencil_attachment;
+	const auto&                 depth_stencil = state.depth_stencil_attachment;
 	vk::RenderingAttachmentInfo depth {};
 	depth.sType       = vk::StructureType::eRenderingAttachmentInfo;
 	depth.imageView   = depth_stencil.image_view;
 	depth.imageLayout = depth_stencil.image_layout;
-	depth.loadOp = depth_stencil.depth_clear ? vk::AttachmentLoadOp::eClear
-	                                        : vk::AttachmentLoadOp::eLoad;
-	depth.storeOp                        = vk::AttachmentStoreOp::eStore;
+	depth.loadOp =
+	    depth_stencil.depth_clear ? vk::AttachmentLoadOp::eClear : vk::AttachmentLoadOp::eLoad;
+	depth.storeOp                       = vk::AttachmentStoreOp::eStore;
 	depth.clearValue.depthStencil.depth = std::bit_cast<float>(depth_stencil.clear_value[0]);
 
 	vk::RenderingAttachmentInfo stencil {};
 	stencil.sType       = vk::StructureType::eRenderingAttachmentInfo;
 	stencil.imageView   = depth_stencil.image_view;
 	stencil.imageLayout = depth_stencil.image_layout;
-	stencil.loadOp = depth_stencil.stencil_clear ? vk::AttachmentLoadOp::eClear
-	                                            : vk::AttachmentLoadOp::eLoad;
-	stencil.storeOp                            = vk::AttachmentStoreOp::eStore;
-	stencil.clearValue.depthStencil.stencil   = depth_stencil.clear_value[1];
+	stencil.loadOp =
+	    depth_stencil.stencil_clear ? vk::AttachmentLoadOp::eClear : vk::AttachmentLoadOp::eLoad;
+	stencil.storeOp                         = vk::AttachmentStoreOp::eStore;
+	stencil.clearValue.depthStencil.stencil = depth_stencil.clear_value[1];
 
 	vk::RenderingInfo rendering {};
 	rendering.sType                = vk::StructureType::eRenderingInfo;

src/graphics/host_gpu/renderer/debug.cpp

@@ -548,14 +548,6 @@ static void ZCheck(const HW::DepthRenderTarget& z) {
 		EXIT_NOT_IMPLEMENTED(z.htile_surface.prefetch_height != 0x00000000);
 		EXIT_NOT_IMPLEMENTED(z.htile_surface.dst_outside_zero_to_one != 0x00000000);
 
-		if (z.depth_view.slice_start != 0x00000000 || z.depth_view.slice_max != 0x00000000) {
-			static std::atomic<uint32_t> log_count {0};
-			if (log_count.fetch_add(1, std::memory_order_relaxed) < 16) {
-				LOGF("DepthTarget: temporary: ignoring PS5 array slice view start=0x%08" PRIx32
-				     ", max=0x%08" PRIx32 "\n",
-				     z.depth_view.slice_start, z.depth_view.slice_max);
-			}
-		}
 		if (z.depth_view.current_mip_level != 0x00000000) {
 			static std::atomic<uint32_t> log_count {0};
 			if (log_count.fetch_add(1, std::memory_order_relaxed) < 16) {

src/graphics/host_gpu/renderer/depthRenderTarget.cpp

@@ -10,10 +10,10 @@
 #include "graphics/guest_gpu/hardwareContext.h"
 #include "graphics/guest_gpu/tile.h"
 #include "graphics/host_gpu/graphicContext.h"
-#include "graphics/host_gpu/renderer/image/textureCommon.h"
 #include "graphics/host_gpu/renderer/debug.h"
-#include "graphics/host_gpu/renderer/pipeline/descriptorCache.h"
 #include "graphics/host_gpu/renderer/image/imageView.h"
+#include "graphics/host_gpu/renderer/image/textureCommon.h"
+#include "graphics/host_gpu/renderer/pipeline/descriptorCache.h"
 #include "graphics/host_gpu/renderer/render.h"
 #include "graphics/host_gpu/renderer/renderContext.h"
 #include "graphics/host_gpu/vulkanCommon.h"
@@ -150,10 +150,8 @@ void RenderExecutor::ResolveRenderDepthTarget(uint64_t submit_id, RenderCommandB
 	    has_stencil, has_htile, z.stencil_info.htile_stencil_disabled);
 	const auto view = ResolveTargetViewInfo(z.depth_view.slice_start, z.depth_view.slice_max);
 	switch (view.type) {
-		case TargetViewType::Image2D: break;
-		case TargetViewType::Image2DArray:
-			DepthFatal("layered depth views are unsupported: base=%u count=%u", view.base_layer,
-			           view.layer_count);
+		case TargetViewType::Image2D:
+		case TargetViewType::Image2DArray: break;
 		case TargetViewType::Unsupported:
 			DepthFatal("invalid depth view: base=%u last=%u", z.depth_view.slice_start,
 			           z.depth_view.slice_max);

src/graphics/host_gpu/renderer/depthRenderTarget.h

@@ -2,9 +2,9 @@
 #define EMULATOR_SRC_GRAPHICS_HOST_GPU_RENDERER_DEPTHRENDERTARGET_H_
 
 #include "common/assert.h"
+#include "graphics/host_gpu/renderer/cache/textureCache.h"
 #include "graphics/host_gpu/renderer/image/imageView.h"
 #include "graphics/host_gpu/renderer/renderTarget.h"
-#include "graphics/host_gpu/renderer/cache/textureCache.h"
 #include "graphics/host_gpu/vulkanCommon.h"
 
 #include <cstdint>

src/graphics/host_gpu/renderer/image/blitHelper.cpp

@@ -182,8 +182,8 @@ void BlitHelper::ReinterpretColorAsMsDepth(Image& source, Image& destination) {
 	auto  command        = command_buffer.Handle();
 	source.Transit(vk::ImageLayout::eShaderReadOnlyOptimal, vk::AccessFlagBits2::eShaderRead, {},
 	               command);
-	destination.Transit(ColorToMsDepthLayout,
-	                    vk::AccessFlagBits2::eDepthStencilAttachmentWrite, {}, command);
+	destination.Transit(ColorToMsDepthLayout, vk::AccessFlagBits2::eDepthStencilAttachmentWrite, {},
+	                    command);
 
 	vk::RenderingAttachmentInfo depth_attachment {};
 	depth_attachment.sType                   = vk::StructureType::eRenderingAttachmentInfo;

src/graphics/host_gpu/renderer/image/imageInfo.h

@@ -19,10 +19,9 @@ struct GuestRange {
 	uint64_t address = 0;
 	uint64_t size    = 0;
 
-	[[nodiscard]] constexpr bool     Empty() const noexcept { return address == 0 || size == 0; }
-	[[nodiscard]] constexpr bool     Valid() const noexcept {
-		return !Empty() && address < TRACKER_ADDRESS_SIZE &&
-		       size <= TRACKER_ADDRESS_SIZE - address;
+	[[nodiscard]] constexpr bool Empty() const noexcept { return address == 0 || size == 0; }
+	[[nodiscard]] constexpr bool Valid() const noexcept {
+		return !Empty() && address < TRACKER_ADDRESS_SIZE && size <= TRACKER_ADDRESS_SIZE - address;
 	}
 	[[nodiscard]] constexpr uint64_t End() const noexcept { return address + size; }
 	auto                             operator<=>(const GuestRange&) const = default;
@@ -47,10 +46,10 @@ struct ImageSubresources {
 };
 
 struct ImageSubresourceRange {
-	uint32_t base_level  = 0;
-	uint32_t level_count = 1;
-	uint32_t base_layer  = 0;
-	uint32_t layer_count = 1;
+	uint32_t base_level                                      = 0;
+	uint32_t level_count                                     = 1;
+	uint32_t base_layer                                      = 0;
+	uint32_t layer_count                                     = 1;
 	auto     operator<=>(const ImageSubresourceRange&) const = default;
 };
 
@@ -67,10 +66,10 @@ struct ImageInfo {
 	GuestRange                   stencil;
 	ImageMetadataInfo            metadata;
 	uint32_t                     htile_clear_mask = UINT32_MAX;
-	vk::Format                   pixel_format = vk::Format::eUndefined;
-	uint32_t                     guest_format = 0;
-	Prospero::ImageType          type         = Prospero::ImageType::kColor2D;
-	vk::Extent3D                 extent       = {1, 1, 1};
+	vk::Format                   pixel_format     = vk::Format::eUndefined;
+	uint32_t                     guest_format     = 0;
+	Prospero::ImageType          type             = Prospero::ImageType::kColor2D;
+	vk::Extent3D                 extent           = {1, 1, 1};
 	ImageSubresources            resources;
 	uint32_t                     pitch           = 0;
 	uint32_t                     bytes_per_block = 0;
@@ -352,8 +351,7 @@ inline bool ImageInfo::IsDepth() const noexcept {
 	}
 	const auto transfer_bytes = DepthAspectTransferBytes(info.pixel_format);
 	return transfer_bytes == info.bytes_per_block ||
-	       (info.bytes_per_block == sizeof(uint16_t) &&
-	        transfer_bytes == sizeof(uint32_t));
+	       (info.bytes_per_block == sizeof(uint16_t) && transfer_bytes == sizeof(uint32_t));
 }
 
 [[nodiscard]] inline VideoOutCompression
@@ -470,18 +468,13 @@ IsSupportedDisplayRenderTargetTileMode(uint32_t tile_mode) noexcept {
                                                  vk::ClearColorValue& clear) {
 	vk::ClearColorValue next {};
 	const auto unorm8 = [](uint32_t value) { return static_cast<float>(value & 0xffu) / 255.0f; };
-	const auto srgb8 = [](uint32_t value) {
+	const auto srgb8  = [](uint32_t value) {
 		const auto encoded = static_cast<float>(value & 0xffu) / 255.0f;
-		return encoded <= 0.04045f ? encoded / 12.92f
-		                           : std::pow((encoded + 0.055f) / 1.055f, 2.4f);
+		return encoded <= 0.04045f ? encoded / 12.92f : std::pow((encoded + 0.055f) / 1.055f, 2.4f);
 	};
 	switch (format) {
-		case vk::Format::eR32Uint:
-			next.uint32[0] = packed;
-			break;
-		case vk::Format::eR32Sint:
-			next.int32[0] = static_cast<int32_t>(packed);
-			break;
+		case vk::Format::eR32Uint: next.uint32[0] = packed; break;
+		case vk::Format::eR32Sint: next.int32[0] = static_cast<int32_t>(packed); break;
 		case vk::Format::eR8G8B8A8Srgb:
 			next.float32[0] = srgb8(packed);
 			next.float32[1] = srgb8(packed >> 8u);

src/graphics/host_gpu/renderer/image/imageView.cpp

@@ -70,15 +70,14 @@ namespace {
 			}
 		case vk::ImageType::e3D:
 			switch (info.type) {
-				case vk::ImageViewType::e3D:
-					return info.base_layer == 0 && info.layer_count == 1;
+				case vk::ImageViewType::e3D: return info.base_layer == 0 && info.layer_count == 1;
 				case vk::ImageViewType::e2D:
-					return static_cast<bool>(
-					           image.flags & vk::ImageCreateFlagBits::e2DArrayCompatible) &&
+					return static_cast<bool>(image.flags &
+					                         vk::ImageCreateFlagBits::e2DArrayCompatible) &&
 					       info.level_count == 1 && info.layer_count == 1;
 				case vk::ImageViewType::e2DArray:
-					return static_cast<bool>(
-					           image.flags & vk::ImageCreateFlagBits::e2DArrayCompatible) &&
+					return static_cast<bool>(image.flags &
+					                         vk::ImageCreateFlagBits::e2DArrayCompatible) &&
 					       info.level_count == 1;
 				default: return false;
 			}
@@ -325,11 +324,10 @@ bool FormatsCompatible(vk::Format base, vk::Format view) noexcept {
 } // namespace ImageViewOps
 
 vk::ImageView Image::FindView(const ImageViewInfo& view_info) {
-	const auto& image = backing;
+	const auto& image      = backing;
 	auto        normalized = view_info;
-	const bool  is_storage =
-	    static_cast<bool>(normalized.usage & vk::ImageUsageFlagBits::eStorage);
-	normalized.aspect = FullAspectMask(image.format);
+	const bool  is_storage = static_cast<bool>(normalized.usage & vk::ImageUsageFlagBits::eStorage);
+	normalized.aspect      = FullAspectMask(image.format);
 	if (normalized.aspect & vk::ImageAspectFlagBits::eDepth &&
 	    IsDepthViewFormat(normalized.format)) {
 		normalized.format = image.format;
@@ -340,28 +338,26 @@ vk::ImageView Image::FindView(const ImageViewInfo& view_info) {
 		normalized.format = image.format;
 		normalized.aspect = vk::ImageAspectFlagBits::eStencil;
 	}
-	normalized.usage =
-	    is_storage ? vk::ImageUsageFlagBits::eStorage : vk::ImageUsageFlags {};
+	normalized.usage = is_storage ? vk::ImageUsageFlagBits::eStorage : vk::ImageUsageFlags {};
 	const bool format_compatible = normalized.format != vk::Format::eUndefined &&
 	                               IsCompatibleViewFormat(image.format, normalized.format);
-	const bool slice_view = image.image_type == vk::ImageType::e3D &&
-	                        (normalized.type == vk::ImageViewType::e2D ||
-	                         normalized.type == vk::ImageViewType::e2DArray);
+	const bool slice_view =
+	    image.image_type == vk::ImageType::e3D && (normalized.type == vk::ImageViewType::e2D ||
+	                                               normalized.type == vk::ImageViewType::e2DArray);
 	const bool levels_valid = normalized.level_count != 0 &&
 	                          normalized.base_level < image.mip_levels &&
 	                          normalized.level_count <= image.mip_levels - normalized.base_level;
-	const auto view_layers = slice_view && levels_valid
-	                             ? std::max(image.extent.depth >> normalized.base_level, 1u)
-	                             : image.layers;
-	const bool ranges_valid = levels_valid &&
-	                          normalized.layer_count != 0 && normalized.base_layer < view_layers &&
+	const auto view_layers  = slice_view && levels_valid
+	                              ? std::max(image.extent.depth >> normalized.base_level, 1u)
+	                              : image.layers;
+	const bool ranges_valid = levels_valid && normalized.layer_count != 0 &&
+	                          normalized.base_layer < view_layers &&
 	                          normalized.layer_count <= view_layers - normalized.base_layer;
 	const bool mapping_valid =
 	    IsComponentSwizzle(normalized.mapping.r) && IsComponentSwizzle(normalized.mapping.g) &&
 	    IsComponentSwizzle(normalized.mapping.b) && IsComponentSwizzle(normalized.mapping.a);
 	if (image.image == nullptr || !format_compatible || !ranges_valid || !mapping_valid ||
-	    !IsValidViewType(image, normalized) ||
-	    !IsValidAspect(image, normalized.aspect)) {
+	    !IsValidViewType(image, normalized) || !IsValidAspect(image, normalized.aspect)) {
 		EXIT("invalid image view: image_format=%d view_format=%d type=%d aspect=0x%x "
 		     "mip=%u+%u layer=%u+%u usage=0x%x image_levels=%u image_layers=%u\n",
 		     static_cast<int>(image.format), static_cast<int>(normalized.format),

src/graphics/host_gpu/renderer/image/imageView.h

@@ -88,7 +88,9 @@ SelectSampledDepthView(vk::Format image_format, vk::Format view_format, uint32_t
 IsSupportedSampledDepthResource(const ShaderRecompiler::IR::ImageResource& resource) noexcept {
 	return resource.kind == ShaderRecompiler::IR::ResourceKind::Image &&
 	       (resource.dimension == ShaderRecompiler::Decoder::ImageDimension::Dim2D ||
-	        resource.dimension == ShaderRecompiler::Decoder::ImageDimension::Dim2DArray) &&
+	        resource.dimension == ShaderRecompiler::Decoder::ImageDimension::Dim2DArray ||
+	        resource.dimension == ShaderRecompiler::Decoder::ImageDimension::Dim2DMsaa ||
+	        resource.dimension == ShaderRecompiler::Decoder::ImageDimension::Dim2DMsaaArray) &&
 	       resource.mip_mode == ShaderRecompiler::IR::ImageMipMode::None && resource.read &&
 	       !resource.written && !resource.atomic;
 }

src/graphics/host_gpu/renderer/image/textureCommon.cpp

@@ -397,10 +397,10 @@ TextureUploadLayout TextureCalcUploadLayout(uint32_t fmt, uint64_t width, uint64
 	return layout;
 }
 
-std::vector<vk::BufferImageCopy>
-TextureBuildImageCopies(const TextureUploadLayout& layout, uint32_t width, uint32_t height,
-                        uint32_t depth, uint64_t levels, bool array_texture,
-                        bool volume_texture) {
+std::vector<vk::BufferImageCopy> TextureBuildImageCopies(const TextureUploadLayout& layout,
+                                                         uint32_t width, uint32_t height,
+                                                         uint32_t depth, uint64_t levels,
+                                                         bool array_texture, bool volume_texture) {
 	uint32_t mip_width  = width;
 	uint32_t mip_height = height;
 	uint32_t mip_pitch  = volume_texture && static_cast<Prospero::TileMode>(layout.tile) !=
@@ -416,14 +416,13 @@ TextureBuildImageCopies(const TextureUploadLayout& layout, uint32_t width, uint3
 		const auto mip_depth = GetTextureLevelDepth(depth, i, volume_texture);
 
 		for (uint32_t z = 0; z < mip_depth; z++) {
-			const auto slice_offset = z * layout.slice_stride;
+			const auto          slice_offset = z * layout.slice_stride;
 			vk::BufferImageCopy region {};
-			region.bufferOffset =
-			    layout.level_sizes[i].offset + slice_offset;
-			region.imageSubresource = {vk::ImageAspectFlagBits::eColor, i,
-			                           array_texture ? z : 0, 1};
-			region.imageOffset.z = volume_texture ? static_cast<int>(z) : 0;
-			region.imageExtent   = {mip_width, mip_height, 1};
+			region.bufferOffset     = layout.level_sizes[i].offset + slice_offset;
+			region.imageSubresource = {vk::ImageAspectFlagBits::eColor, i, array_texture ? z : 0,
+			                           1};
+			region.imageOffset.z    = volume_texture ? static_cast<int>(z) : 0;
+			region.imageExtent      = {mip_width, mip_height, 1};
 			const bool linear =
 			    static_cast<Prospero::TileMode>(layout.tile) == Prospero::TileMode::kLinear;
 			if (linear) {
@@ -433,9 +432,8 @@ TextureBuildImageCopies(const TextureUploadLayout& layout, uint32_t width, uint3
 				const auto align = [](uint32_t value, uint32_t block) {
 					return ((value + block - 1u) / block) * block;
 				};
-				const auto pitch = align(mip_pitch, layout.texel_block);
-				region.bufferRowLength =
-				    pitch > align(mip_width, layout.texel_block) ? pitch : 0;
+				const auto pitch       = align(mip_pitch, layout.texel_block);
+				region.bufferRowLength = pitch > align(mip_width, layout.texel_block) ? pitch : 0;
 			}
 			regions.push_back(region);
 		}
@@ -480,8 +478,7 @@ static bool SetGpuTileSize(uint64_t offset, uint64_t length, uint64_t capacity,
 	return true;
 }
 
-bool TextureBuildGpuTileInfos(uint64_t size,
-                              const std::vector<vk::BufferImageCopy>& regions,
+bool TextureBuildGpuTileInfos(uint64_t size, const std::vector<vk::BufferImageCopy>& regions,
                               const TextureUploadLayout& layout, uint32_t fmt, uint32_t depth,
                               uint64_t levels, std::vector<GpuTileInfo>& out_infos) {
 	if (size == 0 || levels == 0 || levels > 16 || depth == 0 ||
@@ -522,13 +519,12 @@ bool TextureBuildGpuTileInfos(uint64_t size,
 			for (uint32_t z = 0; z < mip_depth; z += block.block_depth) {
 				const uint32_t copy_depth = std::min(block.block_depth, mip_depth - z);
 				const auto&    region     = regions[region_base + z];
-				const auto     pitch = region.bufferRowLength != 0
-				                           ? region.bufferRowLength
-				                           : region.imageExtent.width;
-				const auto logical_height = region.bufferImageHeight != 0
-				                                ? region.bufferImageHeight
-				                                : region.imageExtent.height;
-				GpuTileInfo    info {};
+				const auto     pitch =
+				    region.bufferRowLength != 0 ? region.bufferRowLength : region.imageExtent.width;
+				const auto  logical_height = region.bufferImageHeight != 0
+				                                 ? region.bufferImageHeight
+				                                 : region.imageExtent.height;
+				GpuTileInfo info {};
 				info.family            = block.family;
 				info.bytes_per_element = block.bytes_per_element;
 				info.linear_offset     = region.bufferOffset;
@@ -544,20 +540,17 @@ bool TextureBuildGpuTileInfos(uint64_t size,
 					return false;
 				}
 				info.linear_slice_stride = linear_stride;
-				info.width = std::max(
-				    (region.imageExtent.width + element.wide - 1u) / element.wide, 1u);
-				info.height = std::max(
-				    (logical_height + element.tall - 1u) / element.tall, 1u);
-				info.depth       = copy_depth;
-				info.surface_z = block.block_depth == 1
-				                     ? static_cast<uint32_t>(region.imageOffset.z)
-				                     : 0;
-				info.pitch =
-				    std::max((pitch + element.wide - 1u) / element.wide, 1u);
-				info.tail_x      = tail ? volume.tail_x[level] : 0;
-				info.tail_y      = tail ? volume.tail_y[level] : 0;
-				info.tail        = tail;
-				info.tiled_width = volume.level_widths[level];
+				info.width =
+				    std::max((region.imageExtent.width + element.wide - 1u) / element.wide, 1u);
+				info.height = std::max((logical_height + element.tall - 1u) / element.tall, 1u);
+				info.depth  = copy_depth;
+				info.surface_z =
+				    block.block_depth == 1 ? static_cast<uint32_t>(region.imageOffset.z) : 0;
+				info.pitch        = std::max((pitch + element.wide - 1u) / element.wide, 1u);
+				info.tail_x       = tail ? volume.tail_x[level] : 0;
+				info.tail_y       = tail ? volume.tail_y[level] : 0;
+				info.tail         = tail;
+				info.tiled_width  = volume.level_widths[level];
 				info.tiled_height = volume.level_heights[level];
 				infos.push_back(info);
 			}
@@ -581,12 +574,11 @@ bool TextureBuildGpuTileInfos(uint64_t size,
 			const auto level_depth = GetTextureLevelDepth(depth, level, layout.volume_texture);
 			for (uint32_t z = 0; z < level_depth; z++) {
 				const auto& region = regions[region_index++];
-				const auto pitch = region.bufferRowLength != 0
-				                       ? region.bufferRowLength
-				                       : region.imageExtent.width;
-				const auto logical_height = region.bufferImageHeight != 0
-				                                ? region.bufferImageHeight
-				                                : region.imageExtent.height;
+				const auto  pitch =
+				    region.bufferRowLength != 0 ? region.bufferRowLength : region.imageExtent.width;
+				const auto  logical_height = region.bufferImageHeight != 0
+				                                 ? region.bufferImageHeight
+				                                 : region.imageExtent.height;
 				GpuTileInfo info {};
 				info.family            = block.family;
 				info.bytes_per_element = block.bytes_per_element;
@@ -597,16 +589,14 @@ bool TextureBuildGpuTileInfos(uint64_t size,
 				                    info.tiled_size)) {
 					return false;
 				}
-				info.width = std::max(
-				    (region.imageExtent.width + element.wide - 1u) / element.wide, 1u);
-				info.height = std::max(
-				    (logical_height + element.tall - 1u) / element.tall, 1u);
+				info.width =
+				    std::max((region.imageExtent.width + element.wide - 1u) / element.wide, 1u);
+				info.height    = std::max((logical_height + element.tall - 1u) / element.tall, 1u);
 				info.surface_z = base_family == TileBlockFamily::RenderTarget64KB ||
 				                         base_family == TileBlockFamily::Depth64KB
 				                     ? region.imageSubresource.baseArrayLayer
 				                     : 0;
-				info.pitch =
-				    std::max((pitch + element.wide - 1u) / element.wide, 1u);
+				info.pitch     = std::max((pitch + element.wide - 1u) / element.wide, 1u);
 				info.tail      = tail;
 				info.tail_x    = tail ? level_size.x : 0;
 				info.tail_y    = tail ? level_size.y : 0;

src/graphics/host_gpu/renderer/image/textureCommon.h

@@ -32,20 +32,19 @@ struct TextureUploadLayout {
 	TilePaddedSize  padded_sizes[16]    = {};
 };
 
-vk::ComponentMapping TextureGetComponentMapping(uint32_t swizzle);
+vk::ComponentMapping   TextureGetComponentMapping(uint32_t swizzle);
 vk::Format             TextureGetFormat(uint32_t fmt);
 RenderTargetFormatInfo TextureGetRenderTargetFormat(uint32_t layout, uint32_t type, uint32_t order);
-TextureUploadLayout TextureCalcUploadLayout(uint32_t fmt, uint64_t width, uint64_t height,
-                                            uint64_t levels, uint32_t depth, uint64_t pitch,
-                                            uint64_t tile, uint64_t upload_size,
-                                            bool allow_depth_tile, bool volume_texture,
-                                            const char* owner);
-std::vector<vk::BufferImageCopy>
-TextureBuildImageCopies(const TextureUploadLayout& layout, uint32_t width, uint32_t height,
-                        uint32_t depth, uint64_t levels, bool array_texture,
-                        bool volume_texture);
-bool TextureBuildGpuTileInfos(uint64_t size,
-                              const std::vector<vk::BufferImageCopy>& regions,
+TextureUploadLayout    TextureCalcUploadLayout(uint32_t fmt, uint64_t width, uint64_t height,
+                                               uint64_t levels, uint32_t depth, uint64_t pitch,
+                                               uint64_t tile, uint64_t upload_size,
+                                               bool allow_depth_tile, bool volume_texture,
+                                               const char* owner);
+std::vector<vk::BufferImageCopy> TextureBuildImageCopies(const TextureUploadLayout& layout,
+                                                         uint32_t width, uint32_t height,
+                                                         uint32_t depth, uint64_t levels,
+                                                         bool array_texture, bool volume_texture);
+bool TextureBuildGpuTileInfos(uint64_t size, const std::vector<vk::BufferImageCopy>& regions,
                               const TextureUploadLayout& layout, uint32_t fmt, uint32_t depth,
                               uint64_t levels, std::vector<GpuTileInfo>& infos);
 

src/graphics/host_gpu/renderer/image/tiler.cpp

@@ -14,9 +14,9 @@
 #include "gpu_tiler_shaders/gpu_tiler_standard64_spv.h"
 #include "gpu_tiler_shaders/gpu_tiler_swap_bgra16_spv.h"
 #include "graphics/host_gpu/graphicContext.h"
+#include "graphics/host_gpu/renderer/cache/streamBuffer.h"
 #include "graphics/host_gpu/renderer/commandScheduler.h"
 #include "graphics/host_gpu/renderer/image/image.h"
-#include "graphics/host_gpu/renderer/cache/streamBuffer.h"
 
 #include <algorithm>
 #include <array>

src/graphics/host_gpu/renderer/masterSemaphore.cpp

@@ -26,8 +26,8 @@ MasterSemaphore::~MasterSemaphore() {
 }
 
 void MasterSemaphore::Refresh() {
-	uint64_t counter = 0;
-	const auto result = m_graphics.device.getSemaphoreCounterValue(m_semaphore, &counter);
+	uint64_t   counter = 0;
+	const auto result  = m_graphics.device.getSemaphoreCounterValue(m_semaphore, &counter);
 	EXIT_NOT_IMPLEMENTED(result != vk::Result::eSuccess);
 
 	auto known = m_gpu_tick.load(std::memory_order_acquire);

src/graphics/host_gpu/renderer/masterSemaphore.h

@@ -22,7 +22,7 @@ public:
 	[[nodiscard]] uint64_t KnownGpuTick() const noexcept {
 		return m_gpu_tick.load(std::memory_order_acquire);
 	}
-	[[nodiscard]] bool IsFree(uint64_t tick) const noexcept { return KnownGpuTick() >= tick; }
+	[[nodiscard]] bool     IsFree(uint64_t tick) const noexcept { return KnownGpuTick() >= tick; }
 	[[nodiscard]] uint64_t NextTick() noexcept {
 		return m_current_tick.fetch_add(1, std::memory_order_release);
 	}

src/graphics/host_gpu/renderer/pipeline/descriptorCache.cpp

@@ -26,11 +26,15 @@ bool IsSampledImage(BindingKind kind) {
 		case BindingKind::Sampled1DArray:
 		case BindingKind::Sampled2D:
 		case BindingKind::Sampled2DArray:
+		case BindingKind::Sampled2DMsaa:
+		case BindingKind::Sampled2DMsaaArray:
 		case BindingKind::Sampled3D:
 		case BindingKind::SampledUint1D:
 		case BindingKind::SampledUint1DArray:
 		case BindingKind::SampledUint2D:
 		case BindingKind::SampledUint2DArray:
+		case BindingKind::SampledUint2DMsaa:
+		case BindingKind::SampledUint2DMsaaArray:
 		case BindingKind::SampledUint3D: return true;
 		default: return false;
 	}

src/graphics/host_gpu/renderer/pipeline/descriptorCache.h

@@ -95,7 +95,7 @@ private:
 	};
 
 	static vk::DescriptorImageInfo MakeImageInfo(const TextureBinding& texture);
-	void                 CreatePool();
+	void                           CreatePool();
 	VulkanDescriptorSet* Allocate(Stage stage, const ShaderRecompiler::IR::Program& program);
 	vk::DescriptorSetLayout
 	GetDescriptorSetLayoutInternal(Stage stage, const ShaderRecompiler::IR::Program& program);

src/graphics/host_gpu/renderer/pipeline/descriptors.cpp

@@ -73,6 +73,11 @@ static Prospero::ImageType TextureBaseType(Prospero::ImageType type) {
 	}
 }
 
+static bool IsMultisampledTexture(Prospero::ImageType type) {
+	return type == Prospero::ImageType::kColor2DMsaa ||
+	       type == Prospero::ImageType::kColor2DMsaaArray;
+}
+
 static BufferView NativeStorageBuffer(RenderContext& context, CommandBuffer& command_buffer,
                                       const ShaderBufferResource&                 descriptor,
                                       const ShaderRecompiler::IR::BufferResource& resource,
@@ -159,6 +164,8 @@ static bool IsSupportedSampledColorResource(const ShaderRecompiler::IR::ImageRes
 		case ShaderRecompiler::Decoder::ImageDimension::Dim1DArray:
 		case ShaderRecompiler::Decoder::ImageDimension::Dim2D:
 		case ShaderRecompiler::Decoder::ImageDimension::Dim2DArray:
+		case ShaderRecompiler::Decoder::ImageDimension::Dim2DMsaa:
+		case ShaderRecompiler::Decoder::ImageDimension::Dim2DMsaaArray:
 			supported_dimension = true;
 			break;
 		default: break;
@@ -195,6 +202,22 @@ TargetTextureViewInfo ResolveTargetTextureView(const ShaderRecompiler::IR::Image
 			           ? TargetTextureViewInfo {vk::ImageViewType::e2DArray, base_layer,
 			                                    image_layers - base_layer}
 			           : TargetTextureViewInfo {};
+		case Prospero::ImageType::kColor2DMsaa:
+			return resource.dimension == ShaderRecompiler::Decoder::ImageDimension::Dim2DMsaa &&
+			               base_layer == 0 && image_layers == 1
+			           ? TargetTextureViewInfo {vk::ImageViewType::e2D, 0, 1}
+			           : TargetTextureViewInfo {};
+		case Prospero::ImageType::kColor2DMsaaArray:
+			if (resource.dimension == ShaderRecompiler::Decoder::ImageDimension::Dim2DMsaa &&
+			    base_layer == 0 && image_layers == 1) {
+				return {vk::ImageViewType::e2D, 0, 1};
+			}
+			return resource.dimension ==
+			                   ShaderRecompiler::Decoder::ImageDimension::Dim2DMsaaArray &&
+			               base_layer < image_layers
+			           ? TargetTextureViewInfo {vk::ImageViewType::e2DArray, base_layer,
+			                                    image_layers - base_layer}
+			           : TargetTextureViewInfo {};
 		default: return {};
 	}
 }
@@ -209,41 +232,64 @@ bool IsSupportedSampledVideoOutView(const ShaderRecompiler::IR::ImageResource& r
 }
 
 bool IsSupportedDepthTargetDescriptor(const ShaderTextureResource& descriptor, const Image& image) {
-	const auto width  = static_cast<uint32_t>(descriptor.Width5()) + 1u;
-	const auto height = static_cast<uint32_t>(descriptor.Height5()) + 1u;
-	const auto pitch  = TileGetTexturePitch(descriptor.Format(), width, 1, descriptor.TileMode());
-	const auto type   = static_cast<Prospero::ImageType>(descriptor.Type());
-	const bool supported_single_layer =
-	    image.info.resources.layers == 1 && descriptor.Depth() == 0 &&
-	    descriptor.BaseArray5() == 0 &&
-	    (type == Prospero::ImageType::kColor2D || type == Prospero::ImageType::kColor2DArray);
+	const auto width        = static_cast<uint32_t>(descriptor.Width5()) + 1u;
+	const auto height       = static_cast<uint32_t>(descriptor.Height5()) + 1u;
+	const auto type         = static_cast<Prospero::ImageType>(descriptor.Type());
+	const bool multisampled = IsMultisampledTexture(type);
+	const auto samples      = multisampled ? 1u << descriptor.LastLevel() : 1u;
+	const auto pitch =
+	    multisampled ? TileGetDepthPitch(width, image.info.bytes_per_block, descriptor.LastLevel())
+	                 : TileGetTexturePitch(descriptor.Format(), width, 1, descriptor.TileMode());
+	const bool supported_2d    = type == Prospero::ImageType::kColor2D &&
+	                             image.info.resources.layers == 1 && descriptor.Depth() == 0 &&
+	                             descriptor.BaseArray5() == 0;
+	const bool supported_array = type == Prospero::ImageType::kColor2DArray &&
+	                             descriptor.BaseArray5() <= descriptor.Depth() &&
+	                             descriptor.Depth() < image.info.resources.layers;
 	const bool supported_cube =
 	    type == Prospero::ImageType::kCube && width == height && image.info.resources.layers >= 6 &&
 	    image.info.resources.layers % 6u == 0 &&
 	    static_cast<uint32_t>(descriptor.Depth()) + 1u == image.info.resources.layers &&
 	    descriptor.BaseArray5() == 0;
+	const bool supported_msaa_2d    = type == Prospero::ImageType::kColor2DMsaa &&
+	                                  image.info.resources.layers == 1 && descriptor.Depth() == 0 &&
+	                                  descriptor.BaseArray5() == 0;
+	const bool supported_msaa_array = type == Prospero::ImageType::kColor2DMsaaArray &&
+	                                  descriptor.BaseArray5() <= descriptor.Depth() &&
+	                                  descriptor.Depth() < image.info.resources.layers;
+	const bool levels_ok =
+	    multisampled
+	        ? descriptor.BaseLevel() == 0 && descriptor.LastLevel() >= 1 &&
+	              descriptor.LastLevel() <= 3 && descriptor.MaxMip() == descriptor.LastLevel() &&
+	              image.info.resources.levels == 1 && image.info.samples == samples
+	        : descriptor.BaseLevel() == 0 && descriptor.LastLevel() == 0 &&
+	              descriptor.MaxMip() == 0 && image.info.samples == 1;
 	return image.info.IsDepth() && width == image.info.extent.width &&
-	       height == image.info.extent.height && (supported_single_layer || supported_cube) &&
-	       descriptor.BaseLevel() == 0 && descriptor.LastLevel() == 0 && descriptor.MaxMip() == 0 &&
-	       descriptor.MinLod() == 0 && descriptor.BaseArray5() == 0 &&
+	       height == image.info.extent.height &&
+	       (supported_2d || supported_array || supported_cube || supported_msaa_2d ||
+	        supported_msaa_array) &&
+	       levels_ok && descriptor.MinLod() == 0 &&
 	       descriptor.TileMode() == Prospero::GpuEnumValue(Prospero::TileMode::kDepth) &&
-	       descriptor.BCSwizzle() == 0 && !descriptor.MsaaDepth() && pitch >= width &&
-	       pitch == image.info.pitch;
+	       descriptor.BCSwizzle() == 0 && descriptor.MsaaDepth() == multisampled &&
+	       pitch >= width && pitch == image.info.pitch;
 }
 
 bool IsSupportedDepthTextureEncoding(const ShaderTextureResource& descriptor, const Image& image) {
 	constexpr uint32_t field1_reserved_mask = 0x200fff00u;
 	constexpr uint32_t field2_reserved_mask = 0xf0003000u;
-	constexpr uint32_t field3_common        = 0x01800000u;
-	constexpr uint32_t field5_expected      = 0x00700000u;
-	const uint32_t     field3_expected =
-	    (descriptor.Type() << 28u) | field3_common | descriptor.DstSelXYZW();
-	const uint32_t field4_expected = descriptor.Depth() | (descriptor.BaseArray5() << 16u);
-	const bool     common          = (descriptor.fields[1] & field1_reserved_mask) == 0 &&
-	                                 (descriptor.fields[2] & field2_reserved_mask) == 0 &&
-	                                 descriptor.fields[3] == field3_expected &&
-	                                 descriptor.fields[4] == field4_expected &&
-	                                 descriptor.fields[5] == field5_expected;
+	const uint32_t     field3_expected = descriptor.DstSelXYZW() |
+	                                     (static_cast<uint32_t>(descriptor.BaseLevel()) << 12u) |
+	                                     (static_cast<uint32_t>(descriptor.LastLevel()) << 16u) |
+	                                     (static_cast<uint32_t>(descriptor.TileMode()) << 20u) |
+	                                     (static_cast<uint32_t>(descriptor.Type()) << 28u);
+	const uint32_t     field4_expected = descriptor.Depth() | (descriptor.BaseArray5() << 16u);
+	const uint32_t     field5_expected =
+	    0x00700000u | (static_cast<uint32_t>(descriptor.MaxMip()) << 4u);
+	const bool common = (descriptor.fields[1] & field1_reserved_mask) == 0 &&
+	                    (descriptor.fields[2] & field2_reserved_mask) == 0 &&
+	                    descriptor.fields[3] == field3_expected &&
+	                    descriptor.fields[4] == field4_expected &&
+	                    descriptor.fields[5] == field5_expected;
 	if (!common || (descriptor.fields[6] == 0 && descriptor.fields[7] != 0)) {
 		return false;
 	}
@@ -251,8 +297,9 @@ bool IsSupportedDepthTextureEncoding(const ShaderTextureResource& descriptor, co
 		return true;
 	}
 	constexpr uint32_t htile_control = 0x00280000u;
-	const auto         metadata_addr = descriptor.MetaAddr() << 8u;
-	return (descriptor.fields[6] & 0x00ffffffu) == htile_control && metadata_addr != 0 &&
+	const uint32_t expected_control  = htile_control | (descriptor.MsaaDepth() ? (1u << 10u) : 0u);
+	const auto     metadata_addr     = descriptor.MetaAddr() << 8u;
+	return (descriptor.fields[6] & 0x00ffffffu) == expected_control && metadata_addr != 0 &&
 	       descriptor.TileMode() == Prospero::GpuEnumValue(Prospero::TileMode::kDepth) &&
 	       image.info.tile_mode == Prospero::GpuEnumValue(Prospero::TileMode::kDepth) &&
 	       image.info.metadata.kind == ImageMetadataKind::Htile &&
@@ -518,6 +565,7 @@ static ImageViewInfo TextureViewInfo(const ShaderRecompiler::IR::ImageResource&
 			view.layer_count = 1;
 			break;
 		case ShaderRecompiler::Decoder::ImageDimension::Dim2DArray:
+		case ShaderRecompiler::Decoder::ImageDimension::Dim2DMsaaArray:
 			view.type       = vk::ImageViewType::e2DArray;
 			view.base_layer = descriptor.BaseArray5();
 			if (view.base_layer >= image_layers) {
@@ -526,6 +574,7 @@ static ImageViewInfo TextureViewInfo(const ShaderRecompiler::IR::ImageResource&
 			view.layer_count = image_layers - view.base_layer;
 			break;
 		case ShaderRecompiler::Decoder::ImageDimension::Dim2D:
+		case ShaderRecompiler::Decoder::ImageDimension::Dim2DMsaa:
 			view.type       = vk::ImageViewType::e2D;
 			view.base_layer = descriptor.BaseArray5();
 			if (view.base_layer >= image_layers) {
@@ -556,22 +605,23 @@ RenderExecutor::ResolveTexture(const ShaderRecompiler::IR::ImageResource&   reso
 		return {id, nullptr, std::move(desc)};
 	}
 
