Merge branch 'main' into Red_Theme

CMakeLists.txt

@@ -1119,6 +1119,10 @@ if (APPLE)
         set(MVK_BUNDLE_PATH "Resources/vulkan/icd.d")
         set_property(TARGET shadps4 APPEND PROPERTY BUILD_RPATH "@executable_path/../${MVK_BUNDLE_PATH}")
         set(MVK_DST ${CMAKE_CURRENT_BINARY_DIR}/shadps4.app/Contents/${MVK_BUNDLE_PATH})
+
+        add_custom_command(
+            OUTPUT ${MVK_DST}
+            COMMAND ${CMAKE_COMMAND} -E make_directory ${MVK_DST})
     else()
         set_property(TARGET shadps4 APPEND PROPERTY BUILD_RPATH "@executable_path")
         set(MVK_DST ${CMAKE_CURRENT_BINARY_DIR})
@@ -1129,9 +1133,6 @@ if (APPLE)
     set(MVK_ICD_SRC ${CMAKE_CURRENT_SOURCE_DIR}/externals/MoltenVK/MoltenVK/MoltenVK/icd/MoltenVK_icd.json)
     set(MVK_ICD_DST ${MVK_DST}/MoltenVK_icd.json)
 
-    add_custom_command(
-        OUTPUT ${MVK_DST}
-        COMMAND ${CMAKE_COMMAND} -E make_directory ${MVK_DST})
     add_custom_command(
         OUTPUT ${MVK_ICD_DST}
         DEPENDS ${MVK_ICD_SRC} ${MVK_DST}
@@ -1146,17 +1147,13 @@ if (APPLE)
 
     if (ARCHITECTURE STREQUAL "x86_64")
         # Reserve system-managed memory space.
-        target_link_options(shadps4 PRIVATE -Wl,-no_pie,-no_fixup_chains,-no_huge,-pagezero_size,0x4000,-segaddr,TCB_SPACE,0x4000,-segaddr,SYSTEM_MANAGED,0x400000,-segaddr,SYSTEM_RESERVED,0x7FFFFC000,-image_base,0x20000000000)
+        target_link_options(shadps4 PRIVATE -Wl,-ld_classic,-no_pie,-no_fixup_chains,-no_huge,-pagezero_size,0x4000,-segaddr,TCB_SPACE,0x4000,-segaddr,SYSTEM_MANAGED,0x400000,-segaddr,SYSTEM_RESERVED,0x7FFFFC000,-image_base,0x20000000000)
     endif()
 
     # Replacement for std::chrono::time_zone
     target_link_libraries(shadps4 PRIVATE date::date-tz)
 endif()
 
-if (NOT ENABLE_QT_GUI)
-  target_link_libraries(shadps4 PRIVATE SDL3::SDL3)
-endif()
-
 if (ENABLE_QT_GUI)
     target_link_libraries(shadps4 PRIVATE Qt6::Widgets Qt6::Concurrent Qt6::Network Qt6::Multimedia)
     add_definitions(-DENABLE_QT_GUI)

externals/MoltenVK/MoltenVK

@@ -1 +1 @@
-Subproject commit 3a0b07a24a4a681ffe70b461b1f4333b2729e2ef
+Subproject commit 00abd384ce01cbd439045905d2fa6cf799dfa2f6

externals/MoltenVK/SPIRV-Cross

@@ -1 +1 @@
-Subproject commit 969e75f7cc0718774231d029f9d52fa87d4ae1b2
+Subproject commit 1a69a919fa302e92b337594bd0a8aaea61037d91

src/core/signals.cpp

@@ -11,6 +11,7 @@
 #include <windows.h>
 #else
 #include <csignal>
+#include <pthread.h>
 #ifdef ARCH_X86_64
 #include <Zydis/Formatter.h>
 #endif

src/core/tls.cpp

@@ -51,7 +51,7 @@ Tcb* GetTcbBase() {
 // Apple x86_64
 
 // Reserve space in the 32-bit address range for allocating TCB pages.
-asm(".zerofill TCB_SPACE,TCB_SPACE,__guest_system,0x3FC000");
+asm(".zerofill TCB_SPACE,TCB_SPACE,__tcb_space,0x3FC000");
 
 struct LdtPage {
     void* tcb;

src/shader_recompiler/backend/spirv/spirv_emit_context.cpp

@@ -299,8 +299,7 @@ void EmitContext::DefineInterpolatedAttribs() {
     // Iterate all input attributes, load them and manually interpolate.
     for (s32 i = 0; i < runtime_info.fs_info.num_inputs; i++) {
         const auto& input = runtime_info.fs_info.inputs[i];
-        const u32 semantic = input.param_index;
+        auto& params = input_params[i];
-        auto& params = input_params[semantic];
         if (input.is_flat || params.is_loaded) {
             continue;
         }
@@ -318,7 +317,7 @@ void EmitContext::DefineInterpolatedAttribs() {
         const Id p10_y{OpVectorTimesScalar(F32[4], p10, bary_coord_y)};
         const Id p20_z{OpVectorTimesScalar(F32[4], p20, bary_coord_z)};
         params.id = OpFAdd(F32[4], p0, OpFAdd(F32[4], p10_y, p20_z));
-        Name(params.id, fmt::format("fs_in_attr{}", semantic));
+        Name(params.id, fmt::format("fs_in_attr{}", i));
         params.is_loaded = true;
     }
 }
@@ -427,25