-	const auto address    = descriptor.Base40();
-	const auto width      = static_cast<uint32_t>(descriptor.Width5()) + 1u;
-	const auto height     = static_cast<uint32_t>(descriptor.Height5()) + 1u;
-	const auto base_level = descriptor.BaseLevel();
-	const auto last_level = descriptor.LastLevel();
-	const auto type       = TextureType(descriptor);
-	const bool multisampled =
-	    type == Prospero::ImageType::kColor2DMsaa || type == Prospero::ImageType::kColor2DMsaaArray;
-	const auto levels     = multisampled ? 1u : static_cast<uint32_t>(descriptor.MaxMip()) + 1u;
-	const auto tile       = descriptor.TileMode();
-	const bool msaa_tile  = tile == Prospero::GpuEnumValue(Prospero::TileMode::kRenderTarget);
+	const auto address      = descriptor.Base40();
+	const auto width        = static_cast<uint32_t>(descriptor.Width5()) + 1u;
+	const auto height       = static_cast<uint32_t>(descriptor.Height5()) + 1u;
+	const auto base_level   = descriptor.BaseLevel();
+	const auto last_level   = descriptor.LastLevel();
+	const auto type         = TextureType(descriptor);
+	const bool multisampled = IsMultisampledTexture(type);
+	const auto levels       = multisampled ? 1u : static_cast<uint32_t>(descriptor.MaxMip()) + 1u;
+	const auto tile         = descriptor.TileMode();
+	const bool msaa_tile =
+	    tile == Prospero::GpuEnumValue(descriptor.MsaaDepth() ? Prospero::TileMode::kDepth
+	                                                          : Prospero::TileMode::kRenderTarget);
 	const bool msaa_array = type == Prospero::ImageType::kColor2DMsaaArray;
 	if ((!multisampled && (base_level > last_level || last_level >= levels)) ||
 	    (multisampled &&
 	     (base_level != 0 || last_level == 0 || last_level > 3 ||
-	      descriptor.MaxMip() != last_level || !msaa_tile || descriptor.MsaaDepth() ||
+	      descriptor.MaxMip() != last_level || !msaa_tile ||
 	      (!msaa_array && (descriptor.Depth() != 0 || descriptor.BaseArray5() != 0))))) {
 		EXIT("unsupported texture mip view: base=%u last=%u levels=%u\n", base_level, last_level,
 		     levels);

src/graphics/host_gpu/renderer/pipeline/descriptors.h

@@ -36,7 +36,7 @@ ResolveTargetTextureView(const ShaderRecompiler::IR::ImageResource& resource,
 [[nodiscard]] bool IsSupportedDepthTargetDescriptor(const ShaderTextureResource& descriptor,
                                                     const Image&                 image);
 [[nodiscard]] bool IsSupportedDepthTextureEncoding(const ShaderTextureResource& descriptor,
-                                                    const Image&                 image);
+                                                   const Image&                 image);
 [[nodiscard]] bool
 IsSupportedSampledVideoOutView(const ShaderRecompiler::IR::ImageResource& resource,
                                const ShaderTextureResource& descriptor, const Image& image);

src/graphics/host_gpu/renderer/pipeline/pipelineCache.cpp

@@ -88,12 +88,12 @@ PipelineCache::GraphicsPipeline& PipelineCache::CreateGraphicsPipeline(
 
 	PipelineStaticParameters static_params {};
 	GraphicsPipeline         p {};
-	p.ps_shader_id   = ps_id;
-	p.vs_shader_id   = vs_id;
+	p.ps_shader_id = ps_id;
+	p.vs_shader_id = vs_id;
 
 	static_params.color_count = color_count;
 	PipelineRenderingState rendering {};
-	rendering.color_count = color_count;
+	rendering.color_count       = color_count;
 	uint32_t attachment_samples = 0;
 	for (uint32_t i = 0; i < color_count; i++) {
 		EXIT_IF(!colors[i].image_id || colors[i].format == vk::Format::eUndefined);
@@ -116,8 +116,8 @@ PipelineCache::GraphicsPipeline& PipelineCache::CreateGraphicsPipeline(
 		if (attachment_samples == 0) {
 			attachment_samples = depth.samples;
 		} else if (attachment_samples != depth.samples) {
-			EXIT("mixed color/depth sample counts are unsupported: %u and %u\n",
-			     attachment_samples, depth.samples);
+			EXIT("mixed color/depth sample counts are unsupported: %u and %u\n", attachment_samples,
+			     depth.samples);
 		}
 	}
 	EXIT_IF(attachment_samples == 0 ||
@@ -179,10 +179,10 @@ PipelineCache::GraphicsPipeline& PipelineCache::CreateGraphicsPipeline(
 	NormalizeStaticParamsForDynamicState(static_params);
 
 	GraphicsPipelineKey key {};
-	key.rendering      = rendering;
-	key.vs_shader_id   = p.vs_shader_id;
-	key.ps_shader_id   = p.ps_shader_id;
-	key.static_params  = static_params;
+	key.rendering     = rendering;
+	key.vs_shader_id  = p.vs_shader_id;
+	key.ps_shader_id  = p.ps_shader_id;
+	key.static_params = static_params;
 
 	if (auto iter = m_graphics_pipelines.find(key); iter != m_graphics_pipelines.end()) {
 		return *iter->second;
@@ -203,9 +203,8 @@ PipelineCache::GraphicsPipeline& PipelineCache::CreateGraphicsPipeline(
 	LogPipelineTrace("CreatePipelineInternal begin", vs_id.hash0, vs_id.crc32, ps_id.hash0,
 	                 ps_id.crc32);
 	CreatePipelineInternal(m_graphics, m_descriptor_cache, *cached, rendering, vs_input_info,
-	                       vs_spirv, ps_input_info,
-	                       ps_spirv, static_params, vs_id.hash0, vs_id.crc32, ps_id.hash0,
-	                       ps_id.crc32, ps_active);
+	                       vs_spirv, ps_input_info, ps_spirv, static_params, vs_id.hash0,
+	                       vs_id.crc32, ps_id.hash0, ps_id.crc32, ps_active);
 	LogPipelineTrace("CreatePipelineInternal done", vs_id.hash0, vs_id.crc32, ps_id.hash0,
 	                 ps_id.crc32);
 

src/graphics/host_gpu/renderer/pipeline/pipelineCache.h

@@ -88,9 +88,9 @@ static_assert(sizeof(PipelineStaticParameters) ==
 
 struct PipelineRenderingState {
 	std::array<vk::Format, RENDER_COLOR_ATTACHMENTS_MAX> color_formats {};
-	vk::Format depth_format   = vk::Format::eUndefined;
-	vk::Format stencil_format = vk::Format::eUndefined;
-	uint32_t   color_count    = 0;
+	vk::Format                                           depth_format   = vk::Format::eUndefined;
+	vk::Format                                           stencil_format = vk::Format::eUndefined;
+	uint32_t                                             color_count    = 0;
 
 	bool operator==(const PipelineRenderingState&) const = default;
 };
@@ -118,11 +118,12 @@ public:
 		ShaderId cs_shader_id;
 	};
 
-	GraphicsPipeline& CreateGraphicsPipeline(
-	    RenderColorInfo* colors, uint32_t color_count, RenderDepthInfo& depth,
-	    ShaderVertexInputInfo& vs_input_info, RenderCommandBuffer& command,
-	    ShaderPixelInputInfo* ps_input_info, vk::PrimitiveTopology topology, bool ps_active,
-	    std::span<const uint32_t> vs_spirv, std::span<const uint32_t> ps_spirv);
+	GraphicsPipeline&
+	CreateGraphicsPipeline(RenderColorInfo* colors, uint32_t color_count, RenderDepthInfo& depth,
+	                       ShaderVertexInputInfo& vs_input_info, RenderCommandBuffer& command,
+	                       ShaderPixelInputInfo* ps_input_info, vk::PrimitiveTopology topology,
+	                       bool ps_active, std::span<const uint32_t> vs_spirv,
+	                       std::span<const uint32_t> ps_spirv);
 	ComputePipeline& CreateComputePipeline(ShaderComputeInputInfo&      input_info,
 	                                       const HW::ComputeShaderInfo& cs_regs,
 	                                       std::span<const uint32_t>    cs_spirv);
@@ -199,7 +200,7 @@ private:
 		}
 	};
 
-	GraphicContext& m_graphics;
+	GraphicContext&  m_graphics;
 	DescriptorCache& m_descriptor_cache;
 	std::unordered_map<GraphicsPipelineKey, std::unique_ptr<GraphicsPipeline>,
 	                   GraphicsPipelineKeyHash>
@@ -211,16 +212,13 @@ private:
 
 void LogPipelineTrace(const char* phase, uint32_t vs_hash0, uint32_t vs_crc32, uint32_t ps_hash0,
                       uint32_t ps_crc32);
-void CreatePipelineInternal(GraphicContext& graphics, DescriptorCache& descriptor_cache,
-                            PipelineCache::GraphicsPipeline& pipeline,
-                            const PipelineRenderingState&   rendering,
-                            const ShaderVertexInputInfo&    vs_input_info,
-                            std::span<const uint32_t>       vs_shader,
-                            const ShaderPixelInputInfo*     ps_input_info,
-                            std::span<const uint32_t>       ps_shader,
-                            const PipelineStaticParameters& static_params, uint32_t vs_hash0,
-                            uint32_t vs_crc32, uint32_t ps_hash0, uint32_t ps_crc32,
-                            bool ps_active);
+void CreatePipelineInternal(
+    GraphicContext& graphics, DescriptorCache& descriptor_cache,
+    PipelineCache::GraphicsPipeline& pipeline, const PipelineRenderingState& rendering,
+    const ShaderVertexInputInfo& vs_input_info, std::span<const uint32_t> vs_shader,
+    const ShaderPixelInputInfo* ps_input_info, std::span<const uint32_t> ps_shader,
+    const PipelineStaticParameters& static_params, uint32_t vs_hash0, uint32_t vs_crc32,
+    uint32_t ps_hash0, uint32_t ps_crc32, bool ps_active);
 void CreatePipelineInternal(GraphicContext& graphics, DescriptorCache& descriptor_cache,
                             PipelineCache::ComputePipeline& pipeline,
                             const ShaderComputeInputInfo&   input_info,

src/graphics/host_gpu/renderer/pipeline/shaders.cpp

@@ -8,10 +8,10 @@
 #include "graphics/host_gpu/renderer/debug.h"
 #include "graphics/host_gpu/renderer/pipeline/descriptorCache.h"
 #include "graphics/host_gpu/renderer/pipeline/pipelineCache.h"
+#include "graphics/host_gpu/renderer/pipeline/shaderSubgroup.h"
 #include "graphics/host_gpu/renderer/render.h"
 #include "graphics/host_gpu/renderer/renderContext.h"
 #include "graphics/host_gpu/renderer/renderTarget.h"
-#include "graphics/host_gpu/renderer/pipeline/shaderSubgroup.h"
 #include "graphics/host_gpu/vulkanCommon.h"
 #include "graphics/shader/recompiler/ir/ShaderIR.h"
 #include "graphics/shader/shader.h"
@@ -385,9 +385,8 @@ static vk::BlendOp GetBlendOp(uint32_t op) {
 	return vk::BlendOp::eAdd;
 }
 
-static void CreateLayout(DescriptorCache& descriptor_cache,
-                         std::span<vk::DescriptorSetLayout> set_layouts,
-                         uint32_t& set_layouts_num,
+static void CreateLayout(DescriptorCache&                   descriptor_cache,
+                         std::span<vk::DescriptorSetLayout> set_layouts, uint32_t& set_layouts_num,
                          std::span<vk::PushConstantRange>     push_constant_info,
                          uint32_t&                            push_constant_info_num,
                          const ShaderRecompiler::IR::Program& program,
@@ -412,12 +411,11 @@ static void CreateLayout(DescriptorCache& descriptor_cache,
 	}
 }
 
-static void ConfigureSubgroupSize(const GraphicContext&                                  graphics,
-                                  vk::ShaderStageFlagBits                                vk_stage,
+static void ConfigureSubgroupSize(const GraphicContext& graphics, vk::ShaderStageFlagBits vk_stage,
                                   const ShaderRecompiler::IR::Program&                   program,
                                   vk::PipelineShaderStageRequiredSubgroupSizeCreateInfo& required,
                                   vk::PipelineShaderStageCreateInfo&                     stage) {
-	const auto  config =
+	const auto config =
 	    ConfigureShaderSubgroup(ShaderSubgroupCapabilities {graphics}, vk_stage, program);
 	switch (config.mode) {
 		case ShaderSubgroupMode::Natural: return;
@@ -456,16 +454,13 @@ static void ConfigureSubgroupSize(const GraphicContext&
 }
 
 // NOLINTNEXTLINE(readability-function-cognitive-complexity)
-void CreatePipelineInternal(GraphicContext& graphics, DescriptorCache& descriptor_cache,
-                            PipelineCache::GraphicsPipeline& pipeline,
-                            const PipelineRenderingState&   rendering,
-                            const ShaderVertexInputInfo&    vs_input_info,
-                            std::span<const uint32_t>       vs_shader,
-                            const ShaderPixelInputInfo*     ps_input_info,
-                            std::span<const uint32_t>       ps_shader,
-                            const PipelineStaticParameters& static_params, uint32_t vs_hash0,
-                            uint32_t vs_crc32, uint32_t ps_hash0, uint32_t ps_crc32,
-                            bool ps_active) {
+void CreatePipelineInternal(
+    GraphicContext& graphics, DescriptorCache& descriptor_cache,
+    PipelineCache::GraphicsPipeline& pipeline, const PipelineRenderingState& rendering,
+    const ShaderVertexInputInfo& vs_input_info, std::span<const uint32_t> vs_shader,
+    const ShaderPixelInputInfo* ps_input_info, std::span<const uint32_t> ps_shader,
+    const PipelineStaticParameters& static_params, uint32_t vs_hash0, uint32_t vs_crc32,
+    uint32_t ps_hash0, uint32_t ps_crc32, bool ps_active) {
 	EXIT_IF(ps_active && ps_input_info == nullptr);
 
 	vk::ShaderModule vert_shader_module = nullptr;
@@ -511,8 +506,7 @@ void CreatePipelineInternal(GraphicContext& graphics, DescriptorCache& descripto
 	vert_shader_stage_info.pName               = "main";
 	vert_shader_stage_info.pSpecializationInfo = nullptr;
 	EXIT_IF(!vs_input_info.stage);
-	ConfigureSubgroupSize(graphics, vk::ShaderStageFlagBits::eVertex,
-	                      *vs_input_info.stage.program,
+	ConfigureSubgroupSize(graphics, vk::ShaderStageFlagBits::eVertex, *vs_input_info.stage.program,
 	                      vert_subgroup_size, vert_shader_stage_info);
 
 	vk::PipelineShaderStageCreateInfo                     frag_shader_stage_info {};
@@ -527,8 +521,8 @@ void CreatePipelineInternal(GraphicContext& graphics, DescriptorCache& descripto
 	if (ps_active) {
 		EXIT_IF(!ps_input_info->stage);
 		ConfigureSubgroupSize(graphics, vk::ShaderStageFlagBits::eFragment,
-		                      *ps_input_info->stage.program,
-		                      frag_subgroup_size, frag_shader_stage_info);
+		                      *ps_input_info->stage.program, frag_subgroup_size,
+		                      frag_shader_stage_info);
 	}
 
 	vk::PipelineShaderStageCreateInfo shader_stages[]    = {vert_shader_stage_info,
@@ -728,13 +722,13 @@ void CreatePipelineInternal(GraphicContext& graphics, DescriptorCache& descripto
 	clip_ext.depthClipEnable = static_params.depth_clip_enable ? VK_TRUE : VK_FALSE;
 
 	vk::PipelineRasterizationStateCreateInfo rasterizer {};
-	rasterizer.sType                   = vk::StructureType::ePipelineRasterizationStateCreateInfo;
+	rasterizer.sType = vk::StructureType::ePipelineRasterizationStateCreateInfo;
 	// MoltenVK lacks VK_EXT_depth_clip_enable; omit the depth-clip struct on macOS and accept
 	// Vulkan's default depth clipping (enabled) instead of the PS5's clamp behavior.
 #if defined(__APPLE__)
-	rasterizer.pNext                   = nullptr;
+	rasterizer.pNext = nullptr;
 #else
-	rasterizer.pNext                   = &clip_ext;
+	rasterizer.pNext = &clip_ext;
 #endif
 	rasterizer.flags                   = {};
 	rasterizer.depthClampEnable        = VK_FALSE;
@@ -812,13 +806,13 @@ void CreatePipelineInternal(GraphicContext& graphics, DescriptorCache& descripto
 	color_write.pColorWriteEnables = color_write_enable;
 
 	vk::PipelineColorBlendStateCreateInfo color_blending {};
-	color_blending.sType             = vk::StructureType::ePipelineColorBlendStateCreateInfo;
+	color_blending.sType = vk::StructureType::ePipelineColorBlendStateCreateInfo;
 	// MoltenVK lacks VK_EXT_color_write_enable; drop the dynamic color-write struct on macOS
 	// and rely on each attachment's static colorWriteMask (all channels enabled by default).
 #if defined(__APPLE__)
-	color_blending.pNext             = nullptr;
+	color_blending.pNext = nullptr;
 #else
-	color_blending.pNext             = &color_write;
+	color_blending.pNext = &color_write;
 #endif
 	color_blending.flags             = {};
 	color_blending.logicOpEnable     = VK_FALSE;
@@ -838,15 +832,13 @@ void CreatePipelineInternal(GraphicContext& graphics, DescriptorCache& descripto
 
 	EXIT_IF(!vs_input_info.stage);
 	CreateLayout(descriptor_cache, set_layouts, set_layouts_num, push_constant_info,
-	             push_constant_info_num,
-	             *vs_input_info.stage.program, vk::ShaderStageFlagBits::eVertex,
-	             DescriptorCache::Stage::Vertex);
+	             push_constant_info_num, *vs_input_info.stage.program,
+	             vk::ShaderStageFlagBits::eVertex, DescriptorCache::Stage::Vertex);
 	if (ps_active) {
 		EXIT_IF(!ps_input_info->stage);
 		CreateLayout(descriptor_cache, set_layouts, set_layouts_num, push_constant_info,
-		             push_constant_info_num,
-		             *ps_input_info->stage.program, vk::ShaderStageFlagBits::eFragment,
-		             DescriptorCache::Stage::Pixel);
+		             push_constant_info_num, *ps_input_info->stage.program,
+		             vk::ShaderStageFlagBits::eFragment, DescriptorCache::Stage::Pixel);
 	}
 
 	vk::PipelineLayoutCreateInfo pipeline_layout_info {};
@@ -923,32 +915,32 @@ void CreatePipelineInternal(GraphicContext& graphics, DescriptorCache& descripto
 	dynamic_state.dynamicStateCount = dynamic_states_count;
 	dynamic_state.pDynamicStates    = dynamic_states;
 
-	vk::GraphicsPipelineCreateInfo pipeline_info {};
+	vk::GraphicsPipelineCreateInfo  pipeline_info {};
 	vk::PipelineRenderingCreateInfo rendering_info {};
 	rendering_info.sType                   = vk::StructureType::ePipelineRenderingCreateInfo;
 	rendering_info.colorAttachmentCount    = rendering.color_count;
 	rendering_info.pColorAttachmentFormats = rendering.color_formats.data();
 	rendering_info.depthAttachmentFormat   = rendering.depth_format;
 	rendering_info.stencilAttachmentFormat = rendering.stencil_format;
-	pipeline_info.sType               = vk::StructureType::eGraphicsPipelineCreateInfo;
-	pipeline_info.pNext               = &rendering_info;
-	pipeline_info.flags               = {};
-	pipeline_info.stageCount          = shader_stage_count;
-	pipeline_info.pStages             = shader_stages;
-	pipeline_info.pVertexInputState   = &vertex_input_info;
-	pipeline_info.pInputAssemblyState = &input_assembly;
-	pipeline_info.pTessellationState  = nullptr;
-	pipeline_info.pViewportState      = &viewport_state;
-	pipeline_info.pRasterizationState = &rasterizer;
-	pipeline_info.pMultisampleState   = &multisampling;
-	pipeline_info.pDepthStencilState  = (static_params.with_depth ? &depth_stencil_info : nullptr);
-	pipeline_info.pColorBlendState    = &color_blending;
-	pipeline_info.pDynamicState       = &dynamic_state;
-	pipeline_info.layout              = pipeline.pipeline_layout;
-	pipeline_info.renderPass          = nullptr;
-	pipeline_info.subpass             = 0;
-	pipeline_info.basePipelineHandle  = nullptr;
-	pipeline_info.basePipelineIndex   = -1;
+	pipeline_info.sType                    = vk::StructureType::eGraphicsPipelineCreateInfo;
+	pipeline_info.pNext                    = &rendering_info;
+	pipeline_info.flags                    = {};
+	pipeline_info.stageCount               = shader_stage_count;
+	pipeline_info.pStages                  = shader_stages;
+	pipeline_info.pVertexInputState        = &vertex_input_info;
+	pipeline_info.pInputAssemblyState      = &input_assembly;
+	pipeline_info.pTessellationState       = nullptr;
+	pipeline_info.pViewportState           = &viewport_state;
+	pipeline_info.pRasterizationState      = &rasterizer;
+	pipeline_info.pMultisampleState        = &multisampling;
+	pipeline_info.pDepthStencilState = (static_params.with_depth ? &depth_stencil_info : nullptr);
+	pipeline_info.pColorBlendState   = &color_blending;
+	pipeline_info.pDynamicState      = &dynamic_state;
+	pipeline_info.layout             = pipeline.pipeline_layout;
+	pipeline_info.renderPass         = nullptr;
+	pipeline_info.subpass            = 0;
+	pipeline_info.basePipelineHandle = nullptr;
+	pipeline_info.basePipelineIndex  = -1;
 
 	EXIT_IF(pipeline.pipeline != nullptr);
 
@@ -1012,8 +1004,7 @@ void CreatePipelineInternal(GraphicContext& graphics, DescriptorCache& descripto
 	comp_shader_stage_info.pName               = "main";
 	comp_shader_stage_info.pSpecializationInfo = nullptr;
 	EXIT_IF(!input_info.stage);
-	ConfigureSubgroupSize(graphics, vk::ShaderStageFlagBits::eCompute,
-	                      *input_info.stage.program,
+	ConfigureSubgroupSize(graphics, vk::ShaderStageFlagBits::eCompute, *input_info.stage.program,
 	                      comp_subgroup_size, comp_shader_stage_info);
 
 	vk::DescriptorSetLayout set_layouts[1]  = {};
@@ -1024,9 +1015,8 @@ void CreatePipelineInternal(GraphicContext& graphics, DescriptorCache& descripto
 
 	EXIT_IF(!input_info.stage);
 	CreateLayout(descriptor_cache, set_layouts, set_layouts_num, push_constant_info,
-	             push_constant_info_num,
-	             *input_info.stage.program, vk::ShaderStageFlagBits::eCompute,
-	             DescriptorCache::Stage::Compute);
+	             push_constant_info_num, *input_info.stage.program,
+	             vk::ShaderStageFlagBits::eCompute, DescriptorCache::Stage::Compute);
 
 	vk::PipelineLayoutCreateInfo pipeline_layout_info {};
 	pipeline_layout_info.sType                  = vk::StructureType::ePipelineLayoutCreateInfo;

src/graphics/host_gpu/renderer/renderCompute.cpp

@@ -10,14 +10,14 @@
 #include "graphics/guest_gpu/graphicsRun.h"
 #include "graphics/guest_gpu/hardwareContext.h"
 #include "graphics/host_gpu/graphicContext.h"
+#include "graphics/host_gpu/renderer/image/imageInfo.h"
 #include "graphics/host_gpu/renderer/pipeline/descriptorCache.h"
 #include "graphics/host_gpu/renderer/pipeline/descriptors.h"
-#include "graphics/host_gpu/renderer/image/imageInfo.h"
 #include "graphics/host_gpu/renderer/pipeline/pipelineCache.h"
-#include "graphics/host_gpu/renderer/render.h"
-#include "graphics/host_gpu/renderer/renderContext.h"
 #include "graphics/host_gpu/renderer/pipeline/shaderResourceBarrier.h"
 #include "graphics/host_gpu/renderer/pipeline/shaderSubgroup.h"
+#include "graphics/host_gpu/renderer/render.h"
+#include "graphics/host_gpu/renderer/renderContext.h"
 #include "graphics/host_gpu/vulkanCommon.h"
 #include "graphics/shader/recompiler/ir/ResourceMaterialization.h"
 #include "graphics/shader/recompiler/ir/ShaderIR.h"

src/graphics/host_gpu/renderer/renderContext.cpp

@@ -14,8 +14,7 @@ namespace Libs::Graphics {
 RenderContext::RenderContext(GraphicContext& graphics)
     : m_graphics(graphics), m_render_executor(*this), m_command_scheduler(*this, graphics),
       m_descriptor_cache(graphics), m_pipeline_cache(graphics, m_descriptor_cache),
-      m_sampler_cache(graphics),
-      m_gpu_resources(graphics, m_command_scheduler) {
+      m_sampler_cache(graphics), m_gpu_resources(graphics, m_command_scheduler) {
 	EXIT_NOT_IMPLEMENTED(!Common::Thread::IsMainThread());
 }
 
@@ -27,7 +26,7 @@ RenderContext::~RenderContext() {
 void RenderContext::InitializeGpu(VideoOut::VideoOutDriver* video_out) {
 	EXIT_IF(m_gpu != nullptr);
 	m_video_out = video_out;
-	m_gpu = std::make_unique<Gpu>(*this);
+	m_gpu       = std::make_unique<Gpu>(*this);
 	m_gpu_resources.SetGpu(m_gpu.get());
 }
 
@@ -99,8 +98,7 @@ void RenderContext::TriggerEopEvent(uint32_t context_id) {
 		    registration.eq, static_cast<uintptr_t>(registration.id),
 		    LibKernel::EventQueue::KERNEL_EVFILT_GRAPHICS,
 		    reinterpret_cast<void*>(static_cast<uintptr_t>(context_id)));
-		if (result == LibKernel::KERNEL_ERROR_EBADF ||
-		    result == LibKernel::KERNEL_ERROR_ENOENT) {
+		if (result == LibKernel::KERNEL_ERROR_EBADF || result == LibKernel::KERNEL_ERROR_ENOENT) {
 			DeleteEopEq(registration.eq, registration.id);
 			continue;
 		}

src/graphics/host_gpu/renderer/renderContext.h

@@ -6,12 +6,12 @@
 #include "common/common.h"
 #include "common/threads.h"
 #include "graphics/host_gpu/renderer/cache/bufferCache.h"
-#include "graphics/host_gpu/renderer/commandScheduler.h"
-#include "graphics/host_gpu/renderer/pipeline/descriptorCache.h"
 #include "graphics/host_gpu/renderer/cache/gpuResourceManager.h"
-#include "graphics/host_gpu/renderer/pipeline/pipelineCache.h"
 #include "graphics/host_gpu/renderer/cache/samplerCache.h"
 #include "graphics/host_gpu/renderer/cache/textureCache.h"
+#include "graphics/host_gpu/renderer/commandScheduler.h"
+#include "graphics/host_gpu/renderer/pipeline/descriptorCache.h"
+#include "graphics/host_gpu/renderer/pipeline/pipelineCache.h"
 #include "kernel/eventQueue.h"
 
 #include <memory>
@@ -32,10 +32,10 @@ public:
 	~RenderContext();
 	KYTY_CLASS_NO_COPY(RenderContext);
 
-	[[nodiscard]] GraphicContext& GetGraphics() const noexcept { return m_graphics; }
-	void                          InitializeGpu(VideoOut::VideoOutDriver* video_out);
-	void                          ShutdownGpu();
-	[[nodiscard]] Gpu&            GetGpu() const;
+	[[nodiscard]] GraphicContext&           GetGraphics() const noexcept { return m_graphics; }
+	void                                    InitializeGpu(VideoOut::VideoOutDriver* video_out);
+	void                                    ShutdownGpu();
+	[[nodiscard]] Gpu&                      GetGpu() const;
 	[[nodiscard]] VideoOut::VideoOutDriver& GetVideoOut() const;
 
 	Common::Mutex&      GetMutex() { return m_mutex; }
@@ -56,18 +56,18 @@ private:
 	struct EopEqRegistration {
 		LibKernel::EventQueue::KernelEqueue    eq = LibKernel::EventQueue::KERNEL_EQUEUE_INVALID;
 		LibKernel::EventQueue::KernelEqueueRef queue;
-		int                                   id = 0;
+		int                                    id = 0;
 	};
 
-	GraphicContext&      m_graphics;
-	Common::Mutex        m_mutex;
-	RenderExecutor       m_render_executor;
-	CommandScheduler     m_command_scheduler;
-	DescriptorCache      m_descriptor_cache;
-	PipelineCache        m_pipeline_cache;
-	SamplerCache         m_sampler_cache;
-	GpuResourceManager   m_gpu_resources;
-	std::unique_ptr<Gpu> m_gpu;
+	GraphicContext&           m_graphics;
+	Common::Mutex             m_mutex;
+	RenderExecutor            m_render_executor;
+	CommandScheduler          m_command_scheduler;
+	DescriptorCache           m_descriptor_cache;
+	PipelineCache             m_pipeline_cache;
+	SamplerCache              m_sampler_cache;
+	GpuResourceManager        m_gpu_resources;
+	std::unique_ptr<Gpu>      m_gpu;
 	VideoOut::VideoOutDriver* m_video_out = nullptr;
 
 	Common::Mutex                  m_eop_mutex;

src/graphics/host_gpu/renderer/renderTarget.h

@@ -12,13 +12,13 @@ namespace Libs::Graphics {
 static constexpr uint32_t RENDER_COLOR_ATTACHMENTS_MAX = 8;
 
 struct RenderAttachment {
-	vk::ImageView           image_view   = nullptr;
-	vk::ImageLayout         image_layout = vk::ImageLayout::eUndefined;
-	std::array<uint32_t, 4> clear_value  = {};
+	vk::ImageView           image_view    = nullptr;
+	vk::ImageLayout         image_layout  = vk::ImageLayout::eUndefined;
+	std::array<uint32_t, 4> clear_value   = {};
 	bool                    is_clear      = false;
-	bool                    has_depth    = false;
-	bool                    depth_clear  = false;
-	bool                    has_stencil  = false;
+	bool                    has_depth     = false;
+	bool                    depth_clear   = false;
+	bool                    has_stencil   = false;
 	bool                    stencil_clear = false;
 
 	bool operator==(const RenderAttachment&) const = default;

src/graphics/host_gpu/renderer/sync.cpp

@@ -251,9 +251,9 @@ uint64_t PrepareVideoOutFlip(CommandBuffer& buffer, int handle, int index, int f
                              int64_t flip_arg) {
 	for (;;) {
 		uint64_t   request_id = 0;
-		auto&      video_out = buffer.GetContext().GetVideoOut();
-		const auto result = video_out.SubmitFlipFromGpu(
-		    buffer, handle, index, flip_mode, flip_arg, request_id);
+		auto&      video_out  = buffer.GetContext().GetVideoOut();
+		const auto result =
+		    video_out.SubmitFlipFromGpu(buffer, handle, index, flip_mode, flip_arg, request_id);
 		if (result == OK) {
 			EXIT_IF(request_id == 0);
 			return request_id;

src/graphics/host_gpu/vma.cpp

@@ -122,9 +122,9 @@ uint64_t GraphicContext::GetDeviceMemoryUsage() const {
 	    physical_device_properties.deviceType == vk::PhysicalDeviceType::eDiscreteGpu;
 	uint64_t usage = 0;
 	for (uint32_t heap = 0; heap < physical_device_memory_properties.memoryHeapCount; heap++) {
-		const bool device_local = static_cast<bool>(
-		    physical_device_memory_properties.memoryHeaps[heap].flags &
-		    vk::MemoryHeapFlagBits::eDeviceLocal);
+		const bool device_local =
+		    static_cast<bool>(physical_device_memory_properties.memoryHeaps[heap].flags &
+		                      vk::MemoryHeapFlagBits::eDeviceLocal);
 		if (!discrete || device_local) {
 			usage += budgets[heap].usage;
 		}
@@ -144,7 +144,7 @@ uint64_t GraphicContext::GetTotalMemoryBudget() const {
 	uint64_t local  = 0;
 	uint64_t usage  = 0;
 	for (uint32_t heap = 0; heap < physical_device_memory_properties.memoryHeapCount; heap++) {
-		const auto& properties   = physical_device_memory_properties.memoryHeaps[heap];
+		const auto& properties = physical_device_memory_properties.memoryHeaps[heap];
 		const bool  device_local =
 		    static_cast<bool>(properties.flags & vk::MemoryHeapFlagBits::eDeviceLocal);
 		if (device_local) {
@@ -159,9 +159,8 @@ uint64_t GraphicContext::GetTotalMemoryBudget() const {
 		return budget - std::min<uint64_t>(budget / 8, 1024ull * 1024 * 1024);
 	}
 	constexpr uint64_t system_reserve = 8ull * 1024 * 1024 * 1024;
-	const auto available = budget > usage ? budget - usage : uint64_t {0};
-	return std::max(local,
-	                available > system_reserve ? available - system_reserve : uint64_t {0});
+	const auto         available      = budget > usage ? budget - usage : uint64_t {0};
+	return std::max(local, available > system_reserve ? available - system_reserve : uint64_t {0});
 }
 
 void GraphicContext::CreateBuffer(uint64_t size, VulkanBuffer& buffer) {

src/graphics/host_gpu/vulkanCommon.cpp

@@ -55,6 +55,10 @@ constexpr FormatMapping kFormatMappings[] = {
     {Prospero::BufferFormat::k32_32_32_32UInt, vk::Format::eR32G32B32A32Uint},
     {Prospero::BufferFormat::k32_32_32_32SInt, vk::Format::eR32G32B32A32Sint},
     {Prospero::BufferFormat::k32_32_32_32Float, vk::Format::eR32G32B32A32Sfloat},
+    // Narrow-channel sRGB formats are optional in Vulkan. Keep a same-width fallback until
+    // sampler-aware sRGB emulation is available.
+    {Prospero::BufferFormat::k8Srgb, vk::Format::eR8Unorm},
+    {Prospero::BufferFormat::k8_8Srgb, vk::Format::eR8G8Unorm},
     {Prospero::BufferFormat::k8_8_8_8Srgb, vk::Format::eR8G8B8A8Srgb},
     {Prospero::BufferFormat::k9_9_9_5Float, vk::Format::eE5B9G9R9UfloatPack32},
     {Prospero::BufferFormat::k5_6_5UNorm, vk::Format::eB5G6R5UnormPack16},

src/graphics/presentation/presenter.h

@@ -20,14 +20,14 @@ public:
 	~Presenter();
 	KYTY_CLASS_NO_COPY(Presenter);
 
-	[[nodiscard]] Frame& PrepareFrame(CommandBuffer& command, const ImageInfo& info);
-	[[nodiscard]] Frame& PrepareBlankFrame(uint32_t width, uint32_t height, bool opaque,
-	                                       CommandBuffer* producer = nullptr);
-	[[nodiscard]] Frame* PrepareLastFrame();
-	[[nodiscard]] bool   IsGuestPaused() const noexcept;
+	[[nodiscard]] Frame&         PrepareFrame(CommandBuffer& command, const ImageInfo& info);
+	[[nodiscard]] Frame&         PrepareBlankFrame(uint32_t width, uint32_t height, bool opaque,
+	                                               CommandBuffer* producer = nullptr);
+	[[nodiscard]] Frame*         PrepareLastFrame();
+	[[nodiscard]] bool           IsGuestPaused() const noexcept;
 	[[nodiscard]] RenderContext& Renderer() const noexcept;
-	void                 Present(Frame& frame, bool reuse = false);
-	void                 Discard(Frame& frame);
+	void                         Present(Frame& frame, bool reuse = false);
+	void                         Discard(Frame& frame);
 
 private:
 	struct Impl;

src/graphics/presentation/videoOut.cpp

@@ -69,8 +69,8 @@ enum class FlipRequestSource { Cpu, GpuEop };
 struct VideoOutEventState;
 
 struct VideoOutEventRegistration {
-	EventQueue::KernelEqueue             handle = EventQueue::KERNEL_EQUEUE_INVALID;
-	std::shared_ptr<VideoOutEventState>  state;
+	EventQueue::KernelEqueue            handle = EventQueue::KERNEL_EQUEUE_INVALID;
+	std::shared_ptr<VideoOutEventState> state;
 	uint64_t                            generation = 0;
 	VideoOutEventKind                   kind       = VideoOutEventKind::Flip;
 };
@@ -170,13 +170,13 @@ struct BufferAttributeGroup {
 struct VideoOutConfig {
 	Common::Mutex                       mutex;
 	Common::CondVar                     vblank_cond;
-	std::shared_ptr<VideoOutEventState> events = std::make_shared<VideoOutEventState>();
-	uint32_t                            width      = 0;
-	uint32_t                            height     = 0;
-	uint64_t                            generation = 0;
-	bool                                opened     = false;
-	bool                                closing    = false;
-	int                                 flip_rate  = 0;
+	std::shared_ptr<VideoOutEventState> events      = std::make_shared<VideoOutEventState>();
+	uint32_t                            width       = 0;
+	uint32_t                            height      = 0;
+	uint64_t                            generation  = 0;
+	bool                                opened      = false;
+	bool                                closing     = false;
+	int                                 flip_rate   = 0;
 	uint64_t                            output_mode = VIDEO_OUT_OUTPUT_MODE_DEFAULT;
 	float                               gamma       = 1.0f;
 	VideoOutFlipStatus                  flip_status;
@@ -250,8 +250,8 @@ public:
 	VideoOutConfig* Get(int handle, uint64_t& generation);
 	bool            IsOpened(int handle);
 
-	void Init(uint32_t width, uint32_t height);
-	FlipQueue& GetFlipQueue() { return m_flip_queue; }
+	void                     Init(uint32_t width, uint32_t height);
+	FlipQueue&               GetFlipQueue() { return m_flip_queue; }
 	Graphics::RenderContext& Renderer() const noexcept { return m_renderer; }
 
 	void VblankBegin();
@@ -259,12 +259,12 @@ public:
 	void PresentThread(std::stop_token token);
 
 private:
-	Common::Mutex        m_mutex;
-	VideoOutConfig       m_video_out_ctx[VIDEO_OUT_NUM_MAX];
+	Common::Mutex            m_mutex;
+	VideoOutConfig           m_video_out_ctx[VIDEO_OUT_NUM_MAX];
 	Graphics::RenderContext& m_renderer;
-	Graphics::Presenter& m_presenter;
-	FlipQueue            m_flip_queue;
-	std::jthread         m_present_thread;
+	Graphics::Presenter&     m_presenter;
+	FlipQueue                m_flip_queue;
+	std::jthread             m_present_thread;
 };
 
 static std::unique_ptr<VideoOutDriver> g_video_out_driver;
@@ -279,7 +279,7 @@ static uintptr_t VideoOutEventId(VideoOutEventKind kind) {
 }
 
 static VideoOutEventQueues& VideoOutEventQueuesFor(VideoOutEventState& state,
-                                                   VideoOutEventKind  kind) {
+                                                   VideoOutEventKind   kind) {
 	switch (kind) {
 		case VideoOutEventKind::Flip: return state.flip;
 		case VideoOutEventKind::Vblank: return state.vblank;
@@ -359,9 +359,9 @@ static void TriggerVideoOutEvents(VideoOutConfig& video_out, VideoOutEventKind k
 		if (!registration || registration->generation != video_out.generation) {
 			continue;
 		}
-		const auto result = EventQueue::KernelTriggerEvent(
-		    registration->handle, VideoOutEventId(kind), EventQueue::KERNEL_EVFILT_VIDEO_OUT,
-		    trigger_data);
+		const auto result =
+		    EventQueue::KernelTriggerEvent(registration->handle, VideoOutEventId(kind),
+		                                   EventQueue::KERNEL_EVFILT_VIDEO_OUT, trigger_data);
 		EXIT_NOT_IMPLEMENTED(result != OK && result != LibKernel::KERNEL_ERROR_EBADF &&
 		                     result != LibKernel::KERNEL_ERROR_ENOENT);
 	}
@@ -372,9 +372,8 @@ static void DeleteVideoOutEvents(const VideoOutEventQueues& queues, VideoOutEven
 		if (!registration) {
 			continue;
 		}
-		const auto result =
-		    EventQueue::KernelDeleteEvent(registration->handle, VideoOutEventId(kind),
-		                                  EventQueue::KERNEL_EVFILT_VIDEO_OUT);
+		const auto result = EventQueue::KernelDeleteEvent(
+		    registration->handle, VideoOutEventId(kind), EventQueue::KERNEL_EVFILT_VIDEO_OUT);
 		EXIT_NOT_IMPLEMENTED(result != OK && result != LibKernel::KERNEL_ERROR_EBADF &&
 		                     result != LibKernel::KERNEL_ERROR_ENOENT);
 	}
@@ -383,7 +382,7 @@ static void DeleteVideoOutEvents(const VideoOutEventQueues& queues, VideoOutEven
 static int RegisterVideoOutEvent(int handle, EventQueue::KernelEqueue eq, VideoOutEventKind kind,
                                  void* udata) {
 	uint64_t generation = 0;
-	auto*    video_out   = DriverState().Get(handle, generation);
+	auto*    video_out  = DriverState().Get(handle, generation);
 	if (video_out == nullptr) {
 		return VIDEO_OUT_ERROR_INVALID_HANDLE;
 	}
@@ -425,27 +424,25 @@ static int RegisterVideoOutEvent(int handle, EventQueue::KernelEqueue eq, VideoO
 	bool                         add_queue = false;
 	{
 		Common::LockGuard event_lock(event_state->mutex);
-		const auto existing = std::find_if(queues.begin(), queues.end(), [&](const auto& candidate) {
-			return candidate->handle == eq && candidate->generation == generation;
-		});
+		const auto        existing =
+		    std::find_if(queues.begin(), queues.end(), [&](const auto& candidate) {
+			    return candidate->handle == eq && candidate->generation == generation;
+		    });
 		if (existing != queues.end()) {
 			registration = *existing;
 		} else {
-			registration = std::make_shared<VideoOutEventRegistration>(
-			    VideoOutEventRegistration {.handle     = eq,
-			                               .state      = event_state,
-			                               .generation = generation,
-			                               .kind       = kind});
+			registration = std::make_shared<VideoOutEventRegistration>(VideoOutEventRegistration {
+			    .handle = eq, .state = event_state, .generation = generation, .kind = kind});
 			queues.push_back(registration);
 			add_queue = true;
 		}
 	}
 	event.filter.data  = registration.get();
 	event.filter.owner = registration;
-	const int result = EventQueue::KernelAddEvent(eq, event);
+	const int result   = EventQueue::KernelAddEvent(eq, event);
 	if (result != OK && add_queue) {
 		Common::LockGuard event_lock(event_state->mutex);
-		const auto added = std::find(queues.begin(), queues.end(), registration);
+		const auto        added = std::find(queues.begin(), queues.end(), registration);
 		if (added != queues.end()) {
 			queues.erase(added);
 		}
@@ -455,7 +452,7 @@ static int RegisterVideoOutEvent(int handle, EventQueue::KernelEqueue eq, VideoO
 
 static int DeleteVideoOutEvent(int handle, EventQueue::KernelEqueue eq, VideoOutEventKind kind) {
 	uint64_t generation = 0;
-	auto*    video_out   = DriverState().Get(handle, generation);
+	auto*    video_out  = DriverState().Get(handle, generation);
 	if (video_out == nullptr) {
 		return VIDEO_OUT_ERROR_INVALID_HANDLE;
 	}
@@ -814,8 +811,8 @@ void VideoOutDriver::Impl::PresentThread(std::stop_token token) {
 				m_presenter.Present(*frame, true);
 			}
 			const auto frame_end = Common::Timer::QueryPerformanceCounter();
-			total_wait += static_cast<int64_t>(period) -
-			              static_cast<int64_t>(frame_end - frame_begin);
+			total_wait +=
+			    static_cast<int64_t>(period) - static_cast<int64_t>(frame_end - frame_begin);
 			continue;
 		}
 
@@ -841,8 +838,7 @@ void VideoOutDriver::Impl::PresentThread(std::stop_token token) {
 		VblankEnd();
 
 		const auto frame_end = Common::Timer::QueryPerformanceCounter();
-		total_wait += static_cast<int64_t>(period) -
-		              static_cast<int64_t>(frame_end - frame_begin);
+		total_wait += static_cast<int64_t>(period) - static_cast<int64_t>(frame_end - frame_begin);
 	}
 }
 
@@ -1000,8 +996,8 @@ void FlipQueue::Prepare(uint64_t request_id, Graphics::CommandBuffer& buffer) {
 	}
 	Graphics::Presenter::Frame* frame = nullptr;
 	if (special) {
-		frame = &m_presenter.PrepareBlankFrame(width, height,
-		                                       index == VIDEO_OUT_BUFFER_INDEX_BLACK, &buffer);
+		frame = &m_presenter.PrepareBlankFrame(width, height, index == VIDEO_OUT_BUFFER_INDEX_BLACK,
+		                                       &buffer);
 	} else {
 		frame = &m_presenter.PrepareFrame(buffer, source_info);
 	}

src/graphics/presentation/videoOut.h

@@ -32,13 +32,13 @@ public:
 	~VideoOutDriver();
 	KYTY_CLASS_NO_COPY(VideoOutDriver);
 
-	int SubmitFlipFromGpu(Graphics::CommandBuffer& buffer, int handle, int index, int flip_mode,
-	                      int64_t flip_arg, uint64_t& request_id);
-	void                   PrepareFlip(uint64_t request_id, Graphics::CommandBuffer& buffer);
-	void                   CompleteFlip(uint64_t request_id);
-	void                   SubmitFlipPreparation(uint64_t request_id);
-	void                   WaitForSubmitSlot();
-	void                   WaitFlipDone(int handle, int index);
+	int  SubmitFlipFromGpu(Graphics::CommandBuffer& buffer, int handle, int index, int flip_mode,
+	                       int64_t flip_arg, uint64_t& request_id);
+	void PrepareFlip(uint64_t request_id, Graphics::CommandBuffer& buffer);
+	void CompleteFlip(uint64_t request_id);
+	void SubmitFlipPreparation(uint64_t request_id);
+	void WaitForSubmitSlot();
+	void WaitFlipDone(int handle, int index);
 
 	[[nodiscard]] Impl& State() noexcept;
 

src/graphics/presentation/window/swapchain.cpp

@@ -61,7 +61,7 @@ namespace Libs::Graphics {
 struct Presenter::Frame {
 	VulkanImage                    image;
 	std::unique_ptr<CommandBuffer> present_commands;
-	bool                           busy = false;
+	bool                           busy         = false;
 	bool                           reusing_last = false;
 
 	void Configure(GraphicContext& graphics, vk::Extent2D extent, vk::Format format);
@@ -155,7 +155,7 @@ public:
 			EXIT("last submitted frame is not available for reuse\n");
 		}
 		m_free.erase(free);
-		m_last_frame       = nullptr;
+		m_last_frame        = nullptr;
 		frame->busy         = true;
 		frame->reusing_last = true;
 		m_mutex.Unlock();
@@ -197,30 +197,27 @@ private:
 		}
 	}
 
-	WindowContext&                               m_window;
-	Common::Mutex                               m_mutex;
-	Common::CondVar                             m_available;
+	WindowContext&                                 m_window;
+	Common::Mutex                                  m_mutex;
+	Common::CondVar                                m_available;
 	std::vector<std::unique_ptr<Presenter::Frame>> m_frames;
 	std::deque<Presenter::Frame*>                  m_free;
 	Presenter::Frame*                              m_last_frame = nullptr;
-	vk::Format                                  m_format = vk::Format::eUndefined;
+	vk::Format                                     m_format     = vk::Format::eUndefined;
 };
 
-void Presenter::Frame::Configure(GraphicContext& graphics, vk::Extent2D extent,
-                                 vk::Format format) {
+void Presenter::Frame::Configure(GraphicContext& graphics, vk::Extent2D extent, vk::Format format) {
 	if (extent.width == 0 || extent.height == 0 || format == vk::Format::eUndefined) {
 		EXIT("unsupported prepared frame, extent=%ux%u format=%d\n", extent.width, extent.height,
 		     static_cast<int>(format));
 	}
 	const auto features = graphics.GetFormatProperties(format).optimalTilingFeatures;
-	const auto required = vk::FormatFeatureFlagBits::eBlitSrc |
-	                      vk::FormatFeatureFlagBits::eSampledImageFilterLinear |
-	                      vk::FormatFeatureFlagBits::eTransferSrc |
-	                      vk::FormatFeatureFlagBits::eTransferDst;
+	const auto required =
+	    vk::FormatFeatureFlagBits::eBlitSrc | vk::FormatFeatureFlagBits::eSampledImageFilterLinear |
+	    vk::FormatFeatureFlagBits::eTransferSrc | vk::FormatFeatureFlagBits::eTransferDst;
 	if ((features & required) != required) {
 		EXIT("prepared presentation format lacks optimal blit support: format=%d features=0x%x\n",
-		     static_cast<int>(format),
-		     static_cast<vk::FormatFeatureFlags::MaskType>(features));
+		     static_cast<int>(format), static_cast<vk::FormatFeatureFlags::MaskType>(features));
 	}
 
 	auto&      dst        = image;
@@ -234,11 +231,11 @@ void Presenter::Frame::Configure(GraphicContext& graphics, vk::Extent2D extent,
 		dst.memory = {};
 	}
 
-	dst.extent          = {extent.width, extent.height, 1};
-	dst.format          = format;
-	dst.layers          = 1;
-	dst.mip_levels      = 1;
-	dst.state           = {};
+	dst.extent     = {extent.width, extent.height, 1};
+	dst.format     = format;
+	dst.layers     = 1;
+	dst.mip_levels = 1;
+	dst.state      = {};
 	dst.subresource_states.clear();
 	dst.memory.property = vk::MemoryPropertyFlagBits::eDeviceLocal;
 
@@ -262,13 +259,12 @@ void Presenter::Frame::Configure(GraphicContext& graphics, vk::Extent2D extent,
 
 void Presenter::Frame::Transit(vk::CommandBuffer command, vk::ImageLayout layout,
                                vk::AccessFlags2 access) {
-	const auto stage = access == vk::AccessFlagBits2::eTransferRead ||
-	                           access == vk::AccessFlagBits2::eTransferWrite
-	                       ? vk::PipelineStageFlagBits2::eTransfer
-	                       : vk::PipelineStageFlagBits2::eAllCommands;
+	const auto     stage  = access == vk::AccessFlagBits2::eTransferRead ||
+	                                access == vk::AccessFlagBits2::eTransferWrite
+	                            ? vk::PipelineStageFlagBits2::eTransfer
+	                            : vk::PipelineStageFlagBits2::eAllCommands;
 	constexpr auto writes = vk::AccessFlagBits2::eTransferWrite |
-	                        vk::AccessFlagBits2::eShaderWrite |
-	                        vk::AccessFlagBits2::eMemoryWrite;
+	                        vk::AccessFlagBits2::eShaderWrite | vk::AccessFlagBits2::eMemoryWrite;
 	if (image.state.layout == layout && image.state.access_mask == access &&
 	    !static_cast<bool>(image.state.access_mask & writes)) {
 		return;
@@ -299,35 +295,27 @@ void Presenter::Frame::Transit(vk::CommandBuffer command, vk::ImageLayout layout
 void Presenter::Frame::CopyFrom(CommandBuffer& command_buffer, Image& source) {
 	command_buffer.EndRendering();
 	auto command = command_buffer.Handle();
-	source.Transit(vk::ImageLayout::eTransferSrcOptimal,
-	               vk::AccessFlagBits2::eTransferRead, {}, command);
-	Transit(command, vk::ImageLayout::eTransferDstOptimal,
-	        vk::AccessFlagBits2::eTransferWrite);
+	source.Transit(vk::ImageLayout::eTransferSrcOptimal, vk::AccessFlagBits2::eTransferRead, {},
+	               command);
+	Transit(command, vk::ImageLayout::eTransferDstOptimal, vk::AccessFlagBits2::eTransferWrite);
 	vk::ImageCopy copy {};
-	copy.srcSubresource = {vk::ImageAspectFlagBits::eColor, 0, 0,
-	                       source.backing.layers};
+	copy.srcSubresource = {vk::ImageAspectFlagBits::eColor, 0, 0, source.backing.layers};
 	copy.dstSubresource = {vk::ImageAspectFlagBits::eColor, 0, 0, image.layers};
-	copy.extent = {std::min(source.backing.extent.width, image.extent.width),
-	               std::min(source.backing.extent.height, image.extent.height), 1};
+	copy.extent         = {std::min(source.backing.extent.width, image.extent.width),
+	                       std::min(source.backing.extent.height, image.extent.height), 1};
 	EXIT_IF(copy.srcSubresource.layerCount != copy.dstSubresource.layerCount);
-	command.copyImage(source.backing.image, vk::ImageLayout::eTransferSrcOptimal,
-	                  image.image, vk::ImageLayout::eTransferDstOptimal, copy);
-	Transit(command, vk::ImageLayout::eTransferSrcOptimal,
-	        vk::AccessFlagBits2::eTransferRead);
+	command.copyImage(source.backing.image, vk::ImageLayout::eTransferSrcOptimal, image.image,
+	                  vk::ImageLayout::eTransferDstOptimal, copy);
+	Transit(command, vk::ImageLayout::eTransferSrcOptimal, vk::AccessFlagBits2::eTransferRead);
 }
 
-void Presenter::Frame::Clear(CommandBuffer& command_buffer,
-                             const vk::ClearColorValue& color) {
+void Presenter::Frame::Clear(CommandBuffer& command_buffer, const vk::ClearColorValue& color) {
 	command_buffer.EndRendering();
 	auto command = command_buffer.Handle();
-	Transit(command, vk::ImageLayout::eTransferDstOptimal,
-	        vk::AccessFlagBits2::eTransferWrite);
-	const vk::ImageSubresourceRange range {
-	    vk::ImageAspectFlagBits::eColor, 0, 1, 0, 1};
-	command.clearColorImage(image.image, vk::ImageLayout::eTransferDstOptimal, &color, 1,
-	                        &range);
-	Transit(command, vk::ImageLayout::eTransferSrcOptimal,
-	        vk::AccessFlagBits2::eTransferRead);
+	Transit(command, vk::ImageLayout::eTransferDstOptimal, vk::AccessFlagBits2::eTransferWrite);
+	const vk::ImageSubresourceRange range {vk::ImageAspectFlagBits::eColor, 0, 1, 0, 1};
+	command.clearColorImage(image.image, vk::ImageLayout::eTransferDstOptimal, &color, 1, &range);
+	Transit(command, vk::ImageLayout::eTransferSrcOptimal, vk::AccessFlagBits2::eTransferRead);
 }
 
 class Swapchain final {
@@ -338,8 +326,8 @@ public:
 	~Swapchain();
 	KYTY_CLASS_NO_COPY(Swapchain);
 
-	void Create();
-	void Recreate(bool surface_lost = false);
+	void                 Create();
+	void                 Recreate(bool surface_lost = false);
 	[[nodiscard]] Status AcquireNextImage();
 	void                 RecordPresentCommands(CommandBuffer& command, VulkanImage& source);
 	void                 Submit(CommandBuffer& command);
@@ -395,17 +383,17 @@ struct Presenter::Impl {
 		desc.view_info.usage       = vk::ImageUsageFlagBits::eTransferSrc;
 		desc.type                  = TextureCache::BindingType::VideoOut;
 
-		auto& cache = renderer.GetTextureCache();
-		auto& image = cache.GetImage(cache.FindImage(desc));
+		auto& cache           = renderer.GetTextureCache();
+		auto& image           = cache.GetImage(cache.FindImage(desc));
 		image.usage.video_out = true;
 		return image;
 	}
 
-	RenderContext& renderer;
-	WindowContext& window;
-	Swapchain      swapchain;
+	RenderContext&   renderer;
+	WindowContext&   window;
+	Swapchain        swapchain;
 	CommandScheduler present_scheduler;
-	FramePool      frames;
+	FramePool        frames;
 };
 
 void Swapchain::Create() {
@@ -441,25 +429,20 @@ void Swapchain::Create() {
 	        ? vk::CompositeAlphaFlagBitsKHR::eOpaque
 	        : vk::CompositeAlphaFlagBitsKHR::eInherit;
 
-	vk::SurfaceFormatKHR format {vk::Format::eR8G8B8A8Unorm,
-	                             vk::ColorSpaceKHR::eSrgbNonlinear};
-	if (surface.formats.size() != 1 ||
-	    surface.formats.front().format != vk::Format::eUndefined) {
+	vk::SurfaceFormatKHR format {vk::Format::eR8G8B8A8Unorm, vk::ColorSpaceKHR::eSrgbNonlinear};
+	if (surface.formats.size() != 1 || surface.formats.front().format != vk::Format::eUndefined) {
 		const auto it = std::find_if(surface.formats.begin(), surface.formats.end(),
 		                             [](const vk::SurfaceFormatKHR& candidate) {
-			                             return candidate.format ==
-			                                        vk::Format::eB8G8R8A8Unorm ||
-			                                    candidate.format ==
-			                                        vk::Format::eR8G8B8A8Unorm;
+			                             return candidate.format == vk::Format::eB8G8R8A8Unorm ||
+			                                    candidate.format == vk::Format::eR8G8B8A8Unorm;
 		                             });
 		if (it == surface.formats.end()) {
 			EXIT("no supported UNORM swapchain format\n");
 		}
 		format = *it;
 	}
-	m_format = format.format;
-	const auto swapchain_features =
-	    graphics.GetFormatProperties(m_format).optimalTilingFeatures;
+	m_format                      = format.format;
+	const auto swapchain_features = graphics.GetFormatProperties(m_format).optimalTilingFeatures;
 	if (!static_cast<bool>(swapchain_features & vk::FormatFeatureFlagBits::eBlitDst)) {
 		EXIT("swapchain format cannot be a blit destination: format=%d\n",
 		     static_cast<int>(m_format));
@@ -503,9 +486,8 @@ void Swapchain::Create() {
 		view.subresourceRange.baseMipLevel   = 0;
 		view.subresourceRange.layerCount     = 1;
 		view.subresourceRange.levelCount     = 1;
-		RequireVulkanSuccess(
-		    graphics.device.createImageView(&view, nullptr, &m_image_views[i]),
-		    "vkCreateImageView");
+		RequireVulkanSuccess(graphics.device.createImageView(&view, nullptr, &m_image_views[i]),
+		                     "vkCreateImageView");
 		EXIT_IF(m_image_views[i] == nullptr);
 	}
 
@@ -600,7 +582,7 @@ void Swapchain::Recreate(bool surface_lost) {
 
 Swapchain::Status Swapchain::AcquireNextImage() {
 	EXIT_IF(m_handle == nullptr || m_frame_index >= m_image_acquired.size());
-	m_image_index = static_cast<uint32_t>(-1);
+	m_image_index     = static_cast<uint32_t>(-1);
 	const auto result = m_window.graphic_ctx.device.acquireNextImageKHR(
 	    m_handle, std::numeric_limits<uint64_t>::max(), m_image_acquired[m_frame_index], nullptr,
 	    &m_image_index);
@@ -683,10 +665,9 @@ void Swapchain::RecordPresentCommands(CommandBuffer& command, VulkanImage& sourc
 	to_present.subresourceRange.levelCount     = 1;
 	to_present.subresourceRange.baseArrayLayer = 0;
 	to_present.subresourceRange.layerCount     = 1;
-	vk_command.pipelineBarrier(vk::PipelineStageFlagBits::eAllCommands,
-	                           vk::PipelineStageFlagBits::eAllCommands,
-	                           vk::DependencyFlagBits::eByRegion, 0,
-	                           nullptr, 0, nullptr, 1, &to_present);
+	vk_command.pipelineBarrier(
+	    vk::PipelineStageFlagBits::eAllCommands, vk::PipelineStageFlagBits::eAllCommands,
+	    vk::DependencyFlagBits::eByRegion, 0, nullptr, 0, nullptr, 1, &to_present);
 	command.End();
 }
 
@@ -700,7 +681,7 @@ void Swapchain::Submit(CommandBuffer& command) {
 
 Swapchain::Status Swapchain::Present() {
 	EXIT_IF(m_image_index >= m_render_complete.size());
-	const auto ready = m_render_complete[m_image_index];
+	const auto         ready = m_render_complete[m_image_index];
 	vk::PresentInfoKHR present {};
 	present.sType              = vk::StructureType::ePresentInfoKHR;
 	present.swapchainCount     = 1;
@@ -738,7 +719,7 @@ Presenter::~Presenter() = default;
 Presenter::Frame& Presenter::PrepareFrame(CommandBuffer& buffer, const ImageInfo& info) {
 	KYTY_PROFILER_FUNCTION();
 	EXIT_IF(buffer.IsInvalid());
-	auto* frame = m_impl->frames.Acquire();
+	auto*             frame = m_impl->frames.Acquire();
 	Common::LockGuard render_lock(m_impl->renderer.GetMutex());
 	auto&             image = m_impl->ResolveSurface(info);
 	if (image.backing.format == vk::Format::eUndefined) {
@@ -752,14 +733,13 @@ Presenter::Frame& Presenter::PrepareFrame(CommandBuffer& buffer, const ImageInfo
 		default: break;
 	}
 	frame->Configure(m_impl->window.graphic_ctx,
-	                 {image.backing.extent.width, image.backing.extent.height},
-	                 frame_format);
+	                 {image.backing.extent.width, image.backing.extent.height}, frame_format);
 	frame->CopyFrom(buffer, image);
 	return *frame;
 }
 
 Presenter::Frame& Presenter::PrepareBlankFrame(uint32_t width, uint32_t height, bool opaque,
-                                                CommandBuffer* producer) {
+                                               CommandBuffer* producer) {
 	KYTY_PROFILER_FUNCTION();
 	auto              format = m_impl->frames.GetFormat();
 	auto*             frame  = m_impl->frames.Acquire();
@@ -772,8 +752,7 @@ Presenter::Frame& Presenter::PrepareBlankFrame(uint32_t width, uint32_t height,
 		frame->Clear(*producer, clear);
 	} else {
 		if (frame->present_commands == nullptr) {
-			frame->present_commands =
-			    std::make_unique<CommandBuffer>(m_impl->present_scheduler);
+			frame->present_commands = std::make_unique<CommandBuffer>(m_impl->present_scheduler);
 		}
 		auto& command = *frame->present_commands;
 		command.WaitForFenceAndReset();
@@ -830,8 +809,7 @@ void Presenter::Present(Frame& frame, bool reuse) {
 			continue;
 		}
 		if (frame.present_commands == nullptr) {
-			frame.present_commands =
-			    std::make_unique<CommandBuffer>(m_impl->present_scheduler);
+			frame.present_commands = std::make_unique<CommandBuffer>(m_impl->present_scheduler);
 		}
 		{
 			Common::LockGuard render_lock(m_impl->renderer.GetMutex());

src/graphics/presentation/window/vulkanWindow.cpp

@@ -32,11 +32,11 @@
 #include "graphics/host_gpu/vma.h"
 #include "graphics/host_gpu/vulkanCommon.h"
 #include "graphics/presentation/presenter.h"
-#include "kernel/memory.h"
 #include "graphics/presentation/renderDoc.h"
 #include "graphics/presentation/videoOut.h"
 #include "graphics/presentation/window.h"
 #include "graphics/presentation/window/windowInternal.h"
+#include "kernel/memory.h"
 #include "libs/controller.h"
 #include "loader/systemContent.h"
 
@@ -475,9 +475,9 @@ static void VulkanInitSubgroupSizeControl(vk::PhysicalDevice physical_device,
 }
 
 static vk::Device VulkanCreateDevice(vk::PhysicalDevice physical_device, const VulkanExtensions& r,
-                                     uint32_t queue_family,
+                                     uint32_t                        queue_family,
                                      const std::vector<const char*>& device_extensions,
-                                     GraphicContext& graphics) {
+                                     GraphicContext&                 graphics) {
 	EXIT_IF(physical_device == nullptr);
 	EXIT_IF(queue_family == static_cast<uint32_t>(-1));
 
@@ -551,19 +551,19 @@ static vk::Device VulkanCreateDevice(vk::PhysicalDevice physical_device, const V
 	features12.timelineSemaphore = VK_TRUE;
 
 	vk::PhysicalDeviceFeatures device_features {};
-	device_features.fragmentStoresAndAtomics             = VK_TRUE;
-	device_features.samplerAnisotropy                    = VK_TRUE;
-	device_features.robustBufferAccess                   = VK_TRUE;
+	device_features.fragmentStoresAndAtomics = VK_TRUE;
+	device_features.samplerAnisotropy        = VK_TRUE;
+	device_features.robustBufferAccess       = VK_TRUE;
 #if !defined(__APPLE__)
-	device_features.depthBounds                          = VK_TRUE; // unsupported by MoltenVK
+	device_features.depthBounds = VK_TRUE; // unsupported by MoltenVK
 #endif
 	device_features.shaderStorageImageWriteWithoutFormat = VK_TRUE;
 	device_features.shaderStorageImageReadWithoutFormat  = VK_TRUE;
 	device_features.shaderImageGatherExtended            = VK_TRUE;
 	device_features.independentBlend                     = VK_TRUE;
 	device_features.tessellationShader                   = VK_TRUE;
-	device_features.sampleRateShading    = VK_TRUE;
-	graphics.sample_rate_shading_enabled = true;
+	device_features.sampleRateShading                    = VK_TRUE;
+	graphics.sample_rate_shading_enabled                 = true;
 	device_features.vertexPipelineStoresAndAtomics =
 	    supported_features2.features.vertexPipelineStoresAndAtomics;
 
@@ -909,10 +909,9 @@ void WindowContext::CreateVulkan() {
 	}
 	surface = native_surface;
 
-	std::vector<const char*> device_extensions = {VK_KHR_SWAPCHAIN_EXTENSION_NAME,
-	                                              VK_EXT_DEPTH_CLIP_CONTROL_EXTENSION_NAME,
-	                                              VK_KHR_PUSH_DESCRIPTOR_EXTENSION_NAME,
-	                                              "VK_KHR_maintenance1"};
+	std::vector<const char*> device_extensions = {
+	    VK_KHR_SWAPCHAIN_EXTENSION_NAME, VK_EXT_DEPTH_CLIP_CONTROL_EXTENSION_NAME,
+	    VK_KHR_PUSH_DESCRIPTOR_EXTENSION_NAME, "VK_KHR_maintenance1"};
 
 #if defined(__APPLE__)
 	// MoltenVK lacks VK_EXT_depth_clip_enable and VK_EXT_color_write_enable; the renderer
@@ -932,8 +931,8 @@ void WindowContext::CreateVulkan() {
 
 	uint32_t queue_family = static_cast<uint32_t>(-1);
 
-	VulkanFindPhysicalDevice(graphic_ctx.instance, surface, device_extensions,
-	                         surface_capabilities, graphic_ctx.physical_device, queue_family);
+	VulkanFindPhysicalDevice(graphic_ctx.instance, surface, device_extensions, surface_capabilities,
+	                         graphic_ctx.physical_device, queue_family);
 
 	if (graphic_ctx.physical_device == nullptr) {
 		EXIT("Could not find suitable device");
@@ -949,9 +948,8 @@ void WindowContext::CreateVulkan() {
 		auto available_extensions = EnumerateVulkan<vk::ExtensionProperties>(
 		    "vkEnumerateDeviceExtensionProperties",
 		    [&](uint32_t* count, vk::ExtensionProperties* values) {
-			    return graphic_ctx.physical_device.enumerateDeviceExtensionProperties(nullptr,
-			                                                                           count,
-			                                                                           values);
+			    return graphic_ctx.physical_device.enumerateDeviceExtensionProperties(
+			        nullptr, count, values);
 		    });
 
 		if (HasExtension(available_extensions, VK_EXT_MEMORY_BUDGET_EXTENSION_NAME)) {
@@ -985,7 +983,7 @@ void WindowContext::CreateVulkan() {
 
 	render_context = std::make_unique<RenderContext>(graphic_ctx);
 	LibKernel::Memory::InstallGpuResources(&render_context->GetGpuResources());
-	presenter      = std::make_unique<Presenter>(*this);
+	presenter = std::make_unique<Presenter>(*this);
 	RenderDocSetActiveWindow(graphic_ctx.instance, window);
 }
 

src/graphics/presentation/window/window.cpp

@@ -1,7 +1,5 @@
 #include "graphics/presentation/window.h"
 
-#include <cstdlib>
-
 #include "SDL.h"
 #include "SDL_error.h"
 #include "SDL_events.h"
@@ -40,6 +38,7 @@
 
 #include <algorithm>
 #include <cstdio>
+#include <cstdlib>
 #include <cstring>
 #include <memory>
 #include <string>
@@ -59,7 +58,7 @@
 
 namespace Libs::Graphics {
 
-constexpr int   KEYBOARD_CONTROLLER_ID = -1000;
+constexpr int KEYBOARD_CONTROLLER_ID = -1000;
 
 struct EventKeyboard {
 	bool     down;
@@ -251,9 +250,7 @@ static void GameEventKeyboard(WindowLoopState& game, const EventKeyboard& key) {
 	if (key.down) {
 		switch (key.key_code) {
 			case SDLK_ESCAPE: game.need_exit = true; break;
-			case SDLK_SPACE:
-				SetPause(game, !game.paused.load(std::memory_order_acquire));
-				break;
+			case SDLK_SPACE: SetPause(game, !game.paused.load(std::memory_order_acquire)); break;
 			case SDLK_F1:
 				if (!key.repeat) {
 					RenderDocRequestCapture();
@@ -390,7 +387,9 @@ void WindowContext::Resize(uint32_t new_width, uint32_t new_height) {
 void WindowContext::ProcessWindowEvent(const SDL_WindowEvent& event) {
 	const auto& window_event = event;
 	switch (window_event.event) {
-		case SDL_WINDOWEVENT_SHOWN: LOGF("Window %" PRIu32 " shown\n", window_event.windowID); break;
+		case SDL_WINDOWEVENT_SHOWN:
+			LOGF("Window %" PRIu32 " shown\n", window_event.windowID);
+			break;
 
 		case SDL_WINDOWEVENT_HIDDEN:
 			LOGF("Window %" PRIu32 " hidden\n", window_event.windowID);
@@ -401,13 +400,13 @@ void WindowContext::ProcessWindowEvent(const SDL_WindowEvent& event) {
 			break;
 
 		case SDL_WINDOWEVENT_MOVED:
-			LOGF("Window %" PRIu32 " moved to %" PRId32 ",%" PRId32 "\n",
-			     window_event.windowID, window_event.data1, window_event.data2);
+			LOGF("Window %" PRIu32 " moved to %" PRId32 ",%" PRId32 "\n", window_event.windowID,
+			     window_event.data1, window_event.data2);
 			break;
 
 		case SDL_WINDOWEVENT_RESIZED:
-			LOGF("Window %" PRIu32 " resized to %" PRId32 "x%" PRId32 "\n",
-			     window_event.windowID, window_event.data1, window_event.data2);
+			LOGF("Window %" PRIu32 " resized to %" PRId32 "x%" PRId32 "\n", window_event.windowID,
+			     window_event.data1, window_event.data2);
 
 			LOGF("m: %d\n", static_cast<int>(SDL_ThreadID()));
 			Resize(window_event.data1, window_event.data2);
@@ -807,9 +806,8 @@ static void WindowCreate(WindowContext& context) {
 		window_flags |= static_cast<uint32_t>(SDL_WINDOW_BORDERLESS);
 	}
 #endif
-	context.window =
-	    SDL_CreateWindow(KYTY_SDL_WINDOW_CAPTION, KYTY_SDL_WINDOWPOS_CENTERED,
-	                     KYTY_SDL_WINDOWPOS_CENTERED, width, height, window_flags);
+	context.window = SDL_CreateWindow(KYTY_SDL_WINDOW_CAPTION, KYTY_SDL_WINDOWPOS_CENTERED,
+	                                  KYTY_SDL_WINDOWPOS_CENTERED, width, height, window_flags);
 
 	context.window_hidden = true;
 
@@ -832,7 +830,7 @@ Presenter& WindowInit(uint32_t width, uint32_t height) {
 	WindowCreate(*window);
 	window->CreateVulkan();
 	auto& presenter = *window->presenter;
-	g_window         = std::move(window);
+	g_window        = std::move(window);
 	return presenter;
 }
 
@@ -934,9 +932,9 @@ void WindowContext::UpdateTitle() {
 	    Loader::SystemContentParamSfoGetString("TITLE_ID", title_id, sizeof(title_id));
 	static bool has_app_ver =
 	    Loader::SystemContentParamSfoGetString("APP_VER", app_ver, sizeof(app_ver));
-	static uint64_t fps_start = Common::Timer::QueryPerformanceCounter();
-	static uint64_t frame_num = 0;
-	static uint64_t fps_frames = 0;
+	static uint64_t fps_start   = Common::Timer::QueryPerformanceCounter();
+	static uint64_t frame_num   = 0;
+	static uint64_t fps_frames  = 0;
 	static double   current_fps = 0.0;
 
 	const auto now       = Common::Timer::QueryPerformanceCounter();
@@ -946,15 +944,15 @@ void WindowContext::UpdateTitle() {
 	if (now - fps_start >= frequency) {
 		current_fps = static_cast<double>(fps_frames) * static_cast<double>(frequency) /
 		              static_cast<double>(now - fps_start);
-		fps_start  = now;
-		fps_frames = 0;
+		fps_start   = now;
+		fps_frames  = 0;
 	}
 
-	auto fps = fmt::format("{}{}{}{}{}{}[{}] [{}], frame: {}, fps: {:f}", (has_title ? title : ""),
-	                       (has_title ? ", " : ""), (has_title_id ? title_id : ""),
-	                       (has_title_id ? ", " : ""), (has_app_ver ? app_ver : ""),
-	                       (has_app_ver ? " " : ""), device_name, processor_name,
-	                       frame_num, current_fps);
+	auto fps =
+	    fmt::format("{}{}{}{}{}{}[{}] [{}], frame: {}, fps: {:f}", (has_title ? title : ""),
+	                (has_title ? ", " : ""), (has_title_id ? title_id : ""),
+	                (has_title_id ? ", " : ""), (has_app_ver ? app_ver : ""),
+	                (has_app_ver ? " " : ""), device_name, processor_name, frame_num, current_fps);
 
 #if defined(__APPLE__)
 	// AppKit traps on title changes off the main thread; fire-and-forget keeps present pacing.

src/graphics/presentation/window/windowInternal.h

@@ -28,8 +28,8 @@ struct SurfaceCapabilities {
 };
 
 struct WindowLoopState {
-	SDL_Event       event {};
-	bool            need_exit = false;
+	SDL_Event        event {};
+	bool             need_exit = false;
 	std::atomic_bool paused    = false;
 };
 
@@ -38,14 +38,13 @@ struct WindowContext {
 	~WindowContext();
 	KYTY_CLASS_NO_COPY(WindowContext);
 
-	[[nodiscard]] static vk::PhysicalDeviceVulkan13Features
-	RequiredVulkan13Features() noexcept;
-	void CreateVulkan();
-	void RecreateSurface();
-	void RefreshSurfaceCapabilities();
-	void UpdateIcon();
-	void UpdateTitle();
-	void Resize(uint32_t width, uint32_t height);
+	[[nodiscard]] static vk::PhysicalDeviceVulkan13Features RequiredVulkan13Features() noexcept;
+	void                                                    CreateVulkan();
+	void                                                    RecreateSurface();
+	void                                                    RefreshSurfaceCapabilities();
+	void                                                    UpdateIcon();
+	void                                                    UpdateTitle();
+	void                                                    Resize(uint32_t width, uint32_t height);
 	void ProcessWindowEvent(const SDL_WindowEvent& event);
 	void ProcessDisplayEvent(const SDL_DisplayEvent& event);
 	void ProcessEvent(double time_seconds);
@@ -59,14 +58,14 @@ struct WindowContext {
 	void DrainMainThreadTasks();
 #endif
 
-	GraphicContext      graphic_ctx;
-	SDL_Window*         window        = nullptr;
-	bool                window_hidden = true;
-	vk::SurfaceKHR      surface       = nullptr;
-	SurfaceCapabilities surface_capabilities;
+	GraphicContext                 graphic_ctx;
+	SDL_Window*                    window        = nullptr;
+	bool                           window_hidden = true;
+	vk::SurfaceKHR                 surface       = nullptr;
+	SurfaceCapabilities            surface_capabilities;
 	std::unique_ptr<RenderContext> render_context;
-	std::unique_ptr<Presenter> presenter;
-	WindowLoopState            loop;
+	std::unique_ptr<Presenter>     presenter;
+	WindowLoopState                loop;
 
 	char device_name[VK_MAX_PHYSICAL_DEVICE_NAME_SIZE] = {0};
 	char processor_name[64]                            = {0};
@@ -76,7 +75,7 @@ struct WindowContext {
 #if defined(__APPLE__)
 	Common::Mutex                      main_task_mutex;
 	Common::CondVar                    main_task_done;
-	std::vector<std::function<void()>> main_tasks;        // guarded by main_task_mutex
+	std::vector<std::function<void()>> main_tasks;            // guarded by main_task_mutex
 	uint64_t                           main_tasks_queued = 0; // guarded by main_task_mutex
 	uint64_t                           main_tasks_run    = 0; // guarded by main_task_mutex
 #endif

src/graphics/shader/recompiler/ShaderRecompiler.cpp

@@ -2,15 +2,16 @@
 
 #include "common/assert.h"
 #include "common/logging/log.h"
+#include "graphics/shader/recompiler/cfg/ShaderCFG.h"
+#include "graphics/shader/recompiler/decompiler/ShaderDecoder.h"
+#include "graphics/shader/recompiler/emitter/SpirvEmitter.h"
 #include "graphics/shader/recompiler/ir/BindingLayout.h"
+#include "graphics/shader/recompiler/ir/ReadLaneElimination.h"
 #include "graphics/shader/recompiler/ir/ResourceMaterialization.h"
 #include "graphics/shader/recompiler/ir/ResourceTracking.h"
 #include "graphics/shader/recompiler/ir/ScalarProvenance.h"
-#include "graphics/shader/recompiler/cfg/ShaderCFG.h"
-#include "graphics/shader/recompiler/decompiler/ShaderDecoder.h"
 #include "graphics/shader/recompiler/ir/ShaderIR.h"
 #include "graphics/shader/recompiler/ir/ShaderInfoCollection.h"
-#include "graphics/shader/recompiler/emitter/SpirvEmitter.h"
 #include "graphics/shader/recompiler/ir/SrtPatcher.h"
 #include "graphics/shader/recompiler/ir/SrtWalker.h"
 
@@ -838,6 +839,11 @@ bool TryRecompile(std::span<const uint32_t> code, const CompileOptions& options,
 	if (!IR::AllocateBindings(ir, layout_options, error)) {
 		return false;
 	}
+	const auto read_lane_stats = IR::EliminateReadLane(ir);
+	if (read_lane_stats.rewritten_reads != 0) {
+		LOGF("%s read-lane elimination: reads=%" PRIu32 " shadow_writes=%" PRIu32 "\n",
+		     GetDumpLabel(options), read_lane_stats.rewritten_reads, read_lane_stats.shadow_writes);
+	}
 	std::string ir_dump;
 	if (options.dump_ir) {
 		ir_dump = MakeIrDump(cfg, ir);

src/graphics/shader/recompiler/ShaderRecompiler.h

@@ -44,7 +44,7 @@ struct CompileResult {
 };
 
 bool TryRecompile(std::span<const uint32_t> code, const CompileOptions& options,
-	              CompileResult& result, std::string* error);
+                  CompileResult& result, std::string* error);
 
 } // namespace Libs::Graphics::ShaderRecompiler
 

src/graphics/shader/recompiler/decompiler/ImageOps.cpp

@@ -35,9 +35,9 @@ constexpr ImageDimension DecodeImageDimension(uint32_t dim) {
 		case 2u: return ImageDimension::Dim3D;
 		case 3u: return ImageDimension::Dim2DArray;
 		case 4u: return ImageDimension::Dim1DArray;
-		case 5u:
-		case 7u: return ImageDimension::Dim2DArray;
-		case 6u: return ImageDimension::Dim2D;
+		case 5u: return ImageDimension::Dim2DArray;
+		case 6u: return ImageDimension::Dim2DMsaa;
+		case 7u: return ImageDimension::Dim2DMsaaArray;
 		default: return ImageDimension::Unknown;
 	}
 }
@@ -46,8 +46,10 @@ constexpr uint32_t ImageCoordComponents(ImageDimension dimension) {
 	switch (dimension) {
 		case ImageDimension::Dim1D: return 1u;
 		case ImageDimension::Dim1DArray: return 2u;
+		case ImageDimension::Dim2DMsaa:
 		case ImageDimension::Dim3D:
 		case ImageDimension::Dim2DArray: return 3u;
+		case ImageDimension::Dim2DMsaaArray: return 4u;
 		default: return 2u;
 	}
 }

src/graphics/shader/recompiler/decompiler/ScalarAluOps.cpp

@@ -196,10 +196,10 @@ bool DecodeSopk(uint32_t pc, std::span<const uint32_t> code, uint32_t word_index
 		case Opcode::SMovkI32: return DecodeScalarDestination(sdst, pc, inst.dst, error);
 		case Opcode::SWaitcnt: {
 			const uint32_t waitcnt = word & 0xffffu;
-			inst.dst.kind         = OperandKind::Null;
-			inst.src0.signed_val  = static_cast<int32_t>(waitcnt);
-			inst.src0.value       = waitcnt;
-			inst.src_count        = 1;
+			inst.dst.kind          = OperandKind::Null;
+			inst.src0.signed_val   = static_cast<int32_t>(waitcnt);
+			inst.src0.value        = waitcnt;
+			inst.src_count         = 1;
 			return true;
 		}
 		case Opcode::SSetregB32:
@@ -266,10 +266,10 @@ bool DecodeSopp(uint32_t pc, std::span<const uint32_t> code, uint32_t word_index
 	inst.src0.value      = simm;
 	inst.src0.signed_val = static_cast<int16_t>(simm);
 	inst.src_count = (inst.opcode == Opcode::SNop || inst.opcode == Opcode::SWaitcnt ||
-	                   inst.opcode == Opcode::SSleep || inst.opcode == Opcode::SSendmsg ||
-	                   inst.opcode == Opcode::STtraceData || inst.opcode == Opcode::SInstPrefetch)
-	                      ? 1
-	                      : 0;
+	                  inst.opcode == Opcode::SSleep || inst.opcode == Opcode::SSendmsg ||
+	                  inst.opcode == Opcode::STtraceData || inst.opcode == Opcode::SInstPrefetch)
+	                     ? 1
+	                     : 0;
 	inst.branch_offset = static_cast<int32_t>(static_cast<int16_t>(simm)) * 4;
 	inst.branch_target = pc + 4u + static_cast<uint32_t>(inst.branch_offset);
 	SetRawWords(inst, code, word_index, 1);

src/graphics/shader/recompiler/decompiler/ShaderDecoder.cpp

@@ -194,6 +194,8 @@ const char* ImageDimensionToString(ImageDimension dimension) {
 		case ImageDimension::Dim2D: return "2d";
 		case ImageDimension::Dim3D: return "3d";
 		case ImageDimension::Dim2DArray: return "2d_array";
+		case ImageDimension::Dim2DMsaa: return "2d_msaa";
+		case ImageDimension::Dim2DMsaaArray: return "2d_msaa_array";
 		default: return "unknown";
 	}
 }
@@ -220,9 +222,9 @@ bool DecodeScalarSource(uint32_t code, uint32_t pc, Operand& operand, std::strin
 	}
 	if (code >= 240u && code <= 247u) {
 		constexpr float values[] = {0.5f, -0.5f, 1.0f, -1.0f, 2.0f, -2.0f, 4.0f, -4.0f};
-		operand.kind            = OperandKind::FloatInlineConstant;
-		operand.float_val       = values[code - 240u];
-		operand.value           = FloatBits(operand.float_val);
+		operand.kind             = OperandKind::FloatInlineConstant;
+		operand.float_val        = values[code - 240u];
+		operand.value            = FloatBits(operand.float_val);
 		return true;
 	}
 	if (code >= 256u && code <= 511u) {
@@ -285,7 +287,7 @@ bool DecodeVectorGpr(uint32_t reg, Operand& operand, std::string* error) {
 		SetError(error, "VGPR index is out of range");
 		return false;
 	}
-	operand       = {};
+	operand      = {};
 	operand.kind = OperandKind::Vgpr;
 	operand.reg  = reg;
 	return true;

src/graphics/shader/recompiler/decompiler/ShaderDecoder.h

@@ -575,6 +575,8 @@ enum class ImageDimension : uint32_t {
 	Dim2D,
 	Dim3D,
 	Dim2DArray,
+	Dim2DMsaa,
+	Dim2DMsaaArray,
 };
 
 constexpr uint32_t MaxInstructionRawWords       = 5u;

src/graphics/shader/recompiler/emitter/SpirvEmitter.cpp

@@ -30,10 +30,16 @@ bool ImageBinding(const IR::ImageResource& image, IR::DescriptorBindingKind& kin
 				kind = integer ? Kind::SampledUint1DArray : Kind::Sampled1DArray;
 				return true;
 			case Dim::Dim2D: kind = integer ? Kind::SampledUint2D : Kind::Sampled2D; return true;
+			case Dim::Dim2DMsaa:
+				kind = integer ? Kind::SampledUint2DMsaa : Kind::Sampled2DMsaa;
+				return true;
 			case Dim::Dim3D: kind = integer ? Kind::SampledUint3D : Kind::Sampled3D; return true;
 			case Dim::Dim2DArray:
 				kind = integer ? Kind::SampledUint2DArray : Kind::Sampled2DArray;
 				return true;
+			case Dim::Dim2DMsaaArray:
+				kind = integer ? Kind::SampledUint2DMsaaArray : Kind::Sampled2DMsaaArray;
+				return true;
 			case Dim::Unknown: return false;
 		}
 	}
@@ -51,6 +57,8 @@ bool ImageBinding(const IR::ImageResource& image, IR::DescriptorBindingKind& kin
 		case Dim::Dim2DArray:
 			kind = uint_image ? Kind::StorageUint2DArray : Kind::Storage2DArray;
 			return true;
+		case Dim::Dim2DMsaa:
+		case Dim::Dim2DMsaaArray: return false;
 		case Dim::Unknown: return false;
 	}
 	return false;

src/graphics/shader/recompiler/emitter/SpirvEmitter.h

@@ -12,9 +12,9 @@ namespace Libs::Graphics::ShaderRecompiler::Spirv {
 bool ProgramRequiresExactSubgroupSize(const IR::Program& program);
 
 bool EmitProgram(const IR::Program& program, const IR::ResourceSnapshot& resources,
-	             const ShaderVertexInputInfo*  vertex_input_info,
-	             const ShaderPixelInputInfo*   pixel_input_info,
-	             const ShaderComputeInputInfo* compute_input_info, std::vector<uint32_t>& spirv,
+                 const ShaderVertexInputInfo*  vertex_input_info,
+                 const ShaderPixelInputInfo*   pixel_input_info,
+                 const ShaderComputeInputInfo* compute_input_info, std::vector<uint32_t>& spirv,
                  std::string* error);
 
 } // namespace Libs::Graphics::ShaderRecompiler::Spirv

src/graphics/shader/recompiler/emitter/spirvEmitterAnalysis.cpp

@@ -129,7 +129,7 @@ uint32_t MaxCollectedVectorRegisterEnd(const std::vector<RegisterBinding>& regis
 }
 
 void CollectMoveRelSourceRegisters(const IR::Program&            program,
-	                               std::vector<RegisterBinding>& registers) {
+                                   std::vector<RegisterBinding>& registers) {
 	const auto max_vector_end = MaxCollectedVectorRegisterEnd(registers);
 	for (const auto& block: program.blocks) {
 		for (const auto& inst: block.instructions) {
@@ -276,8 +276,7 @@ void CopyProgramInputsAndOutputs(EmitterState& state, const IR::Program& program
 		if (HasOutput(state.outputs, output.kind, output.index)) {
 			continue;
 		}
-		state.outputs.push_back(
-		    {output.kind, output.index, output.location, 0, output.debug_name});
+		state.outputs.push_back({output.kind, output.index, output.location, 0, output.debug_name});
 	}
 }
 
@@ -576,6 +575,8 @@ ImageViewKind ImageViewKindFromDimension(Decoder::ImageDimension dimension) {
 		case Decoder::ImageDimension::Dim1DArray: return ImageViewKind::Dim1DArray;
 		case Decoder::ImageDimension::Dim2DArray: return ImageViewKind::Dim2DArray;
 		case Decoder::ImageDimension::Dim3D: return ImageViewKind::Dim3D;
+		case Decoder::ImageDimension::Dim2DMsaa: return ImageViewKind::Dim2DMsaa;
+		case Decoder::ImageDimension::Dim2DMsaaArray: return ImageViewKind::Dim2DMsaaArray;
 		default: return ImageViewKind::Dim2D;
 	}
 }
@@ -601,7 +602,9 @@ uint32_t ImageViewCoordinateComponents(ImageViewKind view) {
 		case ImageViewKind::Dim1DArray:
 		case ImageViewKind::Dim2D: return 2u;
 		case ImageViewKind::Dim2DArray:
+		case ImageViewKind::Dim2DMsaaArray:
 		case ImageViewKind::Dim3D: return 3u;
+		case ImageViewKind::Dim2DMsaa: return 2u;
 		default: return 0u;
 	}
 }
@@ -611,7 +614,9 @@ uint32_t ImageViewSpatialComponents(ImageViewKind view) {
 		case ImageViewKind::Dim1D:
 		case ImageViewKind::Dim1DArray: return 1u;
 		case ImageViewKind::Dim2D:
-		case ImageViewKind::Dim2DArray: return 2u;
+		case ImageViewKind::Dim2DArray:
+		case ImageViewKind::Dim2DMsaa:
+		case ImageViewKind::Dim2DMsaaArray: return 2u;
 		case ImageViewKind::Dim3D: return 3u;
 		default: return 0u;
 	}
@@ -663,8 +668,7 @@ uint32_t LoadSampledImageDescriptor(EmitterState& state, const IR::MemoryInfo& m
 
 uint32_t LoadSamplerDescriptor(EmitterState& state, uint32_t sampler, uint32_t use_pc) {
 	(void)use_pc;
-	const auto binding =
-	    ResourceForDescriptor(state, IR::DescriptorBindingKind::Samplers, sampler);
+	const auto binding = ResourceForDescriptor(state, IR::DescriptorBindingKind::Samplers, sampler);
 	const auto pointer = DescriptorElementPointer(
 	    state, state.ptr_uniform_sampler, state.sampler_variable, binding.array_index,
 	    IR::DescriptorBindingKind::Samplers, sampler, "sampler descriptor array was not emitted");

src/graphics/shader/recompiler/emitter/spirvEmitterExport.cpp

@@ -24,7 +24,7 @@ uint32_t EmitExportVec4F32(EmitterState& state, const IR::Instruction& inst) {
 			const auto raw      = EmitValueLoad(state, inst.src[pair_index]);
 			const auto unpacked = state.builder.AllocateId();
 			state.builder.AddFunction({OpExtInst, state.vec2_float_type, unpacked,
-			                            state.glsl_std450, GlslUnpackHalf2x16, raw});
+			                           state.glsl_std450, GlslUnpackHalf2x16, raw});
 			for (uint32_t lane = 0; lane < 2u; lane++) {
 				const auto component = pair_index * 2u + lane;
 				if (((inst.export_info.en >> component) & 1u) == 0) {
@@ -36,8 +36,8 @@ uint32_t EmitExportVec4F32(EmitterState& state, const IR::Instruction& inst) {
 			}
 		}
 		const auto vec = state.builder.AllocateId();
-		state.builder.AddFunction({OpCompositeConstruct, state.vec4_float_type, vec,
-		                            components[0], components[1], components[2], components[3]});
+		state.builder.AddFunction({OpCompositeConstruct, state.vec4_float_type, vec, components[0],
+		                           components[1], components[2], components[3]});
 		return vec;
 	}
 
@@ -50,7 +50,59 @@ uint32_t EmitExportVec4F32(EmitterState& state, const IR::Instruction& inst) {
 	return vec;
 }
 
-uint32_t ApplyMrtExportMapping(EmitterState& state, const IR::Instruction& inst, uint32_t value) {
+uint32_t EmitExportComponentU32(EmitterState& state, const IR::Instruction& inst,
+                                uint32_t component) {
+	const bool enabled = ((inst.export_info.en >> component) & 1u) != 0;
+	if (!enabled || component >= inst.src_count || component >= 4u) {
+		return ConstantU32(state, component == 3u ? 1u : 0u);
+	}
+	return EmitValueLoad(state, inst.src[component]);
+}
+
+uint32_t EmitExportVec4U32(EmitterState& state, const IR::Instruction& inst) {
+	uint32_t components[4] = {
+	    ConstantU32(state, 0u),
+	    ConstantU32(state, 0u),
+	    ConstantU32(state, 0u),
+	    ConstantU32(state, 1u),
+	};
+
+	if (inst.export_info.compr) {
+		for (uint32_t pair_index = 0; pair_index < 2u && pair_index < inst.src_count;
+		     pair_index++) {
+			const auto raw = EmitValueLoad(state, inst.src[pair_index]);
+			for (uint32_t lane = 0; lane < 2u; lane++) {
+				const auto component = pair_index * 2u + lane;
+				if (((inst.export_info.en >> component) & 1u) == 0) {
+					continue;
+				}
+				components[component] = state.builder.AllocateId();
+				state.builder.AddFunction(
+				    {OpBitFieldUExtract, state.uint_type, components[component], raw,
+				     ConstantU32(state, lane * 16u), ConstantU32(state, 16u)});
+			}
+		}
+	} else {
+		for (uint32_t component = 0; component < 4u; component++) {
+			components[component] = EmitExportComponentU32(state, inst, component);
+		}
+	}
+
+	const auto vec = state.builder.AllocateId();
+	state.builder.AddFunction({OpCompositeConstruct, state.vec4_uint_type, vec, components[0],
+	                           components[1], components[2], components[3]});
+	return vec;
+}
+
+static bool MrtUsesUintOutput(const EmitterState& state, const IR::Instruction& inst) {
+	return inst.export_info.kind == IR::ExportTargetKind::Mrt &&
+	       state.pixel_input_info != nullptr &&
+	       inst.export_info.index < std::size(state.pixel_input_info->target_output_mode) &&
+	       state.pixel_input_info->target_output_mode[inst.export_info.index] == 7u;
+}
+
+uint32_t ApplyMrtExportMapping(EmitterState& state, const IR::Instruction& inst, uint32_t value,
+                               uint32_t vector_type) {
 	if (inst.export_info.kind != IR::ExportTargetKind::Mrt || state.pixel_input_info == nullptr ||
 	    inst.export_info.index >= state.pixel_input_info->target_export_mapping.size()) {
 		return value;
@@ -62,8 +114,8 @@ uint32_t ApplyMrtExportMapping(EmitterState& state, const IR::Instruction& inst,
 	}
 
 	const auto mapped = state.builder.AllocateId();
-	state.builder.AddFunction({OpVectorShuffle, state.vec4_float_type, mapped, value, value,
-	                            mapping.Map(0), mapping.Map(1), mapping.Map(2), mapping.Map(3)});
+	state.builder.AddFunction({OpVectorShuffle, vector_type, mapped, value, value, mapping.Map(0),
+	                           mapping.Map(1), mapping.Map(2), mapping.Map(3)});
 	return mapped;
 }
 
@@ -89,7 +141,7 @@ void EmitMrtZExport(EmitterState& state, const IR::Instruction& inst) {
 		const auto ptr  = state.builder.AllocateId();
 		state.builder.AddFunction({OpBitcast, state.int_type, mask, raw});
 		state.builder.AddFunction({OpAccessChain, state.ptr_output_int, ptr,
-		                            state.sample_mask_variable, ConstantU32(state, 0)});
+		                           state.sample_mask_variable, ConstantU32(state, 0)});
 		state.builder.AddFunction({OpStore, ptr, mask});
 	}
 }
@@ -114,11 +166,15 @@ void EmitExport(EmitterState& state, const IR::Instruction& inst) {
 		return;
 	}
 
-	const auto value = ApplyMrtExportMapping(state, inst, EmitExportVec4F32(state, inst));
+	const auto uint_output = MrtUsesUintOutput(state, inst);
+	const auto vector_type = uint_output ? state.vec4_uint_type : state.vec4_float_type;
+	const auto value       = ApplyMrtExportMapping(
+	    state, inst, uint_output ? EmitExportVec4U32(state, inst) : EmitExportVec4F32(state, inst),
+	    vector_type);
 	if (inst.export_info.kind == IR::ExportTargetKind::Position) {
 		const auto pointer = state.builder.AllocateId();
-		state.builder.AddFunction({OpAccessChain, state.ptr_output_vec4_float, pointer, variable,
-		                            ConstantU32(state, 0)});
+		state.builder.AddFunction(
+		    {OpAccessChain, state.ptr_output_vec4_float, pointer, variable, ConstantU32(state, 0)});
 		state.builder.AddFunction({OpStore, pointer, value});
 		return;
 	}

src/graphics/shader/recompiler/emitter/spirvEmitterFlowOps.cpp

@@ -102,7 +102,7 @@ uint32_t EmitWqmLaneU32(EmitterState& state, uint32_t src) {
 		state.builder.AddFunction(
 		    {OpINotEqual, state.bool_type, non_zero, masked, ConstantU32(state, 0)});
 		state.builder.AddFunction({OpSelect, state.uint_type, expanded, non_zero,
-		                            ConstantU32(state, mask), ConstantU32(state, 0)});
+		                           ConstantU32(state, mask), ConstantU32(state, 0)});
 		state.builder.AddFunction({OpBitwiseOr, state.uint_type, combined, ret, expanded});
 		ret = combined;
 	}
@@ -122,8 +122,8 @@ void EmitWqmB64(EmitterState& state, const IR::Instruction& inst) {
 	}
 	const auto ballot = state.builder.AllocateId();
 	state.builder.AddFunction({OpGroupNonUniformBallot, state.vec4_uint_type, ballot,
-	                            ConstantU32(state, ScopeSubgroup),
-	                            EmitLaneMaskOperandActiveBool(state, inst.src[0])});
+	                           ConstantU32(state, ScopeSubgroup),
+	                           EmitLaneMaskOperandActiveBool(state, inst.src[0])});
 	const auto low  = state.builder.AllocateId();
 	const auto high = state.builder.AllocateId();
 	state.builder.AddFunction({OpCompositeExtract, state.uint_type, low, ballot, 0});
@@ -150,8 +150,8 @@ void EmitWqmB64(EmitterState& state, const IR::Instruction& inst) {
 		EmitPerInvocationMask(state, inst.dst, active);
 	} else {
 		const auto result = state.builder.AllocateId();
-		state.builder.AddFunction({OpSelect, state.uint_type, result, active,
-		                            ConstantU32(state, 1), ConstantU32(state, 0)});
+		state.builder.AddFunction({OpSelect, state.uint_type, result, active, ConstantU32(state, 1),
+		                           ConstantU32(state, 0)});
 		EmitStoreU32(state, inst.dst, result);
 		EmitStoreU32(state, OffsetRegisterOperand(inst.dst, 1), ConstantU32(state, 0));
 	}
@@ -205,8 +205,7 @@ void EmitSaveexecB32(EmitterState& state, const IR::Instruction& inst) {
 
 	const auto cond = state.builder.AllocateId();
 	const auto scc  = state.builder.AllocateId();
-	state.builder.AddFunction(
-	    {OpINotEqual, state.bool_type, cond, new_low, ConstantU32(state, 0)});
+	state.builder.AddFunction({OpINotEqual, state.bool_type, cond, new_low, ConstantU32(state, 0)});
 	state.builder.AddFunction(
 	    {OpSelect, state.uint_type, scc, cond, ConstantU32(state, 1), ConstantU32(state, 0)});
 	EmitStoreU32(state, SccOperand(), scc);
@@ -269,18 +268,19 @@ void EmitReadFirstLaneU32(EmitterState& state, const IR::Instruction& inst) {
 	const auto first_lane  = state.builder.AllocateId();
 	const auto first_value = state.builder.AllocateId();
 	state.builder.AddFunction({OpGroupNonUniformBallot, state.vec4_uint_type, ballot,
-	                            ConstantU32(state, ScopeSubgroup), active});
+	                           ConstantU32(state, ScopeSubgroup), active});
 	state.builder.AddFunction({OpGroupNonUniformBallotFindLSB, state.uint_type, first_lane,
-	                            ConstantU32(state, ScopeSubgroup), ballot});
+	                           ConstantU32(state, ScopeSubgroup), ballot});
 	state.builder.AddFunction({OpGroupNonUniformShuffle, state.uint_type, first_value,
-	                            ConstantU32(state, ScopeSubgroup), src, first_lane});
+	                           ConstantU32(state, ScopeSubgroup), src, first_lane});
 	EmitStoreU32(state, inst.dst, first_value);
 }
 
 uint32_t EmitLaneIndex(EmitterState& state, const IR::Operand& operand) {
 	const auto lane = state.builder.AllocateId();
+	const auto mask = state.wave_size == 32u ? 31u : 63u;
 	state.builder.AddFunction({OpBitwiseAnd, state.uint_type, lane, EmitValueLoad(state, operand),
-	                            ConstantU32(state, 63)});
+	                           ConstantU32(state, mask)});
 	return lane;
 }
 
@@ -289,7 +289,7 @@ void EmitReadLaneU32(EmitterState& state, const IR::Instruction& inst) {
 	const auto lane  = EmitLaneIndex(state, inst.src[1]);
 	const auto value = state.builder.AllocateId();
 	state.builder.AddFunction({OpGroupNonUniformShuffle, state.uint_type, value,
-	                            ConstantU32(state, ScopeSubgroup), src, lane});
+	                           ConstantU32(state, ScopeSubgroup), src, lane});
 	EmitStoreU32(state, inst.dst, value);
 }
 
@@ -336,10 +336,8 @@ void EmitPermlaneB32(EmitterState& state, const IR::Instruction& inst, bool x16)
 		state.builder.AddFunction(
 		    {OpBitwiseXor, state.uint_type, row_value, row, ConstantU32(state, 16)});
 	}
-	state.builder.AddFunction(
-	    {OpBitwiseAnd, state.uint_type, lane, subid, ConstantU32(state, 15)});
-	state.builder.AddFunction(
-	    {OpBitwiseAnd, state.uint_type, lane8, lane, ConstantU32(state, 7)});
+	state.builder.AddFunction({OpBitwiseAnd, state.uint_type, lane, subid, ConstantU32(state, 15)});
+	state.builder.AddFunction({OpBitwiseAnd, state.uint_type, lane8, lane, ConstantU32(state, 7)});
 	state.builder.AddFunction(
 	    {OpShiftLeftLogical, state.uint_type, shift, lane8, ConstantU32(state, 2)});
 	state.builder.AddFunction(
@@ -350,7 +348,7 @@ void EmitPermlaneB32(EmitterState& state, const IR::Instruction& inst, bool x16)
 	    {OpBitwiseAnd, state.uint_type, index1, index0, ConstantU32(state, 15)});
 	state.builder.AddFunction({OpBitwiseOr, state.uint_type, target, row_value, index1});
 	state.builder.AddFunction({OpGroupNonUniformShuffle, state.uint_type, shuffled,
-	                            ConstantU32(state, ScopeSubgroup), value, target});
+	                           ConstantU32(state, ScopeSubgroup), value, target});
 	uint32_t ret = shuffled;
 	if (!inst.dst.op_sel) {
 		const auto source_active = EmitLaneIndexActiveBool(state, target);
@@ -375,7 +373,7 @@ void EmitBarrier(EmitterState& state, const IR::Instruction& inst) {
 	(void)inst;
 	const auto semantics = MemorySemanticsAcquireRelease | MemorySemanticsWorkgroupMemory;
 	state.builder.AddFunction({OpControlBarrier, ConstantU32(state, ScopeWorkgroup),
-	                            ConstantU32(state, ScopeWorkgroup), ConstantU32(state, semantics)});
+	                           ConstantU32(state, ScopeWorkgroup), ConstantU32(state, semantics)});
 }
 
 } // namespace Libs::Graphics::ShaderRecompiler::Spirv::Emitter

src/graphics/shader/recompiler/emitter/spirvEmitterImageOps.cpp

@@ -35,9 +35,9 @@ uint32_t ConstantImageGatherHorizontalOffsets(EmitterState& state, ImageViewKind
 uint32_t LoadStorageImageDescriptorAtIndex(EmitterState& state, uint32_t resource,
                                            uint32_t array_index, bool uint_image,
                                            ImageViewKind view) {
-	const auto kind = StorageBindingKind(uint_image, view);
+	const auto  kind        = StorageBindingKind(uint_image, view);
 	const auto& descriptors = state.storage_images[StorageImageIndex(uint_image, view)];
-	const auto pointer =
+	const auto  pointer =
 	    DescriptorElementPointer(state, descriptors.pointer_type, descriptors.variable, array_index,
 	                             kind, resource, "storage image descriptor array was not emitted");
 	const auto image = state.builder.AllocateId();
@@ -133,10 +133,18 @@ void EmitImageLoad(EmitterState& state, const IR::Instruction& inst) {
 	const bool integer = inst.memory.kind == IR::ResourceKind::ImageUint;
 
 	const auto color = state.builder.AllocateId();
-	state.builder.AddFunction({OpImageFetch, integer ? state.vec4_uint_type : state.vec4_float_type,
-	                           color, image, EmitImageLoadCoordU32(state, inst, view),
-	                           ImageOperandsLodMask,
-	                           EmitImageMipLodU32(state, inst, inst.src[0], view)});
+	const auto coord = EmitImageLoadCoordU32(state, inst, view);
+	if (ImageSpirvMultisampled(view) != 0) {
+		const auto sample = EmitImageAddressValueLoad(state, inst, inst.src[0],
+		                                              ImageViewCoordinateComponents(view));
+		state.builder.AddFunction({OpImageFetch,
+		                           integer ? state.vec4_uint_type : state.vec4_float_type, color,
+		                           image, coord, ImageOperandsSampleMask, sample});
+	} else {
+		state.builder.AddFunction(
+		    {OpImageFetch, integer ? state.vec4_uint_type : state.vec4_float_type, color, image,
+		     coord, ImageOperandsLodMask, EmitImageMipLodU32(state, inst, inst.src[0], view)});
+	}
 
 	const auto dmask     = inst.memory.dmask != 0 ? inst.memory.dmask : 1u;
 	uint32_t   dst_index = 0;
@@ -158,8 +166,8 @@ void EmitImageLoad(EmitterState& state, const IR::Instruction& inst) {
 void EmitImageStore(EmitterState& state, const IR::Instruction& inst) {
 	const auto uint_image = inst.memory.kind == IR::ResourceKind::StorageImageUint;
 	const auto view       = StorageImageViewKind(state, inst.memory, uint_image, inst.pc);
-	const auto binding    = ResourceForDescriptor(state, StorageBindingKind(uint_image, view),
-	                                              inst.memory.resource);
+	const auto binding =
+	    ResourceForDescriptor(state, StorageBindingKind(uint_image, view), inst.memory.resource);
 	const auto image = LoadStorageImageDescriptorAtIndex(state, inst.memory.resource,
 	                                                     binding.array_index, uint_image, view);
 
@@ -261,9 +269,9 @@ void EmitImageSample(EmitterState& state, const IR::Instruction& inst) {
 	} else if (integer) {
 		result_type = state.vec4_uint_type;
 	}
-	const auto explicit_lod = ImageSampleNeedsExplicitLod(state, inst);
-	const auto opcode       = ImageSampleOpcode(state, inst);
-	std::vector<uint32_t> words = {opcode, result_type, sample, sampled_image, base_coord};
+	const auto            explicit_lod = ImageSampleNeedsExplicitLod(state, inst);
+	const auto            opcode       = ImageSampleOpcode(state, inst);
+	std::vector<uint32_t> words        = {opcode, result_type, sample, sampled_image, base_coord};
 	if (dref) {
 		words.push_back(EmitImageDrefF32(state, inst, layout));
 	}

src/graphics/shader/recompiler/emitter/spirvEmitterInternal.h

@@ -99,6 +99,7 @@ enum : uint32_t {
 	ImageOperandsGradMask         = 0x00000004u,
 	ImageOperandsOffsetMask       = 0x00000010u,
 	ImageOperandsConstOffsetsMask = 0x00000020u,
+	ImageOperandsSampleMask       = 0x00000040u,
 };
 
 enum : uint32_t {
@@ -150,7 +151,6 @@ enum : uint32_t {
 	OpImageGather                  = 96,
 	OpImageDrefGather              = 97,
 	OpImageWrite                   = 99,
-	OpImage                        = 100,
 	OpImageQuerySizeLod            = 103,
 	OpImageQueryLod                = 105,
 	OpImageQueryLevels             = 106,
@@ -382,7 +382,7 @@ struct EmitterState {
 	uint32_t                                         ptr_workgroup_array       = 0;
 	uint32_t                                         ptr_workgroup_uint        = 0;
 	uint32_t                                         lds_variable              = 0;
-	std::array<SampledImageDescriptors, 10>          sampled_images;
+	std::array<SampledImageDescriptors, 14>          sampled_images;
 	std::array<StorageImageDescriptors, 10>          storage_images;
 	uint32_t                                         sampler_type                          = 0;
 	uint32_t                                         sampler_array_type                    = 0;
@@ -453,17 +453,20 @@ enum class ImageViewKind {
 	Dim2D,
 	Dim2DArray,
 	Dim3D,
+	Dim2DMsaa,
+	Dim2DMsaaArray,
 	Count,
 };
 
-constexpr uint32_t ImageViewKindCount = static_cast<uint32_t>(ImageViewKind::Count);
+constexpr uint32_t SampledImageViewKindCount = static_cast<uint32_t>(ImageViewKind::Count);
+constexpr uint32_t StorageImageViewKindCount = static_cast<uint32_t>(ImageViewKind::Dim2DMsaa);
 
 constexpr uint32_t SampledImageIndex(bool integer, ImageViewKind view) {
-	return static_cast<uint32_t>(view) + (integer ? ImageViewKindCount : 0u);
+	return static_cast<uint32_t>(view) + (integer ? SampledImageViewKindCount : 0u);
 }
 
 constexpr uint32_t StorageImageIndex(bool integer, ImageViewKind view) {
-	return static_cast<uint32_t>(view) + (integer ? ImageViewKindCount : 0u);
+	return static_cast<uint32_t>(view) + (integer ? StorageImageViewKindCount : 0u);
 }
 
 constexpr IR::DescriptorBindingKind SampledBindingKind(bool integer, ImageViewKind view) {
@@ -474,6 +477,9 @@ constexpr IR::DescriptorBindingKind SampledBindingKind(bool integer, ImageViewKi
 			case ImageViewKind::Dim2D: return IR::DescriptorBindingKind::SampledUint2D;
 			case ImageViewKind::Dim2DArray: return IR::DescriptorBindingKind::SampledUint2DArray;
 			case ImageViewKind::Dim3D: return IR::DescriptorBindingKind::SampledUint3D;
+			case ImageViewKind::Dim2DMsaa: return IR::DescriptorBindingKind::SampledUint2DMsaa;
+			case ImageViewKind::Dim2DMsaaArray:
+				return IR::DescriptorBindingKind::SampledUint2DMsaaArray;
 			default: break;
 		}
 	}
@@ -483,6 +489,8 @@ constexpr IR::DescriptorBindingKind SampledBindingKind(bool integer, ImageViewKi
 		case ImageViewKind::Dim2D: return IR::DescriptorBindingKind::Sampled2D;
 		case ImageViewKind::Dim2DArray: return IR::DescriptorBindingKind::Sampled2DArray;
 		case ImageViewKind::Dim3D: return IR::DescriptorBindingKind::Sampled3D;
+		case ImageViewKind::Dim2DMsaa: return IR::DescriptorBindingKind::Sampled2DMsaa;
+		case ImageViewKind::Dim2DMsaaArray: return IR::DescriptorBindingKind::Sampled2DMsaaArray;
 		default: break;
 	}
 	return IR::DescriptorBindingKind::Count;
@@ -516,6 +524,8 @@ constexpr uint32_t ImageSpirvDimension(ImageViewKind view) {
 		case ImageViewKind::Dim1DArray: return Dim1D;
 		case ImageViewKind::Dim2D:
 		case ImageViewKind::Dim2DArray:
+		case ImageViewKind::Dim2DMsaa:
+		case ImageViewKind::Dim2DMsaaArray:
 		case ImageViewKind::Count: return Dim2D;
 		case ImageViewKind::Dim3D: return Dim3D;
 	}
@@ -523,7 +533,14 @@ constexpr uint32_t ImageSpirvDimension(ImageViewKind view) {
 }
 
 constexpr uint32_t ImageSpirvArrayed(ImageViewKind view) {
-	return view == ImageViewKind::Dim1DArray || view == ImageViewKind::Dim2DArray ? 1u : 0u;
+	return view == ImageViewKind::Dim1DArray || view == ImageViewKind::Dim2DArray ||
+	               view == ImageViewKind::Dim2DMsaaArray
+	           ? 1u
+	           : 0u;
+}
+
+constexpr uint32_t ImageSpirvMultisampled(ImageViewKind view) {
+	return view == ImageViewKind::Dim2DMsaa || view == ImageViewKind::Dim2DMsaaArray ? 1u : 0u;
 }
 
 struct AddCarryResult {

src/graphics/shader/recompiler/emitter/spirvEmitterModule.cpp

@@ -174,6 +174,12 @@ uint32_t VertexParameterInputPointerType(const EmitterState& state, VertexInputS
 	}
 }
 
+static bool MrtUsesUintOutput(const EmitterState& state, uint32_t index) {
+	return state.stage == ShaderType::Pixel && state.pixel_input_info != nullptr &&
+	       index < std::size(state.pixel_input_info->target_output_mode) &&
+	       state.pixel_input_info->target_output_mode[index] == 7u;
+}
+
 void AllocateInputVariables(EmitterState& state) {
 	for (auto& binding: state.inputs) {
 		binding.variable_id = state.builder.AllocateId();
@@ -323,23 +329,39 @@ void AddDescriptorAnnotationsAndNames(EmitterState& state) {
 		Decorate(state.address_memory_variable, "address_memory",
 		         IR::DescriptorBindingKind::AddressMemory);
 	}
-	constexpr const char* SampledNames[] = {
-	    "sampled_1d",      "sampled_1d_array",      "sampled_2d",      "sampled_2d_array",
-	    "sampled_3d",      "sampled_uint_1d",       "sampled_uint_1d_array",
-	    "sampled_uint_2d", "sampled_uint_2d_array", "sampled_uint_3d"};
+	constexpr const char* SampledNames[] = {"sampled_1d",
+	                                        "sampled_1d_array",
+	                                        "sampled_2d",
+	                                        "sampled_2d_array",
+	                                        "sampled_3d",
+	                                        "sampled_2d_msaa",
+	                                        "sampled_2d_msaa_array",
+	                                        "sampled_uint_1d",
+	                                        "sampled_uint_1d_array",
+	                                        "sampled_uint_2d",
+	                                        "sampled_uint_2d_array",
+	                                        "sampled_uint_3d",
+	                                        "sampled_uint_2d_msaa",
+	                                        "sampled_uint_2d_msaa_array"};
 	for (uint32_t i = 0; i < state.sampled_images.size(); i++) {
-		const auto view = static_cast<ImageViewKind>(i % ImageViewKindCount);
+		const auto view = static_cast<ImageViewKind>(i % SampledImageViewKindCount);
 		Decorate(state.sampled_images[i].variable, SampledNames[i],
-		         SampledBindingKind(i >= ImageViewKindCount, view));
+		         SampledBindingKind(i >= SampledImageViewKindCount, view));
 	}
-	constexpr const char* StorageNames[] = {
-	    "storage_1d",      "storage_1d_array",      "storage_2d",      "storage_2d_array",
-	    "storage_3d",      "storage_uint_1d",       "storage_uint_1d_array",
-	    "storage_uint_2d", "storage_uint_2d_array", "storage_uint_3d"};
+	constexpr const char* StorageNames[] = {"storage_1d",
+	                                        "storage_1d_array",
+	                                        "storage_2d",
+	                                        "storage_2d_array",
+	                                        "storage_3d",
+	                                        "storage_uint_1d",
+	                                        "storage_uint_1d_array",
+	                                        "storage_uint_2d",
+	                                        "storage_uint_2d_array",
+	                                        "storage_uint_3d"};
 	for (uint32_t i = 0; i < state.storage_images.size(); i++) {
-		const auto view = static_cast<ImageViewKind>(i % ImageViewKindCount);
+		const auto view = static_cast<ImageViewKind>(i % StorageImageViewKindCount);
 		Decorate(state.storage_images[i].variable, StorageNames[i],
-		         StorageBindingKind(i >= ImageViewKindCount, view));
+		         StorageBindingKind(i >= StorageImageViewKindCount, view));
 	}
 	if (state.sampler_variable != 0) {
 		Decorate(state.sampler_variable, "samplers", IR::DescriptorBindingKind::Samplers);
@@ -409,6 +431,7 @@ void EmitHeaderAndTypes(EmitterState& state) {
 	state.ptr_output_sample_mask_array = state.builder.AllocateId();
 	state.ptr_output_float             = state.builder.AllocateId();
 	state.ptr_output_vec4_float        = state.builder.AllocateId();
+	const auto ptr_output_vec4_uint    = state.builder.AllocateId();
 	state.per_vertex_type              = state.builder.AllocateId();
 	state.ptr_output_per_vertex        = state.builder.AllocateId();
 	state.storage_runtime_array_type   = state.builder.AllocateId();
@@ -444,15 +467,15 @@ void EmitHeaderAndTypes(EmitterState& state) {
 		image.array_type         = state.builder.AllocateId();
 		image.array_pointer_type = state.builder.AllocateId();
 	}
-	state.sampler_type                                  = state.builder.AllocateId();
-	state.sampler_array_type                            = state.builder.AllocateId();
-	state.ptr_uniform_sampler                           = state.builder.AllocateId();
-	state.ptr_uniform_sampler_array                     = state.builder.AllocateId();
-	state.ptr_image_uint                                = state.builder.AllocateId();
-	state.func_type                                     = state.builder.AllocateId();
-	state.main_func                                     = state.builder.AllocateId();
-	state.entry_label                                   = state.builder.AllocateId();
-	state.glsl_std450                                   = state.builder.AllocateId();
+	state.sampler_type              = state.builder.AllocateId();
+	state.sampler_array_type        = state.builder.AllocateId();
+	state.ptr_uniform_sampler       = state.builder.AllocateId();
+	state.ptr_uniform_sampler_array = state.builder.AllocateId();
+	state.ptr_image_uint            = state.builder.AllocateId();
+	state.func_type                 = state.builder.AllocateId();
+	state.main_func                 = state.builder.AllocateId();
+	state.entry_label               = state.builder.AllocateId();
+	state.glsl_std450               = state.builder.AllocateId();
 
 	state.builder.AddCapability({CapabilityShader});
 	state.builder.AddCapability({CapabilitySampled1D});
@@ -462,7 +485,7 @@ void EmitHeaderAndTypes(EmitterState& state) {
 		state.builder.AddCapability({CapabilityImageGatherExtended});
 	}
 	if (std::any_of(state.storage_images.begin(),
-	                state.storage_images.begin() + ImageViewKindCount,
+	                state.storage_images.begin() + StorageImageViewKindCount,
 	                [](const auto& image) { return image.variable != 0; })) {
 		state.builder.AddCapability({CapabilityStorageImageReadWithoutFormat});
 		state.builder.AddCapability({CapabilityStorageImageWriteWithoutFormat});
@@ -605,6 +628,8 @@ void EmitHeaderAndTypes(EmitterState& state) {
 	    {OpTypePointer, state.ptr_output_int, StorageClassOutput, state.int_type});
 	state.builder.AddType(
 	    {OpTypePointer, state.ptr_output_vec4_float, StorageClassOutput, state.vec4_float_type});
+	state.builder.AddType(
+	    {OpTypePointer, ptr_output_vec4_uint, StorageClassOutput, state.vec4_uint_type});
 	if (state.per_vertex_variable != 0) {
 		state.builder.AddType({OpTypeStruct, state.per_vertex_type, state.vec4_float_type});
 		state.builder.AddType({OpTypePointer, state.ptr_output_per_vertex, StorageClassOutput,
@@ -615,8 +640,12 @@ void EmitHeaderAndTypes(EmitterState& state) {
 	for (const auto& binding: state.outputs) {
 		if (binding.kind == IR::StageOutputKind::Parameter ||
 		    binding.kind == IR::StageOutputKind::Mrt) {
+			const auto pointer_type =
+			    binding.kind == IR::StageOutputKind::Mrt && MrtUsesUintOutput(state, binding.index)
+			        ? ptr_output_vec4_uint
+			        : state.ptr_output_vec4_float;
 			state.builder.AddType(
-			    {OpVariable, state.ptr_output_vec4_float, binding.variable_id, StorageClassOutput});
+			    {OpVariable, pointer_type, binding.variable_id, StorageClassOutput});
 		}
 	}
 	if (state.depth_variable != 0) {
@@ -700,11 +729,11 @@ void EmitHeaderAndTypes(EmitterState& state) {
 	}
 	for (uint32_t i = 0; i < state.sampled_images.size(); i++) {
 		auto&      image     = state.sampled_images[i];
-		const auto view      = static_cast<ImageViewKind>(i % ImageViewKindCount);
-		const bool integer   = i >= ImageViewKindCount;
+		const auto view      = static_cast<ImageViewKind>(i % SampledImageViewKindCount);
+		const bool integer   = i >= SampledImageViewKindCount;
 		const auto component = integer ? state.uint_type : state.float_type;
-		state.builder.AddType({OpTypeImage, image.image_type, component,
-		                       ImageSpirvDimension(view), 0, ImageSpirvArrayed(view), 0, 1,
+		state.builder.AddType({OpTypeImage, image.image_type, component, ImageSpirvDimension(view),
+		                       0, ImageSpirvArrayed(view), ImageSpirvMultisampled(view), 1,
 		                       ImageFormatUnknown});
 		state.builder.AddType({OpTypeSampledImage, image.sampled_image_type, image.image_type});
 		state.builder.AddType(
@@ -733,13 +762,12 @@ void EmitHeaderAndTypes(EmitterState& state) {
 	}
 	for (uint32_t i = 0; i < state.storage_images.size(); i++) {
 		auto&      image     = state.storage_images[i];
-		const auto view      = static_cast<ImageViewKind>(i % ImageViewKindCount);
-		const bool integer   = i >= ImageViewKindCount;
+		const auto view      = static_cast<ImageViewKind>(i % StorageImageViewKindCount);
+		const bool integer   = i >= StorageImageViewKindCount;
 		const auto component = integer ? state.uint_type : state.float_type;
 		const auto format    = integer ? ImageFormatR32ui : ImageFormatUnknown;
-		state.builder.AddType({OpTypeImage, image.image_type, component,
-		                       ImageSpirvDimension(view), 0, ImageSpirvArrayed(view), 0, 2,
-		                       format});
+		state.builder.AddType({OpTypeImage, image.image_type, component, ImageSpirvDimension(view),
+		                       0, ImageSpirvArrayed(view), 0, 2, format});
 		state.builder.AddType(
 		    {OpTypePointer, image.pointer_type, StorageClassUniformConstant, image.image_type});
 		if (image.variable != 0) {
@@ -786,15 +814,15 @@ void AllocateDescriptorVariables(EmitterState& state) {
 		state.flattened_srt_variable = state.builder.AllocateId();
 	}
 	for (uint32_t i = 0; i < state.sampled_images.size(); i++) {
-		const auto view = static_cast<ImageViewKind>(i % ImageViewKindCount);
-		if (DescriptorBinding(state, SampledBindingKind(i >= ImageViewKindCount, view)) !=
+		const auto view = static_cast<ImageViewKind>(i % SampledImageViewKindCount);
+		if (DescriptorBinding(state, SampledBindingKind(i >= SampledImageViewKindCount, view)) !=
 		    nullptr) {
 			state.sampled_images[i].variable = state.builder.AllocateId();
 		}
 	}
 	for (uint32_t i = 0; i < state.storage_images.size(); i++) {
-		const auto view = static_cast<ImageViewKind>(i % ImageViewKindCount);
-		if (DescriptorBinding(state, StorageBindingKind(i >= ImageViewKindCount, view)) !=
+		const auto view = static_cast<ImageViewKind>(i % StorageImageViewKindCount);
+		if (DescriptorBinding(state, StorageBindingKind(i >= StorageImageViewKindCount, view)) !=
 		    nullptr) {
 			state.storage_images[i].variable = state.builder.AllocateId();
 		}

src/graphics/shader/recompiler/ir/BindingLayout.cpp

@@ -13,16 +13,30 @@ namespace {
 constexpr uint32_t MaxPushConstantBytes = 128;
 
 constexpr std::array ImageBindingKinds = {
-    DescriptorBindingKind::Sampled1D,          DescriptorBindingKind::Sampled1DArray,
-    DescriptorBindingKind::Sampled2D,          DescriptorBindingKind::Sampled2DArray,
-    DescriptorBindingKind::Sampled3D,          DescriptorBindingKind::SampledUint1D,
-    DescriptorBindingKind::SampledUint1DArray, DescriptorBindingKind::SampledUint2D,
-    DescriptorBindingKind::SampledUint2DArray, DescriptorBindingKind::SampledUint3D,
-    DescriptorBindingKind::Storage1D,          DescriptorBindingKind::Storage1DArray,
-    DescriptorBindingKind::Storage2D,          DescriptorBindingKind::Storage2DArray,
-    DescriptorBindingKind::Storage3D,          DescriptorBindingKind::StorageUint1D,
-    DescriptorBindingKind::StorageUint1DArray, DescriptorBindingKind::StorageUint2D,
-    DescriptorBindingKind::StorageUint2DArray, DescriptorBindingKind::StorageUint3D,
+    DescriptorBindingKind::Sampled1D,
+    DescriptorBindingKind::Sampled1DArray,
+    DescriptorBindingKind::Sampled2D,
+    DescriptorBindingKind::Sampled2DArray,
+    DescriptorBindingKind::Sampled2DMsaa,
+    DescriptorBindingKind::Sampled2DMsaaArray,
+    DescriptorBindingKind::Sampled3D,
+    DescriptorBindingKind::SampledUint1D,
+    DescriptorBindingKind::SampledUint1DArray,
+    DescriptorBindingKind::SampledUint2D,
+    DescriptorBindingKind::SampledUint2DArray,
+    DescriptorBindingKind::SampledUint2DMsaa,
+    DescriptorBindingKind::SampledUint2DMsaaArray,
+    DescriptorBindingKind::SampledUint3D,
+    DescriptorBindingKind::Storage1D,
+    DescriptorBindingKind::Storage1DArray,
+    DescriptorBindingKind::Storage2D,
+    DescriptorBindingKind::Storage2DArray,
+    DescriptorBindingKind::Storage3D,
+    DescriptorBindingKind::StorageUint1D,
+    DescriptorBindingKind::StorageUint1DArray,
+    DescriptorBindingKind::StorageUint2D,
+    DescriptorBindingKind::StorageUint2DArray,
+    DescriptorBindingKind::StorageUint3D,
 };
 
 bool ImageBinding(const ImageResource& image, DescriptorBindingKind& result) {
@@ -36,6 +50,8 @@ bool ImageBinding(const ImageResource& image, DescriptorBindingKind& result) {
 				case Dimension::Dim1DArray: result = Kind::Sampled1DArray; return true;
 				case Dimension::Dim2D: result = Kind::Sampled2D; return true;
 				case Dimension::Dim2DArray: result = Kind::Sampled2DArray; return true;
+				case Dimension::Dim2DMsaa: result = Kind::Sampled2DMsaa; return true;
+				case Dimension::Dim2DMsaaArray: result = Kind::Sampled2DMsaaArray; return true;
 				case Dimension::Dim3D: result = Kind::Sampled3D; return true;
 				default: return false;
 			}
@@ -45,6 +61,8 @@ bool ImageBinding(const ImageResource& image, DescriptorBindingKind& result) {
 				case Dimension::Dim1DArray: result = Kind::SampledUint1DArray; return true;
 				case Dimension::Dim2D: result = Kind::SampledUint2D; return true;
 				case Dimension::Dim2DArray: result = Kind::SampledUint2DArray; return true;
+				case Dimension::Dim2DMsaa: result = Kind::SampledUint2DMsaa; return true;
+				case Dimension::Dim2DMsaaArray: result = Kind::SampledUint2DMsaaArray; return true;
 				case Dimension::Dim3D: result = Kind::SampledUint3D; return true;
 				default: return false;
 			}
@@ -71,7 +89,7 @@ bool ImageBinding(const ImageResource& image, DescriptorBindingKind& result) {
 }
 
 bool CollectValue(const ScalarProvenance& provenance, uint32_t id, std::vector<uint8_t>& visited,
-	              std::set<uint32_t>& registers) {
+                  std::set<uint32_t>& registers) {
 	if (id <= ScalarProvenance::Unknown) {
 		return true;
 	}
@@ -104,7 +122,7 @@ bool CollectValue(const ScalarProvenance& provenance, uint32_t id, std::vector<u
 }
 
 bool CollectSource(const Program& program, uint32_t source, bool allow_unknown,
-	               std::vector<uint8_t>& visited, std::set<uint32_t>& registers) {
+                   std::vector<uint8_t>& visited, std::set<uint32_t>& registers) {
 	if (allow_unknown && source == ScalarProvenance::Unknown) {
 		return true;
 	}
@@ -165,8 +183,7 @@ bool CollectUserData(const Program& program, std::vector<uint32_t>& result) {
 				return false;
 			}
 			for (uint32_t i = 0; i < inst.src_count; i++) {
-				if (!CollectValue(program.provenance, inst.scalar_sources[i], visited,
-				                  registers)) {
+				if (!CollectValue(program.provenance, inst.scalar_sources[i], visited, registers)) {
 					return false;
 				}
 			}
@@ -199,7 +216,7 @@ bool AllocateBindings(Program& program, const BindingLayoutOptions& options, std
 	if (!program.shader_info_complete || program.binding_layout_complete) {
 		if (error != nullptr) {
 			*error = !program.shader_info_complete ? "shader info is not ready"
-			                                      : "binding layout already allocated";
+			                                       : "binding layout already allocated";
 		}
 		return false;
 	}

src/graphics/shader/recompiler/ir/ReadLaneElimination.cpp

@@ -0,0 +1,323 @@
+#include "graphics/shader/recompiler/ir/ReadLaneElimination.h"
+
+#include "graphics/shader/recompiler/ir/SrtWalker.h"
+
+#include <algorithm>
+#include <iterator>
+#include <map>
+#include <set>
+#include <utility>
+
+namespace Libs::Graphics::ShaderRecompiler::IR {
+
+namespace {
+
+constexpr uint32_t FirstTemporaryScalarRegister = 128;
+
+struct LaneKey {
+	uint32_t reg  = 0;
+	uint32_t lane = 0;
+
+	auto operator<=>(const LaneKey&) const = default;
+};
+
+using LaneSet = std::set<LaneKey>;
+
+bool PairDwordOpcode(Opcode op) {
+	switch (op) {
+		case Opcode::MoveU64:
+		case Opcode::WqmB64:
+		case Opcode::SaveexecB64:
+		case Opcode::BitwiseAndU64:
+		case Opcode::BitwiseAndNotU64:
+		case Opcode::BitwiseOrU64:
+		case Opcode::BitwiseOrNotU64:
+		case Opcode::BitwiseXorU64:
+		case Opcode::BitwiseNandU64:
+		case Opcode::BitwiseNorU64:
+		case Opcode::BitwiseXnorU64:
+		case Opcode::BitwiseNotU64:
+		case Opcode::BitFieldMaskU64:
+		case Opcode::BitFieldExtractU64:
+		case Opcode::BitReplicateB64B32:
+		case Opcode::ShiftLeftLogicalU64:
+		case Opcode::ShiftRightLogicalU64:
+		case Opcode::SelectU64: return true;
+		default: return false;
+	}
+}
+
+bool ResolveLane(const Program& program, const Instruction& inst, uint32_t source_index,
+                 uint32_t& lane) {
+	if (source_index >= inst.src_count || (program.wave_size != 32 && program.wave_size != 64)) {
+		return false;
+	}
+	const auto& selector = inst.src[source_index];
+	if (selector.kind == OperandKind::ImmediateU32) {
+		lane = selector.imm % program.wave_size;
+		return true;
+	}
+	uint32_t folded = 0;
+	if (!FoldScalarConstant(program.provenance, inst.scalar_sources[source_index], folded)) {
+		return false;
+	}
+	lane = folded % program.wave_size;
+	return true;
+}
+
+bool UniformWriteSource(const Instruction& inst) {
+	if (inst.src_count == 0) {
+		return false;
+	}
+	const auto& source = inst.src[0];
+	if (source.kind == OperandKind::ImmediateU32 || source.kind == OperandKind::PcRelativeU32) {
+		return true;
+	}
+	return source.kind == OperandKind::Register &&
+	       (source.reg.file == RegisterFile::Scalar || source.reg.file == RegisterFile::Scc ||
+	        source.reg.file == RegisterFile::M0);
+}
+
+bool WriteLaneKey(const Program& program, const Instruction& inst, LaneKey& key) {
+	if (inst.op != Opcode::WriteLaneU32 || inst.dst.kind != OperandKind::Register ||
+	    inst.dst.reg.file != RegisterFile::Vector || !UniformWriteSource(inst)) {
+		return false;
+	}
+	uint32_t lane = 0;
+	if (!ResolveLane(program, inst, 1, lane)) {
+		return false;
+	}
+	key = {inst.dst.reg.index, lane};
+	return true;
+}
+
+bool ReadLaneKey(const Program& program, const Instruction& inst, LaneKey& key) {
+	if (inst.op != Opcode::ReadLaneU32 || inst.src_count < 2 ||
+	    inst.src[0].kind != OperandKind::Register || inst.src[0].reg.file != RegisterFile::Vector) {
+		return false;
+	}
+	uint32_t lane = 0;
+	if (!ResolveLane(program, inst, 1, lane)) {
+		return false;
+	}
+	key = {inst.src[0].reg.index, lane};
+	return true;
+}
+
+void InvalidateRegister(LaneSet& valid, uint32_t reg) {
+	const auto first = valid.lower_bound({reg, 0});
+	const auto last  = valid.lower_bound({reg + 1u, 0});
+	valid.erase(first, last);
+}
+
+void ApplyInstruction(const Program& program, const Instruction& inst, LaneSet& valid) {
+	if (inst.op == Opcode::WriteLaneU32 && inst.dst.kind == OperandKind::Register &&
+	    inst.dst.reg.file == RegisterFile::Vector) {
+		LaneKey key;
+		if (WriteLaneKey(program, inst, key)) {
+			valid.insert(key);
+			return;
+		}
+		uint32_t lane = 0;
+		if (ResolveLane(program, inst, 1, lane)) {
+			valid.erase({inst.dst.reg.index, lane});
+		} else {
+			InvalidateRegister(valid, inst.dst.reg.index);
+		}
+		return;
+	}
+
+	if (inst.op == Opcode::MoveRelDestU32 && inst.dst.kind == OperandKind::Register &&
+	    inst.dst.reg.file == RegisterFile::Vector) {
+		valid.clear();
+		return;
+	}
+
+	if (inst.dst.kind == OperandKind::Register && inst.dst.reg.file == RegisterFile::Vector) {
+		uint32_t dwords = std::max(inst.memory.data_dwords, 1u);
+		if (PairDwordOpcode(inst.op) || inst.op == Opcode::UMadU64U32) {
+			dwords = std::max(dwords, 2u);
+		}
+		for (uint32_t i = 0; i < dwords && inst.dst.reg.index <= UINT32_MAX - i; i++) {
+			InvalidateRegister(valid, inst.dst.reg.index + i);
+		}
+	}
+	if (inst.dst2.kind == OperandKind::Register && inst.dst2.reg.file == RegisterFile::Vector) {
+		InvalidateRegister(valid, inst.dst2.reg.index);
+	}
+}
+
+LaneSet TransferBlock(const Program& program, const BasicBlock& block, LaneSet state) {
+	for (const auto& inst: block.instructions) {
+		ApplyInstruction(program, inst, state);
+	}
+	return state;
+}
+
+LaneSet Intersect(const LaneSet& left, const LaneSet& right) {
+	LaneSet result;
+	std::set_intersection(left.begin(), left.end(), right.begin(), right.end(),
+	                      std::inserter(result, result.end()));
+	return result;
+}
+
+uint32_t NextTemporaryScalarRegister(const Program& program) {
+	uint32_t   next     = FirstTemporaryScalarRegister;
+	const auto consider = [&next](const Operand& operand) {
+		if (operand.kind == OperandKind::Register && operand.reg.file == RegisterFile::Scalar &&
+		    operand.reg.index >= next && operand.reg.index != UINT32_MAX) {
+			next = operand.reg.index + 1u;
+		}
+	};
+	for (const auto& block: program.blocks) {
+		for (const auto& inst: block.instructions) {
+			consider(inst.dst);
+			consider(inst.dst2);
+			for (uint32_t i = 0; i < inst.src_count; i++) {
+				consider(inst.src[i]);
+			}
+		}
+	}
+	return next;
+}
+
+Operand ScalarRegisterOperand(uint32_t reg) {
+	Operand operand;
+	operand.kind      = OperandKind::Register;
+	operand.reg.file  = RegisterFile::Scalar;
+	operand.reg.index = reg;
+	return operand;
+}
+
+Instruction ShadowWrite(const Instruction& write, uint32_t temporary) {
+	Instruction shadow;
+	shadow.pc        = write.pc;
+	shadow.op        = Opcode::MoveU32;
+	shadow.dst       = ScalarRegisterOperand(temporary);
+	shadow.src[0]    = write.src[0];
+	shadow.src_count = 1;
+	return shadow;
+}
+
+Instruction ShadowRead(const Instruction& read, uint32_t temporary) {
+	Instruction rewritten;
+	rewritten.pc        = read.pc;
+	rewritten.op        = Opcode::MoveU32;
+	rewritten.dst       = read.dst;
+	rewritten.src[0]    = ScalarRegisterOperand(temporary);
+	rewritten.src_count = 1;
+	return rewritten;
+}
+
+} // namespace
+
+ReadLaneEliminationStats EliminateReadLane(Program& program) {
+	ReadLaneEliminationStats stats;
+	if (program.blocks.empty() || (program.wave_size != 32 && program.wave_size != 64)) {
+		return stats;
+	}
+
+	LaneSet universe;
+	for (const auto& block: program.blocks) {
+		for (const auto& inst: block.instructions) {
+			LaneKey key;
+			if (WriteLaneKey(program, inst, key)) {
+				universe.insert(key);
+			}
+		}
+	}
+	if (universe.empty()) {
+		return stats;
+	}
+
+	const size_t         block_count = program.blocks.size();
+	std::vector<LaneSet> entry(block_count, universe);
+	std::vector<LaneSet> exit(block_count, universe);
+	entry[0].clear();
+	for (size_t block = 0; block < block_count; block++) {
+		exit[block] = TransferBlock(program, program.blocks[block], entry[block]);
+	}
+
+	bool changed = true;
+	while (changed) {
+		changed = false;
+		for (size_t block_index = 0; block_index < block_count; block_index++) {
+			LaneSet     next_entry;
+			const auto& block = program.blocks[block_index];
+			if (block_index != 0 && !block.predecessors.empty()) {
+				next_entry = universe;
+				for (const auto predecessor: block.predecessors) {
+					if (predecessor >= block_count) {
+						next_entry.clear();
+						break;
+					}
+					next_entry = Intersect(next_entry, exit[predecessor]);
+				}
+			}
+			auto next_exit = TransferBlock(program, block, next_entry);
+			if (next_entry != entry[block_index] || next_exit != exit[block_index]) {
+				entry[block_index] = std::move(next_entry);
+				exit[block_index]  = std::move(next_exit);
+				changed            = true;
+			}
+		}
+	}
+
+	LaneSet forwarded;
+	for (size_t block_index = 0; block_index < block_count; block_index++) {
+		auto state = entry[block_index];
+		for (const auto& inst: program.blocks[block_index].instructions) {
+			LaneKey key;
+			if (ReadLaneKey(program, inst, key) && state.contains(key)) {
+				forwarded.insert(key);
+			}
+			ApplyInstruction(program, inst, state);
+		}
+	}
+	if (forwarded.empty()) {
+		return stats;
+	}
+
+	std::map<LaneKey, uint32_t> temporaries;
+	auto                        next_temporary = NextTemporaryScalarRegister(program);
+	for (const auto& key: forwarded) {
+		if (next_temporary == UINT32_MAX) {
+			return {};
+		}
+		temporaries.emplace(key, next_temporary++);
+	}
+
+	for (size_t block_index = 0; block_index < block_count; block_index++) {
+		const auto original  = std::move(program.blocks[block_index].instructions);
+		auto&      rewritten = program.blocks[block_index].instructions;
+		rewritten.clear();
+		rewritten.reserve(original.size() + temporaries.size());
+		auto state = entry[block_index];
+		for (const auto& inst: original) {
+			LaneKey read_key;
+			if (ReadLaneKey(program, inst, read_key) && state.contains(read_key)) {
+				const auto temporary = temporaries.find(read_key);
+				if (temporary != temporaries.end()) {
+					rewritten.push_back(ShadowRead(inst, temporary->second));
+					stats.rewritten_reads++;
+					ApplyInstruction(program, inst, state);
+					continue;
+				}
+			}
+
+			rewritten.push_back(inst);
+			LaneKey write_key;
+			if (WriteLaneKey(program, inst, write_key)) {
+				const auto temporary = temporaries.find(write_key);
+				if (temporary != temporaries.end()) {
+					rewritten.push_back(ShadowWrite(inst, temporary->second));
+					stats.shadow_writes++;
+				}
+			}
+			ApplyInstruction(program, inst, state);
+		}
+	}
+	return stats;
+}
+
+} // namespace Libs::Graphics::ShaderRecompiler::IR

src/graphics/shader/recompiler/ir/ReadLaneElimination.h

@@ -0,0 +1,20 @@
+#ifndef EMULATOR_INCLUDE_EMULATOR_GRAPHICS_SHADER_RECOMPILER_READLANEELIMINATION_H_
+#define EMULATOR_INCLUDE_EMULATOR_GRAPHICS_SHADER_RECOMPILER_READLANEELIMINATION_H_
+
+#include "graphics/shader/recompiler/ir/ShaderIR.h"
+
+namespace Libs::Graphics::ShaderRecompiler::IR {
+
+struct ReadLaneEliminationStats {
+	uint32_t rewritten_reads = 0;
+	uint32_t shadow_writes   = 0;
+};
+
+// Replaces fixed-lane ReadLane operations that are reached by a matching WriteLane on every
+// control-flow path. A synthetic scalar register snapshots the value at WriteLane execution time,
+// so the rewrite remains valid when the source SGPR is subsequently overwritten.
+[[nodiscard]] ReadLaneEliminationStats EliminateReadLane(Program& program);
+
+} // namespace Libs::Graphics::ShaderRecompiler::IR
+
+#endif /* EMULATOR_INCLUDE_EMULATOR_GRAPHICS_SHADER_RECOMPILER_READLANEELIMINATION_H_ */

src/graphics/shader/recompiler/ir/ResourceMaterialization.cpp

@@ -15,7 +15,8 @@ constexpr uint64_t AddressMask = 0x0000ffffffffffffull;
 Decoder::ImageDimension DescriptorDimension(const DescriptorValue&  descriptor,
                                             Decoder::ImageDimension requested) {
 	const bool is_array = requested == Decoder::ImageDimension::Dim1DArray ||
-	                      requested == Decoder::ImageDimension::Dim2DArray;
+	                      requested == Decoder::ImageDimension::Dim2DArray ||
+	                      requested == Decoder::ImageDimension::Dim2DMsaaArray;
 	switch (static_cast<Prospero::ImageType>((descriptor.dwords[3] >> 28u) & 0xfu)) {
 		case Prospero::ImageType::kColor1D: return Decoder::ImageDimension::Dim1D;
 		case Prospero::ImageType::kColor1DArray:
@@ -26,13 +27,17 @@ Decoder::ImageDimension DescriptorDimension(const DescriptorValue&  descriptor,
 		case Prospero::ImageType::kColor3D: return Decoder::ImageDimension::Dim3D;
 		case Prospero::ImageType::kCube: return Decoder::ImageDimension::Dim2DArray;
 		case Prospero::ImageType::kColor2DArray:
-		case Prospero::ImageType::kColor2DMsaaArray:
 			if (is_array) {
 				return Decoder::ImageDimension::Dim2DArray;
 			}
 			return Decoder::ImageDimension::Dim2D;
-		case Prospero::ImageType::kColor2D:
-		case Prospero::ImageType::kColor2DMsaa: return Decoder::ImageDimension::Dim2D;
+		case Prospero::ImageType::kColor2DMsaaArray:
+			if (is_array) {
+				return Decoder::ImageDimension::Dim2DMsaaArray;
+			}
+			return Decoder::ImageDimension::Dim2DMsaa;
+		case Prospero::ImageType::kColor2D: return Decoder::ImageDimension::Dim2D;
+		case Prospero::ImageType::kColor2DMsaa: return Decoder::ImageDimension::Dim2DMsaa;
 		default: return Decoder::ImageDimension::Unknown;
 	}
 }

src/graphics/shader/recompiler/ir/ResourceMaterialization.h

@@ -31,7 +31,7 @@ bool ValidateResourceSpecialization(const Program& program, const ResourceSnapsh
 // Resolves the immutable dense resource topology against one runtime user-data/SRT snapshot.
 // On failure the destination is unchanged.
 bool MaterializeResources(const Program& program, const SrtRuntime& runtime,
-	                      ResourceSnapshot& snapshot, std::string* error);
+                          ResourceSnapshot& snapshot, std::string* error);
 
 // Applies runtime descriptor shape/format facts to a copied dense topology before layout and
 // emission. On failure the program is unchanged.

src/graphics/shader/recompiler/ir/ResourceTracking.cpp

@@ -84,7 +84,7 @@ uint32_t ByteExtent(const Instruction& inst) {
 }
 
 bool ContainsUnknown(const ScalarProvenance& provenance, uint32_t id, std::vector<uint8_t>& visited,
-	                 std::vector<uint32_t>& path) {
+                     std::vector<uint32_t>& path) {
 	path.push_back(id);
 	if (id <= ScalarProvenance::Unknown || id >= provenance.values.size()) {
 		return true;
@@ -115,7 +115,7 @@ bool ContainsUnknown(const ScalarProvenance& provenance, uint32_t id, std::vecto
 }
 
 bool IsLoopInvariantValue(const ScalarProvenance& provenance, uint32_t id,
-	                      std::vector<uint8_t>& visiting) {
+                          std::vector<uint8_t>& visiting) {
 	if (id <= ScalarProvenance::Unknown || id >= provenance.values.size()) {
 		return false;
 	}

src/graphics/shader/recompiler/ir/ShaderIR.h

@@ -679,11 +679,15 @@ enum class DescriptorBindingKind {
 	Sampled1DArray,
 	Sampled2D,
 	Sampled2DArray,
+	Sampled2DMsaa,
+	Sampled2DMsaaArray,
 	Sampled3D,
 	SampledUint1D,
 	SampledUint1DArray,
 	SampledUint2D,
 	SampledUint2DArray,
+	SampledUint2DMsaa,
+	SampledUint2DMsaaArray,
 	SampledUint3D,
 	Storage1D,
 	Storage1DArray,

src/graphics/shader/recompiler/ir/SrtWalker.cpp

@@ -530,8 +530,8 @@ bool BuildSrtPlan(Program& program, std::string* error) {
 }
 
 bool EvaluateDescriptorSource(const Program& program, uint32_t source, uint32_t use_pc,
-	                          const SrtRuntime& runtime, DescriptorValue& result,
-	                          std::string* error) {
+                              const SrtRuntime& runtime, DescriptorValue& result,
+                              std::string* error) {
 	const DescriptorSourceRequest request {source, use_pc};
 	std::vector<DescriptorValue>  results;
 	if (!EvaluateDescriptorSources(program, std::span {&request, 1}, runtime, results, error)) {
@@ -542,11 +542,11 @@ bool EvaluateDescriptorSource(const Program& program, uint32_t source, uint32_t
 }
 
 static bool EvaluateRuntimeSourcesImpl(const Program&                           program,
-	                                   std::span<const DescriptorSourceRequest> requests,
-	                                   const SrtRuntime&                        runtime,
-	                                   std::vector<DescriptorValue>&            results,
-	                                   std::vector<uint32_t>& flat, bool evaluate_flat,
-	                                   std::string* error) {
+                                       std::span<const DescriptorSourceRequest> requests,
+                                       const SrtRuntime&                        runtime,
+                                       std::vector<DescriptorValue>&            results,
+                                       std::vector<uint32_t>& flat, bool evaluate_flat,
+                                       std::string* error) {
 	if (!program.srt_plan_complete) {
 		if (error != nullptr) {
 			*error = Diagnostic(program, 0, "SRT plan is not ready");
@@ -602,22 +602,22 @@ static bool EvaluateRuntimeSourcesImpl(const Program&
 }
 
 bool EvaluateDescriptorSources(const Program&                           program,
-	                           std::span<const DescriptorSourceRequest> requests,
-	                           const SrtRuntime& runtime, std::vector<DescriptorValue>& results,
-	                           std::string* error) {
+                               std::span<const DescriptorSourceRequest> requests,
+                               const SrtRuntime& runtime, std::vector<DescriptorValue>& results,
+                               std::string* error) {
 	std::vector<uint32_t> ignored;
 	return EvaluateRuntimeSourcesImpl(program, requests, runtime, results, ignored, false, error);
 }
 
 bool EvaluateRuntimeSources(const Program&                           program,
-	                        std::span<const DescriptorSourceRequest> requests,
-	                        const SrtRuntime& runtime, std::vector<DescriptorValue>& results,
-	                        std::vector<uint32_t>& flat, std::string* error) {
+                            std::span<const DescriptorSourceRequest> requests,
+                            const SrtRuntime& runtime, std::vector<DescriptorValue>& results,
+                            std::vector<uint32_t>& flat, std::string* error) {
 	return EvaluateRuntimeSourcesImpl(program, requests, runtime, results, flat, true, error);
 }
 
 bool WalkSrt(const Program& program, const SrtRuntime& runtime, std::vector<uint32_t>& flat,
-	         std::string* error) {
+             std::string* error) {
 	std::vector<DescriptorValue> ignored;
 	return EvaluateRuntimeSources(program, {}, runtime, ignored, flat, error);
 }

src/graphics/shader/recompiler/ir/SrtWalker.h

@@ -30,22 +30,22 @@ bool FoldScalarConstant(const ScalarProvenance& provenance, uint32_t value, uint
 bool BuildSrtPlan(Program& program, std::string* error);
 
 bool EvaluateDescriptorSource(const Program& program, uint32_t source, uint32_t use_pc,
-	                          const SrtRuntime& runtime, DescriptorValue& result,
+                              const SrtRuntime& runtime, DescriptorValue& result,
                               std::string* error);
 
 // Evaluates one runtime snapshot transactionally. Scalar values and ReadConst results shared by
 // several descriptors are memoized once across the batch.
 bool EvaluateDescriptorSources(const Program&                           program,
-	                           std::span<const DescriptorSourceRequest> requests,
-	                           const SrtRuntime& runtime, std::vector<DescriptorValue>& results,
+                               std::span<const DescriptorSourceRequest> requests,
+                               const SrtRuntime& runtime, std::vector<DescriptorValue>& results,
                                std::string* error);
 
 // Evaluates descriptor sources and the flattened immediate SRT with one memoized scalar walk.
 // On failure neither destination is changed.
 bool EvaluateRuntimeSources(const Program&                           program,
-	                        std::span<const DescriptorSourceRequest> requests,
-	                        const SrtRuntime& runtime, std::vector<DescriptorValue>& results,
-	                        std::vector<uint32_t>& flat, std::string* error);
+                            std::span<const DescriptorSourceRequest> requests,
+                            const SrtRuntime& runtime, std::vector<DescriptorValue>& results,
+                            std::vector<uint32_t>& flat, std::string* error);
 
 bool WalkSrt(const Program& program, const SrtRuntime& runtime, std::vector<uint32_t>& flat,
              std::string* error);

src/graphics/shader/shader.cpp

@@ -13,8 +13,8 @@
 #include "graphics/guest_gpu/graphicsRun.h"
 #include "graphics/guest_gpu/hardwareContext.h"
 #include "graphics/host_gpu/renderer/renderContext.h"
-#include "graphics/shader/recompiler/decompiler/ShaderDecoder.h"
 #include "graphics/shader/recompiler/ShaderRecompiler.h"
+#include "graphics/shader/recompiler/decompiler/ShaderDecoder.h"
 #include "graphics/shader/shaderVertexMetadata.h"
 #include "libs/errno.h"
 #include "spirv-tools/libspirv.h"
@@ -828,11 +828,10 @@ static void ShaderGetStaticInputInfoPS(
 	    vs_info.stage.program != nullptr && !vs_info.stage.program->bindings.descriptors.empty()
 	        ? 1
 	        : 0;
-	ps_info.push_constant_offset =
-	    vs_info.stage.program != nullptr
-	        ? vs_info.stage.program->bindings.push_constant_offset +
-	              vs_info.stage.program->bindings.push_constant_size
-	        : 0;
+	ps_info.push_constant_offset = vs_info.stage.program != nullptr
+	                                   ? vs_info.stage.program->bindings.push_constant_offset +
+	                                         vs_info.stage.program->bindings.push_constant_size
+	                                   : 0;
 
 	for (int i = 0; i < 8; i++) {
 		ps_info.target_output_mode[i]    = sh.target_output_mode[i];
@@ -1294,9 +1293,8 @@ static void DumpShaderRecompilerSpirv(const char* type, uint64_t shader_hash,
 
 	static std::atomic_int id = 0;
 
-	const auto base_name =
-	    Config::GetShaderLogFolder() /
-	    fmt::format("{:04d}_new_shader_{}_{:016x}", id++, type, shader_hash);
+	const auto base_name = Config::GetShaderLogFolder() /
+	                       fmt::format("{:04d}_new_shader_{}_{:016x}", id++, type, shader_hash);
 	Common::File::CreateDirectories(base_name.parent_path());
 
 	Common::File spv_file;
@@ -1345,9 +1343,8 @@ static void DumpShaderRecompilerOriginal(const char* type, uint64_t shader_hash,
 
 	static std::atomic_int id = 0;
 
-	const auto base_name =
-	    Config::GetShaderLogFolder() / "original" /
-	    fmt::format("{:04d}_new_shader_{}_{:016x}", id++, type, shader_hash);
+	const auto base_name = Config::GetShaderLogFolder() / "original" /
+	                       fmt::format("{:04d}_new_shader_{}_{:016x}", id++, type, shader_hash);
 	Common::File::CreateDirectories(base_name.parent_path());
 
 	Common::File bin_file;

src/graphics/shader/shader.h

@@ -42,8 +42,8 @@ struct ShaderStageRuntime {
 // Resolves an immutable native shader plan against current user data. The prior stage is preserved
 // if any ReadConst, snapshot, or specialization check fails.
 bool ShaderMaterializeStageRuntime(std::shared_ptr<const ShaderRecompiler::IR::Program> program,
-	                               std::span<const uint32_t> user_data, uint64_t shader_base,
-	                               ShaderStageRuntime& stage, std::string* error);
+                                   std::span<const uint32_t> user_data, uint64_t shader_base,
+                                   ShaderStageRuntime& stage, std::string* error);
 
 struct ShaderId {
 	uint32_t              hash0 = 0;

src/graphics/shader/shaderStageRuntime.cpp

@@ -6,8 +6,8 @@
 namespace Libs::Graphics {
 
 bool ShaderMaterializeStageRuntime(std::shared_ptr<const ShaderRecompiler::IR::Program> program,
-	                               std::span<const uint32_t> user_data, uint64_t shader_base,
-	                               ShaderStageRuntime& stage, std::string* error) {
+                                   std::span<const uint32_t> user_data, uint64_t shader_base,
+                                   ShaderStageRuntime& stage, std::string* error) {
 	if (program == nullptr) {
 		if (error != nullptr) {
 			*error = "missing native shader plan";

src/graphics/shader/shaderVertexMetadata.cpp

@@ -17,7 +17,7 @@ bool Fail(std::string* error, const char* message) {
 } // namespace
 
 bool ShaderReadVertexMetadata(const ShaderMappedData& data, uint32_t max_user_sgprs,
-	                          ShaderVertexMetadata& metadata, std::string* error) {
+                              ShaderVertexMetadata& metadata, std::string* error) {
 	if (data.user_data == nullptr) {
 		return Fail(error, "missing AGC user-data header");
 	}

src/graphics/shader/shaderVertexMetadata.h

@@ -17,7 +17,7 @@ struct ShaderVertexMetadata {
 
 // Copies the small AGC metadata subset used by the vertex path after validating every guest range.
 bool ShaderReadVertexMetadata(const ShaderMappedData& data, uint32_t max_user_sgprs,
-	                          ShaderVertexMetadata& metadata, std::string* error);
+                              ShaderVertexMetadata& metadata, std::string* error);
 
 } // namespace Libs::Graphics
 

src/kernel/eventQueue.cpp

@@ -33,8 +33,8 @@ static uint64_t MonotonicTimeNs() {
 }
 
 static std::unordered_map<KernelEqueue, KernelEqueueRef> g_equeues;
-static Common::Mutex                                    g_equeues_mutex;
-static uint64_t                                         g_next_equeue = 1;
+static Common::Mutex                                     g_equeues_mutex;
+static uint64_t                                          g_next_equeue = 1;
 
 class KernelEqueuePrivate {
 public:
@@ -459,8 +459,7 @@ int KYTY_SYSV_ABI KernelAddUserEvent(KernelEqueue eq, int id) {
 int KYTY_SYSV_ABI KernelAddUserEventEdge(KernelEqueue eq, int id) {
 	PRINT_NAME();
 
-	LOGF("\t user event edge add: eq = 0x%016" PRIx64 ", id = %d\n", static_cast<uint64_t>(eq),
-	     id);
+	LOGF("\t user event edge add: eq = 0x%016" PRIx64 ", id = %d\n", static_cast<uint64_t>(eq), id);
 
 	KernelEqueueEvent event {};
 	event.event.ident         = static_cast<uintptr_t>(id);
@@ -485,7 +484,7 @@ int KYTY_SYSV_ABI KernelTriggerUserEvent(KernelEqueue eq, int id, void* udata) {
 }
 
 int KYTY_SYSV_ABI KernelTriggerUserEventForAll(int id, void* udata) {
-	int                        triggered = 0;
+	int                          triggered = 0;
 	std::vector<KernelEqueueRef> queues;
 
 	{
@@ -507,8 +506,7 @@ int KYTY_SYSV_ABI KernelTriggerUserEventForAll(int id, void* udata) {
 int KYTY_SYSV_ABI KernelDeleteUserEvent(KernelEqueue eq, int id) {
 	PRINT_NAME();
 
-	LOGF("\t user event delete: eq = 0x%016" PRIx64 ", id = %d\n", static_cast<uint64_t>(eq),
-	     id);
+	LOGF("\t user event delete: eq = 0x%016" PRIx64 ", id = %d\n", static_cast<uint64_t>(eq), id);
 
 	return KernelDeleteEvent(eq, static_cast<uintptr_t>(id), KERNEL_EVFILT_USER);
 }
@@ -577,8 +575,7 @@ int KYTY_SYSV_ABI KernelAddAmprSystemEvent(KernelEqueue eq, int id, void* udata)
 int KYTY_SYSV_ABI KernelDeleteAmprEvent(KernelEqueue eq, int id) {
 	PRINT_NAME();
 
-	LOGF("\t AMPR event delete: eq = 0x%016" PRIx64 ", id = %d\n", static_cast<uint64_t>(eq),
-	     id);
+	LOGF("\t AMPR event delete: eq = 0x%016" PRIx64 ", id = %d\n", static_cast<uint64_t>(eq), id);
 
 	if (eq != KERNEL_EQUEUE_INVALID) {
 		(void)KernelDeleteEvent(eq, static_cast<uintptr_t>(id), KERNEL_EVFILT_USER);
@@ -590,8 +587,8 @@ int KYTY_SYSV_ABI KernelDeleteAmprEvent(KernelEqueue eq, int id) {
 int KYTY_SYSV_ABI KernelDeleteAmprSystemEvent(KernelEqueue eq, int id) {
 	PRINT_NAME();
 
-	LOGF("\t AMPR system event delete: eq = 0x%016" PRIx64 ", id = %d\n",
-	     static_cast<uint64_t>(eq), id);
+	LOGF("\t AMPR system event delete: eq = 0x%016" PRIx64 ", id = %d\n", static_cast<uint64_t>(eq),
+	     id);
 
 	return KernelDeleteAmprEvent(eq, id);
 }

src/kernel/eventQueue.h

@@ -41,7 +41,7 @@ struct KernelEvent {
 };
 
 struct KernelFilter {
-	void*                 data              = nullptr;
+	void*                 data = nullptr;
 	std::shared_ptr<void> owner;
 	trigger_func_t        trigger_func      = nullptr;
 	reset_func_t          reset_func        = nullptr;

src/launcher/src/configurationItem.cpp

@@ -206,8 +206,8 @@ bool ConfigurationItem::operator<(const QTreeWidgetItem& other) const {
 			       GetStatusText(other_item->m_info->game_status);
 		case GameVersionColumn:
 		case FirmwareVersionColumn: {
-			const auto& version = column == GameVersionColumn ? m_info->gameVersion
-			                                                  : m_info->firmwareVer;
+			const auto& version =
+			    column == GameVersionColumn ? m_info->gameVersion : m_info->firmwareVer;
 			const auto& other_version = column == GameVersionColumn
 			                                ? other_item->m_info->gameVersion
 			                                : other_item->m_info->firmwareVer;

src/launcher/src/configurationListWidget.cpp

@@ -1,6 +1,5 @@
 #include "configurationListWidget.h"
 
-#include "patchesDialog.h"
 #include "common.h"
 #include "compatibilityDatabase.h"
 #include "configuration.h"
@@ -8,6 +7,7 @@
 #include "configurationItem.h"
 #include "gameListTreeWidget.h"
 #include "mainDialog.h"
+#include "patchesDialog.h"
 #include "trophyViewerDialog.h"
 
 #include <QAbstractItemModel>

src/launcher/src/mainDialog.cpp

@@ -272,10 +272,18 @@ static bool FindTerminal(QString* program, QStringList* prefix) {
 	};
 
 	static const TerminalSpec candidates[] = {
-	    {"x-terminal-emulator", "-e"}, {"gnome-terminal", "--"}, {"konsole", "-e"},
-	    {"xfce4-terminal", "-x"},      {"mate-terminal", "--"},  {"tilix", "-e"},
-	    {"alacritty", "-e"},           {"kitty", nullptr},       {"foot", nullptr},
-	    {"wezterm", "-e"},             {"urxvt", "-e"},          {"xterm", "-e"},
+	    {"x-terminal-emulator", "-e"},
+	    {"gnome-terminal", "--"},
+	    {"konsole", "-e"},
+	    {"xfce4-terminal", "-x"},
+	    {"mate-terminal", "--"},
+	    {"tilix", "-e"},
+	    {"alacritty", "-e"},
+	    {"kitty", nullptr},
+	    {"foot", nullptr},
+	    {"wezterm", "-e"},
+	    {"urxvt", "-e"},
+	    {"xterm", "-e"},
 	};
 
 	const auto try_candidate = [program, prefix](const QString& executable, const char* separator) {
@@ -293,7 +301,7 @@ static bool FindTerminal(QString* program, QStringList* prefix) {
 
 	if (const auto from_env = qEnvironmentVariable("TERMINAL"); !from_env.isEmpty()) {
 		// Reuse the known separator for an explicit terminal.
-		const auto env_name  = QFileInfo(from_env).fileName();
+		const auto  env_name  = QFileInfo(from_env).fileName();
 		const char* separator = "-e";
 		for (const auto& candidate: candidates) {
 			if (env_name == QLatin1String(candidate.executable)) {
@@ -379,9 +387,9 @@ void MainDialog::RunInterpreter(QProcess* process, const Configuration& info) {
 #if !defined(_WIN32)
 	// Report immediate launch failures.
 	if (!process->waitForStarted(5000)) {
-		QMessageBox::critical(this, tr("Error"),
-		                      tr("Failed to start:\n%1\n\n%2")
-		                          .arg(process->program(), process->errorString()));
+		QMessageBox::critical(
+		    this, tr("Error"),
+		    tr("Failed to start:\n%1\n\n%2").arg(process->program(), process->errorString()));
 		return;
 	}
 #endif

src/launcher/src/patchesDialog.cpp

@@ -59,16 +59,14 @@ void PatchesDialog::Load() {
 		return;
 	}
 
-	const auto patches = QJsonDocument::fromJson(file.readAll())
-	                         .object()
-	                         .value(QStringLiteral("patches"))
-	                         .toArray();
+	const auto patches =
+	    QJsonDocument::fromJson(file.readAll()).object().value(QStringLiteral("patches")).toArray();
 	for (const auto& value: patches) {
 		const auto patch = value.toObject();
 		auto* item = new QListWidgetItem(patch.value(QStringLiteral("name")).toString(), m_patches);
 		item->setFlags(item->flags() | Qt::ItemIsUserCheckable);
 		item->setCheckState(patch.value(QStringLiteral("enabled")).toBool(true) ? Qt::Checked
-		                                                                      : Qt::Unchecked);
+		                                                                        : Qt::Unchecked);
 	}
 
 	m_apply->setEnabled(!patches.isEmpty());
@@ -84,8 +82,8 @@ void PatchesDialog::Save() {
 
 	auto document = QJsonDocument::fromJson(input.readAll());
 	input.close();
-	auto root     = document.object();
-	auto patches  = root.value(QStringLiteral("patches")).toArray();
+	auto root    = document.object();
+	auto patches = root.value(QStringLiteral("patches")).toArray();
 	for (int index = 0; index < patches.size(); index++) {
 		auto patch = patches[index].toObject();
 		patch.insert(QStringLiteral("enabled"),

src/libs/agc.cpp

@@ -52,7 +52,7 @@ KYTY_SUBSYSTEM_INIT(Graphics) {
 
 	auto& presenter = WindowInit(width, height);
 	auto& video_out = VideoOut::VideoOutInit(width, height, presenter);
-	g_renderer       = &presenter.Renderer();
+	g_renderer      = &presenter.Renderer();
 	g_renderer->InitializeGpu(&video_out);
 	ShaderInit();
 }

src/libs/libAmpr.cpp

@@ -1511,8 +1511,8 @@ static int ExecuteAprCommandBuffer(uint64_t command_buffer, int32_t* execution_r
 			} break;
 			case CommandKind::KernelEvent: {
 				const auto& command = state.kernel_event_commands[entry.index];
-				const auto eq = static_cast<LibKernel::EventQueue::KernelEqueue>(command.eq);
-				auto       result = LibKernel::EventQueue::KernelTriggerUserEvent(
+				const auto  eq      = static_cast<LibKernel::EventQueue::KernelEqueue>(command.eq);
+				auto        result  = LibKernel::EventQueue::KernelTriggerUserEvent(
 				    eq, command.id, reinterpret_cast<void*>(command.data));
 				if (result != OK) {
 					LOGF("\tAPR submit event failed: eq=0x%016" PRIx64 ", id=%" PRId32

src/libs/libKernel.cpp

@@ -1241,7 +1241,7 @@ static int KYTY_SYSV_ABI KernelRaiseException(Pthread thread, int signum) {
 		if (thread == PthreadSelfOrNull()) {
 			SignalDispatchScope scope;
 			auto                ctx = CreateCurrentGuestCallSignalUcontext(
-			                   reinterpret_cast<uint64_t>(__builtin_return_address(0)));
+			    reinterpret_cast<uint64_t>(__builtin_return_address(0)));
 			handler(signum, &ctx);
 			return OK;
 		}

src/libs/libSaveData.cpp

@@ -438,11 +438,12 @@ int KYTY_SYSV_ABI SaveDataMount3(const SaveDataMount3* mount, SaveDataMountResul
 	*mount_result = {};
 
 	Common::LockGuard lock(g_mount_mutex);
-	const std::string dir_name  = mount->dir_name->data;
-	const std::string mount_dir = std::string(SAVE_DATA_DIR) + "/" + get_title_id() + "/" + dir_name;
-	const bool        create    = ((mount->mount_mode & 4u) != 0);
-	const bool        create2   = ((mount->mount_mode & 32u) != 0);
-	const bool        open      = (!create && !create2 && ((mount->mount_mode & 3u) != 0));
+	const std::string dir_name = mount->dir_name->data;
+	const std::string mount_dir =
+	    std::string(SAVE_DATA_DIR) + "/" + get_title_id() + "/" + dir_name;
+	const bool create  = ((mount->mount_mode & 4u) != 0);
+	const bool create2 = ((mount->mount_mode & 32u) != 0);
+	const bool open    = (!create && !create2 && ((mount->mount_mode & 3u) != 0));
 
 	const int slot = g_mount_slots.FindAvailable(dir_name);
 	if (slot == SaveDataMountSlots::BUSY) {
@@ -594,9 +595,10 @@ int KYTY_SYSV_ABI SaveDataTransferringMount(const SaveDataTransferringMount* mou
 	*mount_result = {};
 
 	Common::LockGuard lock(g_mount_mutex);
-	const std::string dir_name  = mount->dir_name->data;
-	const std::string mount_dir = std::string(SAVE_DATA_DIR) + "/" + get_title_id() + "/" + dir_name;
-	const int         slot      = g_mount_slots.FindAvailable(dir_name);
+	const std::string dir_name = mount->dir_name->data;
+	const std::string mount_dir =
+	    std::string(SAVE_DATA_DIR) + "/" + get_title_id() + "/" + dir_name;
+	const int slot = g_mount_slots.FindAvailable(dir_name);
 	if (slot == SaveDataMountSlots::BUSY) {
 		return SAVE_DATA_ERROR_BUSY;
 	}

src/libs/libVideoDec2.cpp

@@ -1,10 +1,12 @@
 #include "common/abi.h"
 #include "libs/errno.h"
 #include "libs/libs.h"
+#include "libs/videoDec2Decoder.h"
 #include "loader/symbolDatabase.h"
 
 #include <cstddef>
 #include <cstdint>
+#include <cstring>
 #include <mutex>
 #include <unordered_set>
 
@@ -14,6 +16,7 @@ LIB_VERSION("Videodec2", 1, "Videodec2", 1, 1);
 
 namespace VideoDec2 {
 
+constexpr int32_t VIDEODEC2_ERROR_API_FAIL             = -2128805632; // 0x811d0100
 constexpr int32_t VIDEODEC2_ERROR_STRUCT_SIZE          = -2128805631; // 0x811d0101
 constexpr int32_t VIDEODEC2_ERROR_ARGUMENT_POINTER     = -2128805630; // 0x811d0102
 constexpr int32_t VIDEODEC2_ERROR_DECODER_INSTANCE     = -2128805629; // 0x811d0103
@@ -21,13 +24,20 @@ constexpr int32_t VIDEODEC2_ERROR_MEMORY_SIZE          = -2128805628; // 0x811d0
 constexpr int32_t VIDEODEC2_ERROR_MEMORY_POINTER       = -2128805627; // 0x811d0105
 constexpr int32_t VIDEODEC2_ERROR_FRAME_BUFFER_SIZE    = -2128805626; // 0x811d0106
 constexpr int32_t VIDEODEC2_ERROR_FRAME_BUFFER_POINTER = -2128805625; // 0x811d0107
+constexpr int32_t VIDEODEC2_ERROR_ACCESS_UNIT_SIZE     = -2128805619; // 0x811d010d
+constexpr int32_t VIDEODEC2_ERROR_ACCESS_UNIT_POINTER  = -2128805618; // 0x811d010e
+constexpr int32_t VIDEODEC2_ERROR_OUTPUT_INFO          = -2128805617; // 0x811d010f
+constexpr int32_t VIDEODEC2_ERROR_COMPUTE_QUEUE        = -2128805616; // 0x811d0110
 constexpr int32_t VIDEODEC2_ERROR_CONFIG_INFO          = -2128805376; // 0x811d0200
 constexpr int32_t VIDEODEC2_ERROR_COMPUTE_PIPE_ID      = -2128805375; // 0x811d0201
 constexpr int32_t VIDEODEC2_ERROR_COMPUTE_QUEUE_ID     = -2128805374; // 0x811d0202
 constexpr int32_t VIDEODEC2_ERROR_RESOURCE_TYPE        = -2128805373; // 0x811d0203
+constexpr int32_t VIDEODEC2_ERROR_CODEC_TYPE           = -2128805372; // 0x811d0204
 constexpr int32_t VIDEODEC2_ERROR_INPUT_QUEUE_DEPTH    = -2128805370; // 0x811d0206
 constexpr int32_t VIDEODEC2_ERROR_DPB_FRAME_COUNT      = -2128805367; // 0x811d0209
 constexpr int32_t VIDEODEC2_ERROR_FRAME_WIDTH_HEIGHT   = -2128805366; // 0x811d020a
+constexpr int32_t VIDEODEC2_ERROR_ACCESS_UNIT          = -2128805119; // 0x811d0301
+constexpr int32_t VIDEODEC2_ERROR_OVERSIZE_DECODE      = -2128805118; // 0x811d0302
 
 constexpr uint32_t VIDEODEC2_RESOURCE_TYPE_COMPUTE = 1;
 constexpr size_t   VIDEODEC2_MIN_MEMORY_SIZE       = 16ull * 1024ull * 1024ull;
@@ -101,6 +111,70 @@ struct Videodec2FrameBuffer {
 	bool   is_accepted;
 };
 
+struct Videodec2AvcPictureInfo {
+	size_t this_size;
+	bool   is_valid;
+
+	uint64_t pts_data;
+	uint64_t dts_data;
+	uint64_t attached_data;
+
+	uint8_t  idr_picture_flag;
+	uint8_t  profile_idc;
+	uint8_t  level_idc;
+	uint32_t pic_width_in_mbs_minus1;
+	uint32_t pic_height_in_map_units_minus1;
+	uint8_t  frame_mbs_only_flag;
+
+	uint8_t  frame_cropping_flag;
+	uint32_t frame_crop_left_offset;
+	uint32_t frame_crop_right_offset;
+	uint32_t frame_crop_top_offset;
+	uint32_t frame_crop_bottom_offset;
+
+	uint8_t  aspect_ratio_info_present_flag;
+	uint8_t  aspect_ratio_idc;
+	uint16_t sar_width;
+	uint16_t sar_height;
+
+	uint8_t video_signal_type_present_flag;
+	uint8_t video_format;
+	uint8_t video_full_range_flag;
+	uint8_t colour_description_present_flag;
+	uint8_t colour_primaries;
+	uint8_t transfer_characteristics;
+	uint8_t matrix_coefficients;
+
+	uint8_t  timing_info_present_flag;
+	uint32_t num_units_in_tick;
+	uint32_t time_scale;
+	uint8_t  fixed_frame_rate_flag;
+
+	uint8_t bitstream_restriction_flag;
+	uint8_t max_dec_frame_buffering;
+
+	uint8_t pic_struct_present_flag;
+	uint8_t pic_struct;
+	uint8_t field_pic_flag;
+	uint8_t bottom_field_flag;
+
+	uint8_t sequence_parameter_set_present_flag;
+	uint8_t picture_parameter_set_present_flag;
+	uint8_t au_delimiter_present_flag;
+	uint8_t end_of_sequence_present_flag;
+	uint8_t end_of_stream_present_flag;
+	uint8_t filler_data_present_flag;
+	uint8_t picture_timing_sei_present_flag;
+	uint8_t buffering_period_sei_present_flag;
+
+	uint8_t constraint_set0_flag;
+	uint8_t constraint_set1_flag;
+	uint8_t constraint_set2_flag;
+	uint8_t constraint_set3_flag;
+	uint8_t constraint_set4_flag;
+	uint8_t constraint_set5_flag;
+};
+
 struct Videodec2ComputeMemoryInfo {
 	size_t this_size;
 	size_t cpu_gpu_memory_size;
@@ -116,10 +190,7 @@ struct Videodec2ComputeConfigInfo {
 	uint16_t reserved1;
 };
 
-struct DecoderState {
-	uint64_t magic;
-	uint32_t codec_type;
-};
+using DecoderState = Decoder::Instance;
 
 static_assert(sizeof(Videodec2ComputeMemoryInfo) == 24);
 static_assert(sizeof(Videodec2ComputeConfigInfo) == 16);
@@ -128,8 +199,7 @@ static_assert(sizeof(Videodec2DecoderMemoryInfo) == 72);
 static_assert(sizeof(Videodec2InputData) == 48);
 static_assert(sizeof(Videodec2OutputInfo) == 56);
 static_assert(sizeof(Videodec2FrameBuffer) == 32);
-
-constexpr uint64_t DECODER_MAGIC = 0x4b59545956444543ull; // KYTYVDEC
+static_assert(sizeof(Videodec2AvcPictureInfo) == 120);
 
 static std::mutex                g_decoder_mutex;
 static std::unordered_set<void*> g_decoders;
@@ -156,15 +226,54 @@ static void FillNoPictureOutput(const Videodec2FrameBuffer* frame_buffer,
 	output_info->frame_height       = 0;
 	output_info->frame_buffer      = frame_buffer != nullptr ? frame_buffer->frame_buffer : nullptr;
 	output_info->frame_buffer_size = frame_buffer != nullptr ? frame_buffer->frame_buffer_size : 0;
-	output_info->frame_format      = VIDEODEC2_FRAME_FORMAT_DEFAULT;
-	output_info->frame_pitch_in_bytes = 0;
+	if (output_info->this_size == sizeof(Videodec2OutputInfo)) {
+		output_info->frame_format         = VIDEODEC2_FRAME_FORMAT_DEFAULT;
+		output_info->frame_pitch_in_bytes = 0;
+	}
 }
 
-static int32_t ValidateDecoderConfig(const Videodec2DecoderConfigInfo* config) {
+static int32_t MapDecoderResult(Decoder::Result result) {
+	switch (result) {
+		case Decoder::Result::Ok: return OK;
+		case Decoder::Result::ApiFail: return VIDEODEC2_ERROR_API_FAIL;
+		case Decoder::Result::AccessUnit: return VIDEODEC2_ERROR_ACCESS_UNIT;
+		case Decoder::Result::FrameBufferSize: return VIDEODEC2_ERROR_FRAME_BUFFER_SIZE;
+		case Decoder::Result::OversizeDecode: return VIDEODEC2_ERROR_OVERSIZE_DECODE;
+	}
+	return VIDEODEC2_ERROR_API_FAIL;
+}
+
+static void ApplyDecodedOutput(const Decoder::Output& decoded, Videodec2FrameBuffer* frame_buffer,
+                               Videodec2OutputInfo* output_info) {
+	frame_buffer->is_accepted = decoded.buffer_accepted;
+	if (!decoded.valid) {
+		return;
+	}
+	output_info->is_valid          = true;
+	output_info->is_error_frame    = decoded.error_frame;
+	output_info->picture_count     = 1;
+	output_info->codec_type        = decoded.codec_type;
+	output_info->frame_width       = decoded.width;
+	output_info->frame_pitch       = decoded.pitch;
+	output_info->frame_height      = decoded.height;
+	output_info->frame_buffer      = decoded.buffer;
+	output_info->frame_buffer_size = decoded.buffer_size;
+	if (output_info->this_size == sizeof(Videodec2OutputInfo)) {
+		output_info->frame_format         = VIDEODEC2_FRAME_FORMAT_DEFAULT;
+		output_info->frame_pitch_in_bytes = decoded.pitch;
+	}
+}
+
+static int32_t ValidateDecoderConfig(const Videodec2DecoderConfigInfo* config,
+                                     bool                              require_compute_queue) {
 	if (config->resource_type != VIDEODEC2_RESOURCE_TYPE_COMPUTE) {
 		return VIDEODEC2_ERROR_RESOURCE_TYPE;
 	}
 
+	if (!Decoder::IsCodecSupported(config->codec_type)) {
+		return VIDEODEC2_ERROR_CODEC_TYPE;
+	}
+
 	if (config->reserved0 != 0 || config->reserved1 != 0) {
 		return VIDEODEC2_ERROR_CONFIG_INFO;
 	}
@@ -182,8 +291,8 @@ static int32_t ValidateDecoderConfig(const Videodec2DecoderConfigInfo* config) {
 		return VIDEODEC2_ERROR_FRAME_WIDTH_HEIGHT;
 	}
 
-	if (config->compute_queue == nullptr) {
-		return VIDEODEC2_ERROR_CONFIG_INFO;
+	if (require_compute_queue && config->compute_queue == nullptr) {
+		return VIDEODEC2_ERROR_COMPUTE_QUEUE;
 	}
 
 	return OK;
@@ -243,7 +352,6 @@ static int32_t KYTY_SYSV_ABI AllocateComputeQueue(
 	}
 
 	*compute_queue = compute_memory_info->cpu_gpu_memory;
-
 	return OK;
 }
 
@@ -266,7 +374,7 @@ static int32_t KYTY_SYSV_ABI QueryDecoderMemoryInfo(const Videodec2DecoderConfig
 		return VIDEODEC2_ERROR_STRUCT_SIZE;
 	}
 
-	const auto validation_result = ValidateDecoderConfig(config);
+	const auto validation_result = ValidateDecoderConfig(config, false);
 	if (validation_result != OK) {
 		return validation_result;
 	}
@@ -298,7 +406,7 @@ static int32_t KYTY_SYSV_ABI CreateDecoder(const Videodec2DecoderConfigInfo* con
 		return VIDEODEC2_ERROR_STRUCT_SIZE;
 	}
 
-	const auto validation_result = ValidateDecoderConfig(config);
+	const auto validation_result = ValidateDecoderConfig(config, true);
 	if (validation_result != OK) {
 		return validation_result;
 	}
@@ -315,9 +423,11 @@ static int32_t KYTY_SYSV_ABI CreateDecoder(const Videodec2DecoderConfigInfo* con
 		return VIDEODEC2_ERROR_MEMORY_POINTER;
 	}
 
-	auto* state       = new DecoderState {};
-	state->magic      = DECODER_MAGIC;
-	state->codec_type = config->codec_type;
+	auto* state =
+	    Decoder::Create({config->codec_type, config->max_frame_width, config->max_frame_height});
+	if (state == nullptr) {
+		return VIDEODEC2_ERROR_API_FAIL;
+	}
 
 	{
 		std::scoped_lock lock(g_decoder_mutex);
@@ -325,7 +435,6 @@ static int32_t KYTY_SYSV_ABI CreateDecoder(const Videodec2DecoderConfigInfo* con
 	}
 
 	*decoder = state;
-
 	return OK;
 }
 
@@ -343,7 +452,7 @@ static int32_t KYTY_SYSV_ABI DeleteDecoder(Videodec2Decoder decoder) {
 		g_decoders.erase(it);
 	}
 
-	delete state;
+	Decoder::Destroy(state);
 
 	return OK;
 }
@@ -353,8 +462,8 @@ static int32_t KYTY_SYSV_ABI Decode(Videodec2Decoder decoder, const Videodec2Inp
                                     Videodec2OutputInfo*  output_info) {
 	PRINT_NAME();
 
-	const auto* state = GetDecoder(decoder);
-	if (state == nullptr || state->magic != DECODER_MAGIC) {
+	auto* state = GetDecoder(decoder);
+	if (state == nullptr) {
 		return VIDEODEC2_ERROR_DECODER_INSTANCE;
 	}
 
@@ -368,8 +477,12 @@ static int32_t KYTY_SYSV_ABI Decode(Videodec2Decoder decoder, const Videodec2Inp
 		return VIDEODEC2_ERROR_STRUCT_SIZE;
 	}
 
-	if (input_data->au_size != 0 && input_data->au_data == nullptr) {
-		return VIDEODEC2_ERROR_ARGUMENT_POINTER;
+	if (input_data->au_size == 0) {
+		return VIDEODEC2_ERROR_ACCESS_UNIT_SIZE;
+	}
+
+	if (input_data->au_data == nullptr) {
+		return VIDEODEC2_ERROR_ACCESS_UNIT_POINTER;
 	}
 
 	if (frame_buffer->frame_buffer_size == 0) {
@@ -381,17 +494,24 @@ static int32_t KYTY_SYSV_ABI Decode(Videodec2Decoder decoder, const Videodec2Inp
 	}
 
 	frame_buffer->is_accepted = false;
-	FillNoPictureOutput(frame_buffer, output_info, state->codec_type);
+	FillNoPictureOutput(frame_buffer, output_info, Decoder::GetCodecType(state));
 
-	return OK;
+	Decoder::Output decoded {};
+	const auto      result =
+	    Decoder::Decode(state,
+	                    {input_data->au_data, input_data->au_size, input_data->pts_data,
+	                     input_data->dts_data, input_data->attached_data},
+	                    {frame_buffer->frame_buffer, frame_buffer->frame_buffer_size}, &decoded);
+	ApplyDecodedOutput(decoded, frame_buffer, output_info);
+	return MapDecoderResult(result);
 }
 
 static int32_t KYTY_SYSV_ABI Flush(Videodec2Decoder decoder, Videodec2FrameBuffer* frame_buffer,
                                    Videodec2OutputInfo* output_info) {
 	PRINT_NAME();
 
-	const auto* state = GetDecoder(decoder);
-	if (state == nullptr || state->magic != DECODER_MAGIC) {
+	auto* state = GetDecoder(decoder);
+	if (state == nullptr) {
 		return VIDEODEC2_ERROR_DECODER_INSTANCE;
 	}
 
@@ -404,26 +524,40 @@ static int32_t KYTY_SYSV_ABI Flush(Videodec2Decoder decoder, Videodec2FrameBuffe
 		return VIDEODEC2_ERROR_STRUCT_SIZE;
 	}
 
-	frame_buffer->is_accepted = false;
-	FillNoPictureOutput(frame_buffer, output_info, state->codec_type);
+	if (frame_buffer->frame_buffer_size == 0) {
+		return VIDEODEC2_ERROR_FRAME_BUFFER_SIZE;
+	}
 
-	return OK;
+	if (frame_buffer->frame_buffer == nullptr) {
+		return VIDEODEC2_ERROR_FRAME_BUFFER_POINTER;
+	}
+
+	frame_buffer->is_accepted = false;
+	FillNoPictureOutput(frame_buffer, output_info, Decoder::GetCodecType(state));
+
+	Decoder::Output decoded {};
+	const auto      result = Decoder::Flush(
+	    state, {frame_buffer->frame_buffer, frame_buffer->frame_buffer_size}, &decoded);
+	ApplyDecodedOutput(decoded, frame_buffer, output_info);
+	return MapDecoderResult(result);
 }
 
 static int32_t KYTY_SYSV_ABI Reset(Videodec2Decoder decoder) {
 	PRINT_NAME();
 
-	const auto* state = GetDecoder(decoder);
-	return state != nullptr && state->magic == DECODER_MAGIC ? OK
-	                                                         : VIDEODEC2_ERROR_DECODER_INSTANCE;
+	auto* state = GetDecoder(decoder);
+	if (state == nullptr) {
+		return VIDEODEC2_ERROR_DECODER_INSTANCE;
+	}
+	Decoder::Reset(state);
+	return OK;
 }
 
 static int32_t KYTY_SYSV_ABI GetPictureInfo(const Videodec2OutputInfo* output_info,
-                                            void* /*first_picture_info*/,
-                                            void* /*second_picture_info*/) {
+                                            void* first_picture_info, void* second_picture_info) {
 	PRINT_NAME();
 
-	if (output_info == nullptr) {
+	if (output_info == nullptr || first_picture_info == nullptr) {
 		return VIDEODEC2_ERROR_ARGUMENT_POINTER;
 	}
 
@@ -431,6 +565,88 @@ static int32_t KYTY_SYSV_ABI GetPictureInfo(const Videodec2OutputInfo* output_in
 		return VIDEODEC2_ERROR_STRUCT_SIZE;
 	}
 
+	if (!output_info->is_valid || output_info->picture_count == 0 ||
+	    output_info->frame_buffer == nullptr) {
+		return VIDEODEC2_ERROR_OUTPUT_INFO;
+	}
+
+	Decoder::PictureInfo decoded {};
+	if (!Decoder::GetPictureInfo(output_info->frame_buffer, &decoded) ||
+	    decoded.codec_type != output_info->codec_type) {
+		return VIDEODEC2_ERROR_OUTPUT_INFO;
+	}
+
+	auto fill_common = [&decoded](void* destination, bool valid) -> int32_t {
+		auto*      bytes = static_cast<uint8_t*>(destination);
+		const auto size  = *static_cast<const size_t*>(destination);
+		if (size < 40 || size > 256) {
+			return VIDEODEC2_ERROR_STRUCT_SIZE;
+		}
+		std::memset(bytes + sizeof(size_t), 0, size - sizeof(size_t));
+		bytes[8] = valid ? 1 : 0;
+		if (valid) {
+			std::memcpy(bytes + 16, &decoded.pts, sizeof(decoded.pts));
+			std::memcpy(bytes + 24, &decoded.dts, sizeof(decoded.dts));
+			std::memcpy(bytes + 32, &decoded.attached_data, sizeof(decoded.attached_data));
+		}
+		return OK;
+	};
+
+	if (output_info->codec_type == 1) {
+		const auto requested_size = *static_cast<const size_t*>(first_picture_info);
+		if (requested_size != sizeof(Videodec2AvcPictureInfo) &&
+		    (requested_size | 16u) != sizeof(Videodec2AvcPictureInfo)) {
+			return VIDEODEC2_ERROR_STRUCT_SIZE;
+		}
+
+		Videodec2AvcPictureInfo picture {};
+		picture.this_size                      = requested_size;
+		picture.is_valid                       = true;
+		picture.pts_data                       = decoded.pts;
+		picture.dts_data                       = decoded.dts;
+		picture.attached_data                  = decoded.attached_data;
+		picture.idr_picture_flag               = decoded.key_frame ? 1 : 0;
+		picture.profile_idc                    = static_cast<uint8_t>(decoded.profile);
+		picture.level_idc                      = static_cast<uint8_t>(decoded.level);
+		picture.pic_width_in_mbs_minus1        = (decoded.width + 15u) / 16u - 1u;
+		picture.pic_height_in_map_units_minus1 = (decoded.height + 15u) / 16u - 1u;
+		picture.frame_mbs_only_flag            = 1;
+		picture.frame_cropping_flag      = decoded.crop_left != 0 || decoded.crop_right != 0 ||
+		                                           decoded.crop_top != 0 || decoded.crop_bottom != 0
+		                                       ? 1
+		                                       : 0;
+		picture.frame_crop_left_offset   = decoded.crop_left;
+		picture.frame_crop_right_offset  = decoded.crop_right;
+		picture.frame_crop_top_offset    = decoded.crop_top;
+		picture.frame_crop_bottom_offset = decoded.crop_bottom;
+		picture.aspect_ratio_info_present_flag =
+		    decoded.sar_width != 0 && decoded.sar_height != 0 ? 1 : 0;
+		picture.aspect_ratio_idc               = picture.aspect_ratio_info_present_flag ? 255 : 0;
+		picture.sar_width                      = decoded.sar_width;
+		picture.sar_height                     = decoded.sar_height;
+		picture.video_signal_type_present_flag = 1;
+		picture.video_format                   = 5;
+		picture.video_full_range_flag          = decoded.color_range == 2 ? 1 : 0;
+		picture.colour_description_present_flag =
+		    decoded.color_primaries != 0 || decoded.color_trc != 0 || decoded.color_space != 0 ? 1
+		                                                                                       : 0;
+		picture.colour_primaries         = decoded.color_primaries;
+		picture.transfer_characteristics = decoded.color_trc;
+		picture.matrix_coefficients      = decoded.color_space;
+		std::memcpy(first_picture_info, &picture, requested_size);
+	} else {
+		const auto result = fill_common(first_picture_info, true);
+		if (result != OK) {
+			return result;
+		}
+	}
+
+	if (second_picture_info != nullptr) {
+		const auto result = fill_common(second_picture_info, false);
+		if (result != OK) {
+			return result;
+		}
+	}
 	return OK;
 }
 

src/libs/videoDec2Decoder.cpp

@@ -0,0 +1,412 @@
+#include "libs/videoDec2Decoder.h"
+
+#include "common/logging/log.h"
+
+#include <algorithm>
+#include <cstddef>
+#include <cstdint>
+#include <cstring>
+#include <limits>
+#include <mutex>
+#include <unordered_map>
+#include <unordered_set>
+
+extern "C" {
+#include <libavcodec/avcodec.h>
+#include <libavutil/buffer.h>
+#include <libavutil/error.h>
+#include <libavutil/frame.h>
+#include <libavutil/pixfmt.h>
+#include <libswscale/swscale.h>
+}
+
+namespace Libs::VideoDec2::Decoder {
+
+namespace {
+
+constexpr uint32_t CODEC_TYPE_AVC  = 1;
+constexpr uint32_t CODEC_TYPE_HEVC = 974921;
+constexpr uint32_t CODEC_TYPE_VP9  = 2382845;
+
+struct PacketMetadata {
+	uint64_t pts           = TIMESTAMP_INVALID;
+	uint64_t dts           = TIMESTAMP_INVALID;
+	uint64_t attached_data = 0;
+};
+
+struct StoredPicture {
+	const Instance* owner = nullptr;
+	PictureInfo     info;
+};
+
+std::mutex                               g_picture_mutex;
+std::unordered_map<void*, StoredPicture> g_picture_infos;
+
+AVCodecID GetAvCodecId(uint32_t codec_type) {
+	switch (codec_type) {
+		case CODEC_TYPE_AVC: return AV_CODEC_ID_H264;
+		case CODEC_TYPE_HEVC: return AV_CODEC_ID_HEVC;
+		case CODEC_TYPE_VP9: return AV_CODEC_ID_VP9;
+		default: return AV_CODEC_ID_NONE;
+	}
+}
+
+const char* AvErrorString(int error) {
+	thread_local char text[AV_ERROR_MAX_STRING_SIZE] {};
+	if (av_strerror(error, text, sizeof(text)) != 0) {
+		std::strcpy(text, "unknown FFmpeg error");
+	}
+	return text;
+}
+
+uint32_t AlignUp(uint32_t value, uint32_t alignment) {
+	return (value + alignment - 1u) & ~(alignment - 1u);
+}
+
+int64_t ToAvTimestamp(uint64_t timestamp) {
+	return timestamp == TIMESTAMP_INVALID ||
+	               timestamp > static_cast<uint64_t>(std::numeric_limits<int64_t>::max())
+	           ? AV_NOPTS_VALUE
+	           : static_cast<int64_t>(timestamp);
+}
+
+} // namespace
+
+class Instance {
+public:
+	explicit Instance(const Config& config): m_config(config) {}
+
+	~Instance() {
+		ClearPictureMetadata();
+		if (m_sws != nullptr) {
+			sws_freeContext(m_sws);
+		}
+		if (m_codec != nullptr) {
+			avcodec_free_context(&m_codec);
+		}
+	}
+
+	Instance(const Instance&)            = delete;
+	Instance& operator=(const Instance&) = delete;
+
+	[[nodiscard]] bool Initialize() {
+		const AVCodec* decoder = avcodec_find_decoder(GetAvCodecId(m_config.codec_type));
+		if (decoder == nullptr) {
+			LOGF("Videodec2: FFmpeg decoder is unavailable for codec type %u\n",
+			     m_config.codec_type);
+			return false;
+		}
+		m_codec = avcodec_alloc_context3(decoder);
+		if (m_codec == nullptr) {
+			LOGF("Videodec2: avcodec_alloc_context3 failed\n");
+			return false;
+		}
+
+		// This carries PTS/DTS/attachedData through codecs that reorder B frames.
+		m_codec->flags |= AV_CODEC_FLAG_COPY_OPAQUE;
+		const int result = avcodec_open2(m_codec, decoder, nullptr);
+		if (result < 0) {
+			LOGF("Videodec2: avcodec_open2 failed: %s (%d)\n", AvErrorString(result), result);
+			return false;
+		}
+		return true;
+	}
+
+	[[nodiscard]] uint32_t CodecType() const { return m_config.codec_type; }
+
+	[[nodiscard]] Result DecodeInput(const Input& input, const FrameBuffer& frame_buffer,
+	                                 Output* output) {
+		std::scoped_lock lock(m_mutex);
+		*output    = {};
+		m_draining = false;
+
+		AVPacket* packet = av_packet_alloc();
+		AVFrame*  frame  = av_frame_alloc();
+		if (packet == nullptr || frame == nullptr ||
+		    input.size > static_cast<size_t>(std::numeric_limits<int>::max())) {
+			av_packet_free(&packet);
+			av_frame_free(&frame);
+			return Result::ApiFail;
+		}
+
+		int result = av_new_packet(packet, static_cast<int>(input.size));
+		if (result < 0) {
+			LOGF("Videodec2: av_new_packet failed: %s (%d)\n", AvErrorString(result), result);
+			av_packet_free(&packet);
+			av_frame_free(&frame);
+			return Result::ApiFail;
+		}
+		std::memcpy(packet->data, input.data, input.size);
+		packet->pts = ToAvTimestamp(input.pts);
+		packet->dts = ToAvTimestamp(input.dts);
+
+		packet->opaque_ref = av_buffer_alloc(sizeof(PacketMetadata));
+		if (packet->opaque_ref == nullptr) {
+			av_packet_free(&packet);
+			av_frame_free(&frame);
+			return Result::ApiFail;
+		}
+		const PacketMetadata metadata {input.pts, input.dts, input.attached_data};
+		std::memcpy(packet->opaque_ref->data, &metadata, sizeof(metadata));
+
+		bool have_pending_frame = false;
+		result                  = avcodec_send_packet(m_codec, packet);
+		if (result == AVERROR(EAGAIN)) {
+			result = avcodec_receive_frame(m_codec, frame);
+			if (result < 0) {
+				LOGF("Videodec2: decoder rejected an AU while no output was available: %s (%d)\n",
+				     AvErrorString(result), result);
+				av_packet_free(&packet);
+				av_frame_free(&frame);
+				return Result::AccessUnit;
+			}
+			have_pending_frame = true;
+			result             = avcodec_send_packet(m_codec, packet);
+		}
+		if (result < 0) {
+			LOGF("Videodec2: avcodec_send_packet failed: %s (%d)\n", AvErrorString(result), result);
+			av_packet_free(&packet);
+			av_frame_free(&frame);
+			return Result::AccessUnit;
+		}
+
+		Result decode_result = Result::Ok;
+		if (!have_pending_frame) {
+			result = avcodec_receive_frame(m_codec, frame);
+			if (result != AVERROR(EAGAIN) && result != AVERROR_EOF) {
+				if (result < 0) {
+					LOGF("Videodec2: avcodec_receive_frame failed: %s (%d)\n",
+					     AvErrorString(result), result);
+					decode_result = Result::AccessUnit;
+				} else {
+					decode_result = CopyFrame(frame, frame_buffer, output);
+				}
+			}
+		} else {
+			decode_result = CopyFrame(frame, frame_buffer, output);
+		}
+
+		av_packet_free(&packet);
+		av_frame_free(&frame);
+		return decode_result;
+	}
+
+	[[nodiscard]] Result FlushOutput(const FrameBuffer& frame_buffer, Output* output) {
+		std::scoped_lock lock(m_mutex);
+		*output = {};
+
+		AVFrame* frame = av_frame_alloc();
+		if (frame == nullptr) {
+			return Result::ApiFail;
+		}
+
+		if (!m_draining) {
+			const int send_result = avcodec_send_packet(m_codec, nullptr);
+			if (send_result == 0 || send_result == AVERROR_EOF) {
+				m_draining = true;
+			} else if (send_result != AVERROR(EAGAIN)) {
+				LOGF("Videodec2: flushing decoder failed: %s (%d)\n", AvErrorString(send_result),
+				     send_result);
+				av_frame_free(&frame);
+				return Result::ApiFail;
+			}
+		}
+
+		const int receive_result = avcodec_receive_frame(m_codec, frame);
+		if (receive_result == AVERROR(EAGAIN) || receive_result == AVERROR_EOF) {
+			av_frame_free(&frame);
+			return Result::Ok;
+		}
+		if (receive_result < 0) {
+			LOGF("Videodec2: receiving a flushed frame failed: %s (%d)\n",
+			     AvErrorString(receive_result), receive_result);
+			av_frame_free(&frame);
+			return Result::ApiFail;
+		}
+
+		const auto result = CopyFrame(frame, frame_buffer, output);
+		av_frame_free(&frame);
+		return result;
+	}
+
+	void ResetDecoder() {
+		std::scoped_lock lock(m_mutex);
+		avcodec_flush_buffers(m_codec);
+		m_draining = false;
+		ClearPictureMetadata();
+	}
+
+private:
+	[[nodiscard]] PictureInfo MakePictureInfo(const AVFrame* frame) const {
+		PictureInfo result {};
+		if (frame->opaque_ref != nullptr && frame->opaque_ref->size >= sizeof(PacketMetadata)) {
+			PacketMetadata metadata {};
+			std::memcpy(&metadata, frame->opaque_ref->data, sizeof(metadata));
+			result.pts           = metadata.pts;
+			result.dts           = metadata.dts;
+			result.attached_data = metadata.attached_data;
+		} else {
+			result.pts = frame->pts == AV_NOPTS_VALUE ? TIMESTAMP_INVALID
+			                                          : static_cast<uint64_t>(frame->pts);
+			result.dts = frame->pkt_dts == AV_NOPTS_VALUE ? TIMESTAMP_INVALID
+			                                              : static_cast<uint64_t>(frame->pkt_dts);
+		}
+		result.codec_type  = m_config.codec_type;
+		result.width       = static_cast<uint32_t>(frame->width);
+		result.height      = static_cast<uint32_t>(frame->height);
+		result.crop_left   = static_cast<uint32_t>(frame->crop_left);
+		result.crop_right  = static_cast<uint32_t>(frame->crop_right);
+		result.crop_top    = static_cast<uint32_t>(frame->crop_top);
+		result.crop_bottom = static_cast<uint32_t>(frame->crop_bottom);
+		result.profile     = m_codec->profile > 0 ? static_cast<uint32_t>(m_codec->profile) : 0;
+		result.level       = m_codec->level > 0 ? static_cast<uint32_t>(m_codec->level) : 0;
+		result.sar_width =
+		    frame->sample_aspect_ratio.num > 0
+		        ? static_cast<uint16_t>(std::min(frame->sample_aspect_ratio.num, 65535))
+		        : 0;
+		result.sar_height =
+		    frame->sample_aspect_ratio.den > 0
+		        ? static_cast<uint16_t>(std::min(frame->sample_aspect_ratio.den, 65535))
+		        : 0;
+		result.color_range     = static_cast<uint8_t>(frame->color_range);
+		result.color_primaries = static_cast<uint8_t>(frame->color_primaries);
+		result.color_trc       = static_cast<uint8_t>(frame->color_trc);
+		result.color_space     = static_cast<uint8_t>(frame->colorspace);
+		result.key_frame       = (frame->flags & AV_FRAME_FLAG_KEY) != 0;
+		return result;
+	}
+
+	[[nodiscard]] Result CopyFrame(const AVFrame* frame, const FrameBuffer& frame_buffer,
+	                               Output* output) {
+		if (frame->width <= 0 || frame->height <= 0) {
+			return Result::ApiFail;
+		}
+		if ((m_config.max_width > 0 && frame->width > m_config.max_width) ||
+		    (m_config.max_height > 0 && frame->height > m_config.max_height)) {
+			return Result::OversizeDecode;
+		}
+
+		const auto width       = static_cast<uint32_t>(frame->width);
+		const auto height      = static_cast<uint32_t>(frame->height);
+		const auto pitch       = AlignUp(width, 256);
+		const auto chroma_rows = (static_cast<uint64_t>(height) + 1u) / 2u;
+		const auto required =
+		    static_cast<uint64_t>(pitch) * height + static_cast<uint64_t>(pitch) * chroma_rows;
+		if (required > frame_buffer.size) {
+			return Result::FrameBufferSize;
+		}
+
+		auto* dst = static_cast<uint8_t*>(frame_buffer.data);
+		std::memset(dst, 0, static_cast<size_t>(required));
+		if (frame->format == AV_PIX_FMT_NV12) {
+			for (uint32_t y = 0; y < height; y++) {
+				std::memcpy(dst + static_cast<size_t>(y) * pitch,
+				            frame->data[0] + static_cast<ptrdiff_t>(y) * frame->linesize[0], width);
+			}
+			auto* chroma = dst + static_cast<size_t>(pitch) * height;
+			for (uint32_t y = 0; y < chroma_rows; y++) {
+				std::memcpy(chroma + static_cast<size_t>(y) * pitch,
+				            frame->data[1] + static_cast<ptrdiff_t>(y) * frame->linesize[1], width);
+			}
+		} else {
+			m_sws = sws_getCachedContext(m_sws, frame->width, frame->height,
+			                             static_cast<AVPixelFormat>(frame->format), frame->width,
+			                             frame->height, AV_PIX_FMT_NV12, SWS_FAST_BILINEAR, nullptr,
+			                             nullptr, nullptr);
+			if (m_sws == nullptr) {
+				return Result::ApiFail;
+			}
+			uint8_t* output_planes[4]  = {dst, dst + static_cast<size_t>(pitch) * height, nullptr,
+			                              nullptr};
+			int      output_strides[4] = {static_cast<int>(pitch), static_cast<int>(pitch), 0, 0};
+			if (sws_scale(m_sws, frame->data, frame->linesize, 0, frame->height, output_planes,
+			              output_strides) != frame->height) {
+				return Result::ApiFail;
+			}
+		}
+
+		output->valid           = true;
+		output->error_frame     = (frame->flags & AV_FRAME_FLAG_CORRUPT) != 0;
+		output->buffer_accepted = true;
+		output->codec_type      = m_config.codec_type;
+		output->width           = width;
+		output->pitch           = pitch;
+		output->height          = height;
+		output->buffer          = frame_buffer.data;
+		output->buffer_size     = frame_buffer.size;
+
+		{
+			std::scoped_lock lock(g_picture_mutex);
+			g_picture_infos[frame_buffer.data] = {this, MakePictureInfo(frame)};
+			m_picture_buffers.insert(frame_buffer.data);
+		}
+		return Result::Ok;
+	}
+
+	void ClearPictureMetadata() {
+		std::scoped_lock lock(g_picture_mutex);
+		for (auto* buffer: m_picture_buffers) {
+			const auto it = g_picture_infos.find(buffer);
+			if (it != g_picture_infos.end() && it->second.owner == this) {
+				g_picture_infos.erase(it);
+			}
+		}
+		m_picture_buffers.clear();
+	}
+
+	Config                    m_config;
+	AVCodecContext*           m_codec    = nullptr;
+	SwsContext*               m_sws      = nullptr;
+	bool                      m_draining = false;
+	std::mutex                m_mutex;
+	std::unordered_set<void*> m_picture_buffers;
+};
+
+bool IsCodecSupported(uint32_t codec_type) {
+	return GetAvCodecId(codec_type) != AV_CODEC_ID_NONE;
+}
+
+Instance* Create(const Config& config) {
+	if (!IsCodecSupported(config.codec_type)) {
+		return nullptr;
+	}
+	auto* instance = new Instance(config);
+	if (!instance->Initialize()) {
+		delete instance;
+		return nullptr;
+	}
+	return instance;
+}
+
+void Destroy(Instance* instance) {
+	delete instance;
+}
+
+uint32_t GetCodecType(const Instance* instance) {
+	return instance->CodecType();
+}
+
+Result Decode(Instance* instance, const Input& input, const FrameBuffer& frame_buffer,
+              Output* output) {
+	return instance->DecodeInput(input, frame_buffer, output);
+}
+
+Result Flush(Instance* instance, const FrameBuffer& frame_buffer, Output* output) {
+	return instance->FlushOutput(frame_buffer, output);
+}
+
+void Reset(Instance* instance) {
+	instance->ResetDecoder();
+}
+
+bool GetPictureInfo(void* frame_buffer, PictureInfo* picture_info) {
+	std::scoped_lock lock(g_picture_mutex);
+	const auto       it = g_picture_infos.find(frame_buffer);
+	if (it == g_picture_infos.end()) {
+		return false;
+	}
+	*picture_info = it->second.info;
+	return true;
+}
+
+} // namespace Libs::VideoDec2::Decoder

src/libs/videoDec2Decoder.h

@@ -0,0 +1,86 @@
+#ifndef EMULATOR_INCLUDE_EMULATOR_LIBS_VIDEODEC2DECODER_H_
+#define EMULATOR_INCLUDE_EMULATOR_LIBS_VIDEODEC2DECODER_H_
+
+#include <cstddef>
+#include <cstdint>
+
+namespace Libs::VideoDec2::Decoder {
+
+constexpr uint64_t TIMESTAMP_INVALID = UINT64_MAX;
+
+enum class Result {
+	Ok,
+	ApiFail,
+	AccessUnit,
+	FrameBufferSize,
+	OversizeDecode,
+};
+
+struct Config {
+	uint32_t codec_type = 0;
+	int32_t  max_width  = -1;
+	int32_t  max_height = -1;
+};
+
+struct Input {
+	const void* data          = nullptr;
+	size_t      size          = 0;
+	uint64_t    pts           = TIMESTAMP_INVALID;
+	uint64_t    dts           = TIMESTAMP_INVALID;
+	uint64_t    attached_data = 0;
+};
+
+struct FrameBuffer {
+	void*  data = nullptr;
+	size_t size = 0;
+};
+
+struct Output {
+	bool     valid           = false;
+	bool     error_frame     = false;
+	bool     buffer_accepted = false;
+	uint32_t codec_type      = 0;
+	uint32_t width           = 0;
+	uint32_t pitch           = 0;
+	uint32_t height          = 0;
+	void*    buffer          = nullptr;
+	size_t   buffer_size     = 0;
+};
+
+struct PictureInfo {
+	uint64_t pts             = TIMESTAMP_INVALID;
+	uint64_t dts             = TIMESTAMP_INVALID;
+	uint64_t attached_data   = 0;
+	uint32_t codec_type      = 0;
+	uint32_t width           = 0;
+	uint32_t height          = 0;
+	uint32_t crop_left       = 0;
+	uint32_t crop_right      = 0;
+	uint32_t crop_top        = 0;
+	uint32_t crop_bottom     = 0;
+	uint32_t profile         = 0;
+	uint32_t level           = 0;
+	uint16_t sar_width       = 0;
+	uint16_t sar_height      = 0;
+	uint8_t  color_range     = 0;
+	uint8_t  color_primaries = 0;
+	uint8_t  color_trc       = 0;
+	uint8_t  color_space     = 0;
+	bool     key_frame       = false;
+};
+
+class Instance;
+
+[[nodiscard]] bool      IsCodecSupported(uint32_t codec_type);
+[[nodiscard]] Instance* Create(const Config& config);
+void                    Destroy(Instance* instance);
+[[nodiscard]] uint32_t  GetCodecType(const Instance* instance);
+[[nodiscard]] Result Decode(Instance* instance, const Input& input, const FrameBuffer& frame_buffer,
+                            Output* output);
+[[nodiscard]] Result Flush(Instance* instance, const FrameBuffer& frame_buffer, Output* output);
+void                 Reset(Instance* instance);
+[[nodiscard]] bool   GetPictureInfo(void* frame_buffer, PictureInfo* picture_info);
+
+} // namespace Libs::VideoDec2::Decoder
+
+#endif // EMULATOR_INCLUDE_EMULATOR_LIBS_VIDEODEC2DECODER_H_

src/loader/x64InstructionEmulator.cpp

@@ -79,10 +79,10 @@ static void Sha1Msg1(XmmWords& dest, const XmmWords& src2) {
 	const uint32_t w3 = dest.w[0];
 	const uint32_t w4 = src2.w[3];
 	const uint32_t w5 = src2.w[2];
-	dest.w[3] = w2 ^ w0;
-	dest.w[2] = w3 ^ w1;
-	dest.w[1] = w4 ^ w2;
-	dest.w[0] = w5 ^ w3;
+	dest.w[3]         = w2 ^ w0;
+	dest.w[2]         = w3 ^ w1;
+	dest.w[1]         = w4 ^ w2;
+	dest.w[0]         = w5 ^ w3;
 }
 
 static void Sha1Msg2(XmmWords& dest, const XmmWords& src2) {
@@ -93,18 +93,18 @@ static void Sha1Msg2(XmmWords& dest, const XmmWords& src2) {
 	const uint32_t w17 = Rol32(dest.w[2] ^ w14, 1u);
 	const uint32_t w18 = Rol32(dest.w[1] ^ w15, 1u);
 	const uint32_t w19 = Rol32(dest.w[0] ^ w16, 1u);
-	dest.w[3] = w16;
-	dest.w[2] = w17;
-	dest.w[1] = w18;
-	dest.w[0] = w19;
+	dest.w[3]          = w16;
+	dest.w[2]          = w17;
+	dest.w[1]          = w18;
+	dest.w[0]          = w19;
 }
 
 static void Sha1Nexte(XmmWords& dest, const XmmWords& src2) {
-	const uint32_t tmp  = Rol32(dest.w[3], 30u);
-	dest.w[3]           = src2.w[3] + tmp;
-	dest.w[2]           = src2.w[2];
-	dest.w[1]           = src2.w[1];
-	dest.w[0]           = src2.w[0];
+	const uint32_t tmp = Rol32(dest.w[3], 30u);
+	dest.w[3]          = src2.w[3] + tmp;
+	dest.w[2]          = src2.w[2];
+	dest.w[1]          = src2.w[1];
+	dest.w[0]          = src2.w[0];
 }
 
 static uint32_t Sha1RoundFunc(uint8_t group, uint32_t b, uint32_t c, uint32_t d) {
@@ -185,10 +185,10 @@ static void Sha256Msg1(XmmWords& dest, const XmmWords& src2) {
 	const uint32_t w2 = dest.w[2];
 	const uint32_t w1 = dest.w[1];
 	const uint32_t w0 = dest.w[0];
-	dest.w[3]           = w3 + Sha256Sigma0(w4);
-	dest.w[2]           = w2 + Sha256Sigma0(w3);
-	dest.w[1]           = w1 + Sha256Sigma0(w2);
-	dest.w[0]           = w0 + Sha256Sigma0(w1);
+	dest.w[3]         = w3 + Sha256Sigma0(w4);
+	dest.w[2]         = w2 + Sha256Sigma0(w3);
+	dest.w[1]         = w1 + Sha256Sigma0(w2);
+	dest.w[0]         = w0 + Sha256Sigma0(w1);
 }
 
 static void Sha256Msg2(XmmWords& dest, const XmmWords& src2) {
@@ -312,7 +312,9 @@ static bool DecodeShaNiInsn(const uint8_t* rip, ShaNiInsn& insn) {
 	return true;
 }
 
-static bool ShaNiModrmIsRegister(uint8_t modrm) { return (modrm & 0xc0u) == 0xc0u; }
+static bool ShaNiModrmIsRegister(uint8_t modrm) {
+	return (modrm & 0xc0u) == 0xc0u;
+}
 
 static uint8_t ShaNiRegIndex(uint8_t modrm, uint8_t rex, bool reg_field) {
 	if (reg_field) {
@@ -321,7 +323,7 @@ static uint8_t ShaNiRegIndex(uint8_t modrm, uint8_t rex, bool reg_field) {
 	return (modrm & 0x07u) | ((rex & 0x01u) << 3u);
 }
 
-static bool ResolveShaNiMemoryAddress(const uint8_t* rip, const ShaNiInsn& insn,
+static bool ResolveShaNiMemoryAddress(const uint8_t* rip, const ShaNiInsn&    insn,
                                       const uint64_t (&gpr)[16], const void*& address) {
 	const uint8_t modrm = rip[insn.modrm_offset];
 	const uint8_t mod   = modrm >> 6u;
@@ -334,12 +336,12 @@ static bool ResolveShaNiMemoryAddress(const uint8_t* rip, const ShaNiInsn& insn,
 	uint64_t result = 0;
 
 	if (rm == 4u) {
-		const uint8_t sib        = rip[offset++];
-		const uint8_t scale      = sib >> 6u;
-		const uint8_t index_low  = (sib >> 3u) & 0x07u;
-		const uint8_t base_low   = sib & 0x07u;
-		const bool    has_index  = index_low != 4u || (insn.rex & 0x02u) != 0;
-		const bool    has_base   = mod != 0u || base_low != 5u;
+		const uint8_t sib       = rip[offset++];
+		const uint8_t scale     = sib >> 6u;
+		const uint8_t index_low = (sib >> 3u) & 0x07u;
+		const uint8_t base_low  = sib & 0x07u;
+		const bool    has_index = index_low != 4u || (insn.rex & 0x02u) != 0;
+		const bool    has_base  = mod != 0u || base_low != 5u;
 
 		if (has_base) {
 			const uint8_t base = base_low | ((insn.rex & 0x01u) << 3u);

tests/EventQueueLifetimeTests.cpp

@@ -1,5 +1,4 @@
 #include "kernel/eventQueue.h"
-
 #include "libs/errno.h"
 
 #include <algorithm>
@@ -17,484 +16,428 @@ namespace EventQueue = Libs::LibKernel::EventQueue;
 using Libs::LibKernel::KERNEL_ERROR_EBADF;
 using Libs::LibKernel::KERNEL_ERROR_ENOENT;
 
-void Check(bool value, const char *text) {
-  if (!value) {
-    std::fprintf(stderr, "EventQueueLifetimeTests: failed: %s\n", text);
-    std::abort();
-  }
+void Check(bool value, const char* text) {
+	if (!value) {
+		std::fprintf(stderr, "EventQueueLifetimeTests: failed: %s\n", text);
+		std::abort();
+	}
 }
 
-void CheckConcurrentResult(int result, const char *text) {
-  Check(result == OK || result == KERNEL_ERROR_EBADF ||
-            result == KERNEL_ERROR_ENOENT,
-        text);
+void CheckConcurrentResult(int result, const char* text) {
+	Check(result == OK || result == KERNEL_ERROR_EBADF || result == KERNEL_ERROR_ENOENT, text);
 }
 
-void CountDeletedEvent(EventQueue::KernelEqueue,
-                       EventQueue::KernelEqueueEvent *event) {
-  auto *count = static_cast<std::atomic_uint32_t *>(event->filter.data);
-  count->fetch_add(1, std::memory_order_relaxed);
+void CountDeletedEvent(EventQueue::KernelEqueue, EventQueue::KernelEqueueEvent* event) {
+	auto* count = static_cast<std::atomic_uint32_t*>(event->filter.data);
+	count->fetch_add(1, std::memory_order_relaxed);
 }
 
 struct DuplicateEventOwner {
-  std::atomic_uint32_t delete_count{0};
+	std::atomic_uint32_t delete_count {0};
 };
 
-void QueueDuplicateEvent(EventQueue::KernelEqueueEvent *event,
-                         void *trigger_data) {
-  auto next = event->event;
-  next.data = reinterpret_cast<intptr_t>(trigger_data);
-  if (event->triggered) {
-    event->pending_events.push_back(next);
-  } else {
-    event->event = next;
-    event->triggered = true;
-  }
+void QueueDuplicateEvent(EventQueue::KernelEqueueEvent* event, void* trigger_data) {
+	auto next = event->event;
+	next.data = reinterpret_cast<intptr_t>(trigger_data);
+	if (event->triggered) {
+		event->pending_events.push_back(next);
+	} else {
+		event->event     = next;
+		event->triggered = true;
+	}
 }
 
-void ResetDuplicateEvent(EventQueue::KernelEqueueEvent *event) {
-  event->triggered = false;
-  event->event.data = 0;
+void ResetDuplicateEvent(EventQueue::KernelEqueueEvent* event) {
+	event->triggered  = false;
+	event->event.data = 0;
 }
 
-void DeleteDuplicateEvent(EventQueue::KernelEqueue,
-                          EventQueue::KernelEqueueEvent *event) {
-  auto *owner = static_cast<DuplicateEventOwner *>(event->filter.data);
-  owner->delete_count.fetch_add(1, std::memory_order_relaxed);
+void DeleteDuplicateEvent(EventQueue::KernelEqueue, EventQueue::KernelEqueueEvent* event) {
+	auto* owner = static_cast<DuplicateEventOwner*>(event->filter.data);
+	owner->delete_count.fetch_add(1, std::memory_order_relaxed);
 }
 
-void PoisonDuplicateEvent(EventQueue::KernelEqueueEvent *, void *) {
-  Check(false, "duplicate add replaced trigger callback");
+void PoisonDuplicateEvent(EventQueue::KernelEqueueEvent*, void*) {
+	Check(false, "duplicate add replaced trigger callback");
 }
 
 void TestDuplicateAddPreservesEventState() {
-  EventQueue::KernelEqueue queue = EventQueue::KERNEL_EQUEUE_INVALID;
-  Check(EventQueue::KernelCreateEqueue(&queue, "duplicate-add") == OK,
-        "create duplicate add queue");
+	EventQueue::KernelEqueue queue = EventQueue::KERNEL_EQUEUE_INVALID;
+	Check(EventQueue::KernelCreateEqueue(&queue, "duplicate-add") == OK,
+	      "create duplicate add queue");
 
-  auto original_owner = std::make_shared<DuplicateEventOwner>();
-  std::weak_ptr<DuplicateEventOwner> weak_original = original_owner;
-  EventQueue::KernelEqueueEvent original{};
-  original.event.ident = 17;
-  original.event.filter = EventQueue::KERNEL_EVFILT_VIDEO_OUT;
-  original.event.udata = reinterpret_cast<void *>(0x1111);
-  original.filter.data = original_owner.get();
-  original.filter.owner = original_owner;
-  original.filter.trigger_func = QueueDuplicateEvent;
-  original.filter.reset_func = ResetDuplicateEvent;
-  original.filter.delete_event_func = DeleteDuplicateEvent;
-  Check(EventQueue::KernelAddEvent(queue, original) == OK,
-        "add original duplicate event");
-  Check(EventQueue::KernelTriggerEvent(queue, 17,
-                                       EventQueue::KERNEL_EVFILT_VIDEO_OUT,
-                                       reinterpret_cast<void *>(0x1234)) == OK,
-        "queue first trigger");
-  Check(EventQueue::KernelTriggerEvent(queue, 17,
-                                       EventQueue::KERNEL_EVFILT_VIDEO_OUT,
-                                       reinterpret_cast<void *>(0x5678)) == OK,
-        "queue pending trigger");
+	auto                               original_owner = std::make_shared<DuplicateEventOwner>();
+	std::weak_ptr<DuplicateEventOwner> weak_original  = original_owner;
+	EventQueue::KernelEqueueEvent      original {};
+	original.event.ident              = 17;
+	original.event.filter             = EventQueue::KERNEL_EVFILT_VIDEO_OUT;
+	original.event.udata              = reinterpret_cast<void*>(0x1111);
+	original.filter.data              = original_owner.get();
+	original.filter.owner             = original_owner;
+	original.filter.trigger_func      = QueueDuplicateEvent;
+	original.filter.reset_func        = ResetDuplicateEvent;
+	original.filter.delete_event_func = DeleteDuplicateEvent;
+	Check(EventQueue::KernelAddEvent(queue, original) == OK, "add original duplicate event");
+	Check(EventQueue::KernelTriggerEvent(queue, 17, EventQueue::KERNEL_EVFILT_VIDEO_OUT,
+	                                     reinterpret_cast<void*>(0x1234)) == OK,
+	      "queue first trigger");
+	Check(EventQueue::KernelTriggerEvent(queue, 17, EventQueue::KERNEL_EVFILT_VIDEO_OUT,
+	                                     reinterpret_cast<void*>(0x5678)) == OK,
+	      "queue pending trigger");
 
-  auto replacement_owner = std::make_shared<DuplicateEventOwner>();
-  std::weak_ptr<DuplicateEventOwner> weak_replacement = replacement_owner;
-  EventQueue::KernelEqueueEvent duplicate{};
-  duplicate.triggered = false;
-  duplicate.deadline_ns = 1;
-  duplicate.event.ident = 17;
-  duplicate.event.filter = EventQueue::KERNEL_EVFILT_VIDEO_OUT;
-  duplicate.event.data = 0x7fffffff;
-  duplicate.event.udata = reinterpret_cast<void *>(0x2222);
-  duplicate.filter.data = replacement_owner.get();
-  duplicate.filter.owner = replacement_owner;
-  duplicate.filter.trigger_func = PoisonDuplicateEvent;
-  Check(EventQueue::KernelAddEvent(queue, duplicate) == OK,
-        "update duplicate event");
+	auto                               replacement_owner = std::make_shared<DuplicateEventOwner>();
+	std::weak_ptr<DuplicateEventOwner> weak_replacement  = replacement_owner;
+	EventQueue::KernelEqueueEvent      duplicate {};
+	duplicate.triggered           = false;
+	duplicate.deadline_ns         = 1;
+	duplicate.event.ident         = 17;
+	duplicate.event.filter        = EventQueue::KERNEL_EVFILT_VIDEO_OUT;
+	duplicate.event.data          = 0x7fffffff;
+	duplicate.event.udata         = reinterpret_cast<void*>(0x2222);
+	duplicate.filter.data         = replacement_owner.get();
+	duplicate.filter.owner        = replacement_owner;
+	duplicate.filter.trigger_func = PoisonDuplicateEvent;
+	Check(EventQueue::KernelAddEvent(queue, duplicate) == OK, "update duplicate event");
 
-  duplicate.filter.owner.reset();
-  replacement_owner.reset();
-  Check(weak_replacement.expired(), "duplicate owner is not retained");
-  original.filter.owner.reset();
-  original_owner.reset();
-  Check(!weak_original.expired(), "original event owner remains retained");
+	duplicate.filter.owner.reset();
+	replacement_owner.reset();
+	Check(weak_replacement.expired(), "duplicate owner is not retained");
+	original.filter.owner.reset();
+	original_owner.reset();
+	Check(!weak_original.expired(), "original event owner remains retained");
 
-  EventQueue::KernelEvent events[2]{};
-  int out = 0;
-  Libs::LibKernel::KernelUseconds timeout = 0;
-  Check(EventQueue::KernelWaitEqueue(queue, events, 2, &out, &timeout) == OK,
-        "read queued duplicate triggers");
-  Check(out == 2, "duplicate add preserves pending event count");
-  Check(events[0].data == 0x1234 && events[1].data == 0x5678,
-        "duplicate add preserves current and pending event data");
-  Check(events[0].udata == reinterpret_cast<void *>(0x2222),
-        "duplicate add updates current user data");
-  Check(events[1].udata == reinterpret_cast<void *>(0x2222),
-        "duplicate add updates pending user data");
+	EventQueue::KernelEvent         events[2] {};
+	int                             out     = 0;
+	Libs::LibKernel::KernelUseconds timeout = 0;
+	Check(EventQueue::KernelWaitEqueue(queue, events, 2, &out, &timeout) == OK,
+	      "read queued duplicate triggers");
+	Check(out == 2, "duplicate add preserves pending event count");
+	Check(events[0].data == 0x1234 && events[1].data == 0x5678,
+	      "duplicate add preserves current and pending event data");
+	Check(events[0].udata == reinterpret_cast<void*>(0x2222),
+	      "duplicate add updates current user data");
+	Check(events[1].udata == reinterpret_cast<void*>(0x2222),
+	      "duplicate add updates pending user data");
 
-  EventQueue::KernelEvent timer_event{};
-  Check(EventQueue::KernelWaitEqueue(queue, &timer_event, 1, &out, &timeout) ==
-            OK &&
-            out == 1,
-        "duplicate add updates deadline metadata");
-  Check(timer_event.data == 0 &&
-            timer_event.udata == reinterpret_cast<void *>(0x2222),
-        "deadline trigger retains updated duplicate metadata");
+	EventQueue::KernelEvent timer_event {};
+	Check(EventQueue::KernelWaitEqueue(queue, &timer_event, 1, &out, &timeout) == OK && out == 1,
+	      "duplicate add updates deadline metadata");
+	Check(timer_event.data == 0 && timer_event.udata == reinterpret_cast<void*>(0x2222),
+	      "deadline trigger retains updated duplicate metadata");
 
-  auto retained_owner = weak_original.lock();
-  Check(retained_owner != nullptr, "original owner alive before delete");
-  Check(EventQueue::KernelDeleteEvent(queue, 17,
-                                      EventQueue::KERNEL_EVFILT_VIDEO_OUT) ==
-            OK,
-        "delete duplicate event");
-  Check(retained_owner->delete_count.load(std::memory_order_relaxed) == 1,
-        "duplicate add preserves delete callback");
-  retained_owner.reset();
-  Check(weak_original.expired(), "original owner released on delete");
-  Check(EventQueue::KernelDeleteEqueue(queue) == OK,
-        "delete duplicate add queue");
+	auto retained_owner = weak_original.lock();
+	Check(retained_owner != nullptr, "original owner alive before delete");
+	Check(EventQueue::KernelDeleteEvent(queue, 17, EventQueue::KERNEL_EVFILT_VIDEO_OUT) == OK,
+	      "delete duplicate event");
+	Check(retained_owner->delete_count.load(std::memory_order_relaxed) == 1,
+	      "duplicate add preserves delete callback");
+	retained_owner.reset();
+	Check(weak_original.expired(), "original owner released on delete");
+	Check(EventQueue::KernelDeleteEqueue(queue) == OK, "delete duplicate add queue");
 }
 
 struct SimulatedVideoOutEventState;
 
 struct SimulatedVideoOutRegistration {
-  EventQueue::KernelEqueue handle = EventQueue::KERNEL_EQUEUE_INVALID;
-  std::shared_ptr<SimulatedVideoOutEventState> state;
-  uint64_t marker = 0x123456789abcdef0ull;
+	EventQueue::KernelEqueue                     handle = EventQueue::KERNEL_EQUEUE_INVALID;
+	std::shared_ptr<SimulatedVideoOutEventState> state;
+	uint64_t                                     marker = 0x123456789abcdef0ull;
 };
 
 struct SimulatedVideoOutEventState {
-  SimulatedVideoOutEventState(std::atomic_uint32_t &stage,
-                              std::atomic_uint32_t &destroy_count)
-      : stage(stage), destroy_count(destroy_count) {}
+	SimulatedVideoOutEventState(std::atomic_uint32_t& stage, std::atomic_uint32_t& destroy_count)
+	    : stage(stage), destroy_count(destroy_count) {}
 
-  ~SimulatedVideoOutEventState() {
-    destroy_count.fetch_add(1, std::memory_order_relaxed);
-  }
+	~SimulatedVideoOutEventState() { destroy_count.fetch_add(1, std::memory_order_relaxed); }
 
-  std::mutex mutex;
-  std::vector<std::shared_ptr<SimulatedVideoOutRegistration>> queues;
-  std::atomic_uint32_t &stage;
-  std::atomic_uint32_t &destroy_count;
-  uint64_t marker = 0xfedcba9876543210ull;
+	std::mutex                                                  mutex;
+	std::vector<std::shared_ptr<SimulatedVideoOutRegistration>> queues;
+	std::atomic_uint32_t&                                       stage;
+	std::atomic_uint32_t&                                       destroy_count;
+	uint64_t                                                    marker = 0xfedcba9876543210ull;
 };
 
-void DetachSimulatedVideoOutEvent(EventQueue::KernelEqueue queue,
-                                  EventQueue::KernelEqueueEvent *event) {
-  auto *registration =
-      static_cast<SimulatedVideoOutRegistration *>(event->filter.data);
-  Check(registration != nullptr && registration->handle == queue,
-        "simulated registration identity");
-  auto state = registration->state;
-  Check(state != nullptr, "simulated event owns shared state");
-  state->stage.store(1, std::memory_order_release);
-  while (state->stage.load(std::memory_order_acquire) != 2) {
-    std::this_thread::yield();
-  }
+void DetachSimulatedVideoOutEvent(EventQueue::KernelEqueue       queue,
+                                  EventQueue::KernelEqueueEvent* event) {
+	auto* registration = static_cast<SimulatedVideoOutRegistration*>(event->filter.data);
+	Check(registration != nullptr && registration->handle == queue,
+	      "simulated registration identity");
+	auto state = registration->state;
+	Check(state != nullptr, "simulated event owns shared state");
+	state->stage.store(1, std::memory_order_release);
+	while (state->stage.load(std::memory_order_acquire) != 2) {
+		std::this_thread::yield();
+	}
 
-  {
-    std::lock_guard lock(state->mutex);
-    const auto entry =
-        std::find_if(state->queues.begin(), state->queues.end(),
-                     [registration](const auto &candidate) {
-                       return candidate.get() == registration;
-                     });
-    if (entry != state->queues.end()) {
-      state->queues.erase(entry);
-    }
-  }
-  event->filter.owner.reset();
-  Check(state->marker == 0xfedcba9876543210ull &&
-            registration->marker == 0x123456789abcdef0ull,
-        "callback state survives simulated port destruction");
+	{
+		std::lock_guard lock(state->mutex);
+		const auto      entry = std::find_if(
+		    state->queues.begin(), state->queues.end(),
+		    [registration](const auto& candidate) { return candidate.get() == registration; });
+		if (entry != state->queues.end()) {
+			state->queues.erase(entry);
+		}
+	}
+	event->filter.owner.reset();
+	Check(state->marker == 0xfedcba9876543210ull && registration->marker == 0x123456789abcdef0ull,
+	      "callback state survives simulated port destruction");
 }
 
 void TestCallbackStateOutlivesPort() {
-  EventQueue::KernelEqueue queue = EventQueue::KERNEL_EQUEUE_INVALID;
-  Check(EventQueue::KernelCreateEqueue(&queue, "shared-port-state") == OK,
-        "create shared port state queue");
+	EventQueue::KernelEqueue queue = EventQueue::KERNEL_EQUEUE_INVALID;
+	Check(EventQueue::KernelCreateEqueue(&queue, "shared-port-state") == OK,
+	      "create shared port state queue");
 
-  std::atomic_uint32_t stage{0};
-  std::atomic_uint32_t destroy_count{0};
-  auto port_state =
-      std::make_shared<SimulatedVideoOutEventState>(stage, destroy_count);
-  std::weak_ptr<SimulatedVideoOutEventState> weak_state = port_state;
-  auto registration = std::make_shared<SimulatedVideoOutRegistration>();
-  std::weak_ptr<SimulatedVideoOutRegistration> weak_registration =
-      registration;
-  registration->handle = queue;
-  registration->state = port_state;
-  port_state->queues.push_back(registration);
+	std::atomic_uint32_t stage {0};
+	std::atomic_uint32_t destroy_count {0};
+	auto port_state = std::make_shared<SimulatedVideoOutEventState>(stage, destroy_count);
+	std::weak_ptr<SimulatedVideoOutEventState> weak_state = port_state;
+	auto registration = std::make_shared<SimulatedVideoOutRegistration>();
+	std::weak_ptr<SimulatedVideoOutRegistration> weak_registration = registration;
+	registration->handle                                           = queue;
+	registration->state                                            = port_state;
+	port_state->queues.push_back(registration);
 
-  EventQueue::KernelEqueueEvent event{};
-  event.event.ident = 8;
-  event.event.filter = EventQueue::KERNEL_EVFILT_VIDEO_OUT;
-  event.filter.data = registration.get();
-  event.filter.owner = registration;
-  event.filter.delete_event_func = DetachSimulatedVideoOutEvent;
-  Check(EventQueue::KernelAddEvent(queue, event) == OK,
-        "add shared port state event");
-  event.filter.owner.reset();
+	EventQueue::KernelEqueueEvent event {};
+	event.event.ident              = 8;
+	event.event.filter             = EventQueue::KERNEL_EVFILT_VIDEO_OUT;
+	event.filter.data              = registration.get();
+	event.filter.owner             = registration;
+	event.filter.delete_event_func = DetachSimulatedVideoOutEvent;
+	Check(EventQueue::KernelAddEvent(queue, event) == OK, "add shared port state event");
+	event.filter.owner.reset();
 
-  std::jthread close([&] {
-    Check(EventQueue::KernelDeleteEqueue(queue) == OK,
-          "delete shared port state queue");
-  });
-  while (stage.load(std::memory_order_acquire) != 1) {
-    std::this_thread::yield();
-  }
+	std::jthread close([&] {
+		Check(EventQueue::KernelDeleteEqueue(queue) == OK, "delete shared port state queue");
+	});
+	while (stage.load(std::memory_order_acquire) != 1) {
+		std::this_thread::yield();
+	}
 
-  std::vector<std::shared_ptr<SimulatedVideoOutRegistration>> detached;
-  {
-    std::lock_guard lock(port_state->mutex);
-    detached = std::move(port_state->queues);
-  }
-  registration.reset();
-  detached.clear();
-  port_state.reset();
-  Check(!weak_state.expired(),
-        "callback state outlives simulated port object");
-  Check(!weak_registration.expired(),
-        "registration outlives simulated port object");
+	std::vector<std::shared_ptr<SimulatedVideoOutRegistration>> detached;
+	{
+		std::lock_guard lock(port_state->mutex);
+		detached = std::move(port_state->queues);
+	}
+	registration.reset();
+	detached.clear();
+	port_state.reset();
+	Check(!weak_state.expired(), "callback state outlives simulated port object");
+	Check(!weak_registration.expired(), "registration outlives simulated port object");
 
-  stage.store(2, std::memory_order_release);
-  close.join();
-  Check(weak_registration.expired(), "detached registration is released");
-  Check(weak_state.expired(), "detached event state is released");
-  Check(destroy_count.load(std::memory_order_relaxed) == 1,
-        "shared event state is destroyed exactly once");
+	stage.store(2, std::memory_order_release);
+	close.join();
+	Check(weak_registration.expired(), "detached registration is released");
+	Check(weak_state.expired(), "detached event state is released");
+	Check(destroy_count.load(std::memory_order_relaxed) == 1,
+	      "shared event state is destroyed exactly once");
 }
 
 struct OwnedCallbackPayload {
-  OwnedCallbackPayload(std::atomic_uint32_t &stage,
-                       std::atomic_uint32_t &delete_count,
-                       std::atomic_uint32_t &destroy_count)
-      : stage(stage), delete_count(delete_count), destroy_count(destroy_count) {}
+	OwnedCallbackPayload(std::atomic_uint32_t& stage, std::atomic_uint32_t& delete_count,
+	                     std::atomic_uint32_t& destroy_count)
+	    : stage(stage), delete_count(delete_count), destroy_count(destroy_count) {}
 
-  std::atomic_uint32_t &stage;
-  std::atomic_uint32_t &delete_count;
-  std::atomic_uint32_t &destroy_count;
-  uint64_t marker = 0xc0dec0dec0dec0deull;
+	std::atomic_uint32_t& stage;
+	std::atomic_uint32_t& delete_count;
+	std::atomic_uint32_t& destroy_count;
+	uint64_t              marker = 0xc0dec0dec0dec0deull;
 
-  ~OwnedCallbackPayload() {
-    destroy_count.fetch_add(1, std::memory_order_relaxed);
-  }
+	~OwnedCallbackPayload() { destroy_count.fetch_add(1, std::memory_order_relaxed); }
 };
 
-void DeleteOwnedEvent(EventQueue::KernelEqueue queue,
-                      EventQueue::KernelEqueueEvent *event) {
-  auto *payload = static_cast<OwnedCallbackPayload *>(event->filter.data);
-  Check(payload != nullptr, "owned callback payload");
-  Check(!EventQueue::KernelPinEqueue(queue),
-        "owned callback runs after registry removal");
-  payload->delete_count.fetch_add(1, std::memory_order_relaxed);
-  event->filter.owner.reset();
-  payload->stage.store(1, std::memory_order_release);
-  while (payload->stage.load(std::memory_order_acquire) != 2) {
-    std::this_thread::yield();
-  }
-  Check(payload->marker == 0xc0dec0dec0dec0deull,
-        "owned callback payload remains valid");
+void DeleteOwnedEvent(EventQueue::KernelEqueue queue, EventQueue::KernelEqueueEvent* event) {
+	auto* payload = static_cast<OwnedCallbackPayload*>(event->filter.data);
+	Check(payload != nullptr, "owned callback payload");
+	Check(!EventQueue::KernelPinEqueue(queue), "owned callback runs after registry removal");
+	payload->delete_count.fetch_add(1, std::memory_order_relaxed);
+	event->filter.owner.reset();
+	payload->stage.store(1, std::memory_order_release);
+	while (payload->stage.load(std::memory_order_acquire) != 2) {
+		std::this_thread::yield();
+	}
+	Check(payload->marker == 0xc0dec0dec0dec0deull, "owned callback payload remains valid");
 }
 
 void TestCallbackOwnsPayload() {
-  EventQueue::KernelEqueue queue = EventQueue::KERNEL_EQUEUE_INVALID;
-  Check(EventQueue::KernelCreateEqueue(&queue, "owned-callback") == OK,
-        "create owned callback queue");
+	EventQueue::KernelEqueue queue = EventQueue::KERNEL_EQUEUE_INVALID;
+	Check(EventQueue::KernelCreateEqueue(&queue, "owned-callback") == OK,
+	      "create owned callback queue");
 
-  std::atomic_uint32_t stage{0};
-  std::atomic_uint32_t delete_count{0};
-  std::atomic_uint32_t destroy_count{0};
-  auto registration =
-      std::make_shared<OwnedCallbackPayload>(stage, delete_count, destroy_count);
-  std::weak_ptr<OwnedCallbackPayload> weak_registration = registration;
-  std::vector<std::shared_ptr<OwnedCallbackPayload>> port_registrations{
-      registration};
-  {
-    EventQueue::KernelEqueueEvent event{};
-    event.event.ident = 2;
-    event.event.filter = EventQueue::KERNEL_EVFILT_VIDEO_OUT;
-    event.filter.data = registration.get();
-    event.filter.owner = registration;
-    event.filter.delete_event_func = DeleteOwnedEvent;
-    Check(EventQueue::KernelAddEvent(queue, event) == OK,
-          "add owned callback event");
-  }
+	std::atomic_uint32_t stage {0};
+	std::atomic_uint32_t delete_count {0};
+	std::atomic_uint32_t destroy_count {0};
+	auto registration = std::make_shared<OwnedCallbackPayload>(stage, delete_count, destroy_count);
+	std::weak_ptr<OwnedCallbackPayload>                weak_registration = registration;
+	std::vector<std::shared_ptr<OwnedCallbackPayload>> port_registrations {registration};
+	{
+		EventQueue::KernelEqueueEvent event {};
+		event.event.ident              = 2;
+		event.event.filter             = EventQueue::KERNEL_EVFILT_VIDEO_OUT;
+		event.filter.data              = registration.get();
+		event.filter.owner             = registration;
+		event.filter.delete_event_func = DeleteOwnedEvent;
+		Check(EventQueue::KernelAddEvent(queue, event) == OK, "add owned callback event");
+	}
 
-  std::jthread close([&] {
-    Check(EventQueue::KernelDeleteEqueue(queue) == OK,
-          "delete owned callback queue");
-  });
-  while (stage.load(std::memory_order_acquire) != 1) {
-    std::this_thread::yield();
-  }
+	std::jthread close(
+	    [&] { Check(EventQueue::KernelDeleteEqueue(queue) == OK, "delete owned callback queue"); });
+	while (stage.load(std::memory_order_acquire) != 1) {
+		std::this_thread::yield();
+	}
 
-  Check(!EventQueue::KernelPinEqueue(queue),
-        "owned callback queue removed while callback blocked");
-  port_registrations.clear();
-  registration.reset();
-  Check(!weak_registration.expired(),
-        "delete callback retains detached payload");
+	Check(!EventQueue::KernelPinEqueue(queue),
+	      "owned callback queue removed while callback blocked");
+	port_registrations.clear();
+	registration.reset();
+	Check(!weak_registration.expired(), "delete callback retains detached payload");
 
-  stage.store(2, std::memory_order_release);
-  close.join();
-  Check(delete_count.load(std::memory_order_relaxed) == 1,
-        "owned callback runs exactly once");
-  Check(weak_registration.expired(),
-        "owned callback payload released with event");
-  Check(destroy_count.load(std::memory_order_relaxed) == 1,
-        "owned callback payload destroyed exactly once");
+	stage.store(2, std::memory_order_release);
+	close.join();
+	Check(delete_count.load(std::memory_order_relaxed) == 1, "owned callback runs exactly once");
+	Check(weak_registration.expired(), "owned callback payload released with event");
+	Check(destroy_count.load(std::memory_order_relaxed) == 1,
+	      "owned callback payload destroyed exactly once");
 }
 
 void TestPinnedClose() {
-  EventQueue::KernelEqueue queue = EventQueue::KERNEL_EQUEUE_INVALID;
-  Check(EventQueue::KernelCreateEqueue(&queue, "pinned-close") == OK,
-        "create pinned queue");
+	EventQueue::KernelEqueue queue = EventQueue::KERNEL_EQUEUE_INVALID;
+	Check(EventQueue::KernelCreateEqueue(&queue, "pinned-close") == OK, "create pinned queue");
 
-  std::atomic_uint32_t delete_count{0};
-  EventQueue::KernelEqueueEvent event{};
-  event.event.ident = 1;
-  event.event.filter = EventQueue::KERNEL_EVFILT_VIDEO_OUT;
-  event.filter.data = &delete_count;
-  event.filter.delete_event_func = CountDeletedEvent;
-  Check(EventQueue::KernelAddEvent(queue, event) == OK, "add callback event");
+	std::atomic_uint32_t          delete_count {0};
+	EventQueue::KernelEqueueEvent event {};
+	event.event.ident              = 1;
+	event.event.filter             = EventQueue::KERNEL_EVFILT_VIDEO_OUT;
+	event.filter.data              = &delete_count;
+	event.filter.delete_event_func = CountDeletedEvent;
+	Check(EventQueue::KernelAddEvent(queue, event) == OK, "add callback event");
 
-  auto owner = EventQueue::KernelPinEqueue(queue);
-  Check(owner != nullptr, "pin live queue");
-  Check(EventQueue::KernelDeleteEqueue(queue) == OK, "delete pinned queue");
-  Check(delete_count.load(std::memory_order_relaxed) == 1,
-        "close invokes callback once");
-  Check(!EventQueue::KernelPinEqueue(queue), "deleted queue leaves registry");
-  Check(EventQueue::KernelTriggerEvent(queue, 1,
-                                       EventQueue::KERNEL_EVFILT_VIDEO_OUT,
-                                       nullptr) == KERNEL_ERROR_EBADF,
-        "stale trigger rejected");
-  Check(EventQueue::KernelDeleteEqueue(queue) == KERNEL_ERROR_EBADF,
-        "second queue delete rejected");
+	auto owner = EventQueue::KernelPinEqueue(queue);
+	Check(owner != nullptr, "pin live queue");
+	Check(EventQueue::KernelDeleteEqueue(queue) == OK, "delete pinned queue");
+	Check(delete_count.load(std::memory_order_relaxed) == 1, "close invokes callback once");
+	Check(!EventQueue::KernelPinEqueue(queue), "deleted queue leaves registry");
+	Check(EventQueue::KernelTriggerEvent(queue, 1, EventQueue::KERNEL_EVFILT_VIDEO_OUT, nullptr) ==
+	          KERNEL_ERROR_EBADF,
+	      "stale trigger rejected");
+	Check(EventQueue::KernelDeleteEqueue(queue) == KERNEL_ERROR_EBADF,
+	      "second queue delete rejected");
 
-  owner.reset();
-  Check(delete_count.load(std::memory_order_relaxed) == 1,
-        "deferred destruction does not repeat callback");
+	owner.reset();
+	Check(delete_count.load(std::memory_order_relaxed) == 1,
+	      "deferred destruction does not repeat callback");
 }
 
 void TestStaleHandleNeverAliasesNewQueue() {
-  EventQueue::KernelEqueue stale = EventQueue::KERNEL_EQUEUE_INVALID;
-  Check(EventQueue::KernelCreateEqueue(&stale, "stale-handle") == OK,
-        "create stale queue");
-  Check(EventQueue::KernelDeleteEqueue(stale) == OK, "delete stale queue");
+	EventQueue::KernelEqueue stale = EventQueue::KERNEL_EQUEUE_INVALID;
+	Check(EventQueue::KernelCreateEqueue(&stale, "stale-handle") == OK, "create stale queue");
+	Check(EventQueue::KernelDeleteEqueue(stale) == OK, "delete stale queue");
 
-  EventQueue::KernelEqueue replacement = EventQueue::KERNEL_EQUEUE_INVALID;
-  Check(EventQueue::KernelCreateEqueue(&replacement, "replacement") == OK,
-        "create replacement queue");
-  Check(stale != replacement, "queue handles are never recycled");
-  Check(!EventQueue::KernelPinEqueue(stale), "stale handle does not pin");
-  Check(EventQueue::KernelAddUserEvent(stale, 11) == KERNEL_ERROR_EBADF,
-        "stale handle cannot mutate replacement");
-  Check(EventQueue::KernelAddUserEvent(replacement, 11) == OK,
-        "replacement handle remains valid");
-  Check(EventQueue::KernelTriggerUserEvent(stale, 11, nullptr) ==
-            KERNEL_ERROR_EBADF,
-        "stale handle cannot trigger replacement");
-  Check(EventQueue::KernelTriggerUserEvent(replacement, 11, nullptr) == OK,
-        "replacement event triggers");
-  Check(EventQueue::KernelDeleteEqueue(replacement) == OK,
-        "delete replacement queue");
+	EventQueue::KernelEqueue replacement = EventQueue::KERNEL_EQUEUE_INVALID;
+	Check(EventQueue::KernelCreateEqueue(&replacement, "replacement") == OK,
+	      "create replacement queue");
+	Check(stale != replacement, "queue handles are never recycled");
+	Check(!EventQueue::KernelPinEqueue(stale), "stale handle does not pin");
+	Check(EventQueue::KernelAddUserEvent(stale, 11) == KERNEL_ERROR_EBADF,
+	      "stale handle cannot mutate replacement");
+	Check(EventQueue::KernelAddUserEvent(replacement, 11) == OK,
+	      "replacement handle remains valid");
+	Check(EventQueue::KernelTriggerUserEvent(stale, 11, nullptr) == KERNEL_ERROR_EBADF,
+	      "stale handle cannot trigger replacement");
+	Check(EventQueue::KernelTriggerUserEvent(replacement, 11, nullptr) == OK,
+	      "replacement event triggers");
+	Check(EventQueue::KernelDeleteEqueue(replacement) == OK, "delete replacement queue");
 }
 
 void TestConcurrentCloseCallback() {
-  for (uint32_t iteration = 0; iteration < 64; iteration++) {
-    EventQueue::KernelEqueue queue = EventQueue::KERNEL_EQUEUE_INVALID;
-    Check(EventQueue::KernelCreateEqueue(&queue, "callback-race") == OK,
-          "create callback race queue");
+	for (uint32_t iteration = 0; iteration < 64; iteration++) {
+		EventQueue::KernelEqueue queue = EventQueue::KERNEL_EQUEUE_INVALID;
+		Check(EventQueue::KernelCreateEqueue(&queue, "callback-race") == OK,
+		      "create callback race queue");
 
-    std::atomic_uint32_t delete_count{0};
-    EventQueue::KernelEqueueEvent callback_event{};
-    callback_event.event.ident = 9;
-    callback_event.event.filter = EventQueue::KERNEL_EVFILT_GRAPHICS;
-    callback_event.filter.data = &delete_count;
-    callback_event.filter.delete_event_func = CountDeletedEvent;
-    Check(EventQueue::KernelAddEvent(queue, callback_event) == OK,
-          "add callback race event");
+		std::atomic_uint32_t          delete_count {0};
+		EventQueue::KernelEqueueEvent callback_event {};
+		callback_event.event.ident              = 9;
+		callback_event.event.filter             = EventQueue::KERNEL_EVFILT_GRAPHICS;
+		callback_event.filter.data              = &delete_count;
+		callback_event.filter.delete_event_func = CountDeletedEvent;
+		Check(EventQueue::KernelAddEvent(queue, callback_event) == OK, "add callback race event");
 
-    std::atomic_bool start{false};
-    std::jthread trigger([&] {
-      while (!start.load(std::memory_order_acquire)) {
-        std::this_thread::yield();
-      }
-      for (uint32_t i = 0; i < 256; i++) {
-        CheckConcurrentResult(
-            EventQueue::KernelTriggerEvent(
-                queue, 9, EventQueue::KERNEL_EVFILT_GRAPHICS, nullptr),
-            "callback race trigger result");
-      }
-    });
+		std::atomic_bool start {false};
+		std::jthread     trigger([&] {
+			while (!start.load(std::memory_order_acquire)) {
+				std::this_thread::yield();
+			}
+			for (uint32_t i = 0; i < 256; i++) {
+				CheckConcurrentResult(EventQueue::KernelTriggerEvent(
+				                          queue, 9, EventQueue::KERNEL_EVFILT_GRAPHICS, nullptr),
+				                      "callback race trigger result");
+			}
+		});
 
-    start.store(true, std::memory_order_release);
-    Check(EventQueue::KernelDeleteEqueue(queue) == OK,
-          "callback race queue delete");
-    trigger.join();
-    Check(delete_count.load(std::memory_order_relaxed) == 1,
-          "concurrent close invokes callback exactly once");
-  }
+		start.store(true, std::memory_order_release);
+		Check(EventQueue::KernelDeleteEqueue(queue) == OK, "callback race queue delete");
+		trigger.join();
+		Check(delete_count.load(std::memory_order_relaxed) == 1,
+		      "concurrent close invokes callback exactly once");
+	}
 }
 
 void TestConcurrentDelete() {
-  for (uint32_t iteration = 0; iteration < 64; iteration++) {
-    EventQueue::KernelEqueue queue = EventQueue::KERNEL_EQUEUE_INVALID;
-    Check(EventQueue::KernelCreateEqueue(&queue, "concurrent-delete") == OK,
-          "create concurrent queue");
-    EventQueue::KernelEqueueEvent event{};
-    event.event.ident = 7;
-    event.event.filter = EventQueue::KERNEL_EVFILT_USER;
-    Check(EventQueue::KernelAddEvent(queue, event) == OK,
-          "add concurrent event");
+	for (uint32_t iteration = 0; iteration < 64; iteration++) {
+		EventQueue::KernelEqueue queue = EventQueue::KERNEL_EQUEUE_INVALID;
+		Check(EventQueue::KernelCreateEqueue(&queue, "concurrent-delete") == OK,
+		      "create concurrent queue");
+		EventQueue::KernelEqueueEvent event {};
+		event.event.ident  = 7;
+		event.event.filter = EventQueue::KERNEL_EVFILT_USER;
+		Check(EventQueue::KernelAddEvent(queue, event) == OK, "add concurrent event");
 
-    std::atomic_bool start{false};
-    std::jthread mutate([&] {
-      while (!start.load(std::memory_order_acquire)) {
-        std::this_thread::yield();
-      }
-      for (uint32_t i = 0; i < 64; i++) {
-        CheckConcurrentResult(EventQueue::KernelAddEvent(queue, event),
-                              "concurrent add result");
-        CheckConcurrentResult(
-            EventQueue::KernelTriggerEvent(
-                queue, 7, EventQueue::KERNEL_EVFILT_USER, nullptr),
-            "concurrent trigger result");
-        CheckConcurrentResult(EventQueue::KernelDeleteEvent(
-                                  queue, 7, EventQueue::KERNEL_EVFILT_USER),
-                              "concurrent event delete result");
-      }
-    });
-    std::jthread trigger([&] {
-      while (!start.load(std::memory_order_acquire)) {
-        std::this_thread::yield();
-      }
-      for (uint32_t i = 0; i < 128; i++) {
-        CheckConcurrentResult(
-            EventQueue::KernelTriggerEvent(
-                queue, 7, EventQueue::KERNEL_EVFILT_USER, nullptr),
-            "parallel trigger result");
-      }
-    });
+		std::atomic_bool start {false};
+		std::jthread     mutate([&] {
+			while (!start.load(std::memory_order_acquire)) {
+				std::this_thread::yield();
+			}
+			for (uint32_t i = 0; i < 64; i++) {
+				CheckConcurrentResult(EventQueue::KernelAddEvent(queue, event),
+				                      "concurrent add result");
+				CheckConcurrentResult(EventQueue::KernelTriggerEvent(
+				                          queue, 7, EventQueue::KERNEL_EVFILT_USER, nullptr),
+				                      "concurrent trigger result");
+				CheckConcurrentResult(
+				    EventQueue::KernelDeleteEvent(queue, 7, EventQueue::KERNEL_EVFILT_USER),
+				    "concurrent event delete result");
+			}
+		});
+		std::jthread     trigger([&] {
+			while (!start.load(std::memory_order_acquire)) {
+				std::this_thread::yield();
+			}
+			for (uint32_t i = 0; i < 128; i++) {
+				CheckConcurrentResult(EventQueue::KernelTriggerEvent(
+				                          queue, 7, EventQueue::KERNEL_EVFILT_USER, nullptr),
+				                      "parallel trigger result");
+			}
+		});
 
-    start.store(true, std::memory_order_release);
-    Check(EventQueue::KernelDeleteEqueue(queue) == OK,
-          "concurrent queue delete");
-    mutate.join();
-    trigger.join();
-    Check(!EventQueue::KernelPinEqueue(queue),
-          "concurrent queue removed from registry");
-  }
+		start.store(true, std::memory_order_release);
+		Check(EventQueue::KernelDeleteEqueue(queue) == OK, "concurrent queue delete");
+		mutate.join();
+		trigger.join();
+		Check(!EventQueue::KernelPinEqueue(queue), "concurrent queue removed from registry");
+	}
 }
 
 } // namespace
 
 int main() {
-  TestDuplicateAddPreservesEventState();
-  TestCallbackStateOutlivesPort();
-  TestCallbackOwnsPayload();
-  TestPinnedClose();
-  TestStaleHandleNeverAliasesNewQueue();
-  TestConcurrentCloseCallback();
-  TestConcurrentDelete();
-  std::printf("EventQueueLifetimeTests: all cases passed\n");
-  return 0;
+	TestDuplicateAddPreservesEventState();
+	TestCallbackStateOutlivesPort();
+	TestCallbackOwnsPayload();
+	TestPinnedClose();
+	TestStaleHandleNeverAliasesNewQueue();
+	TestConcurrentCloseCallback();
+	TestConcurrentDelete();
+	std::printf("EventQueueLifetimeTests: all cases passed\n");
+	return 0;
 }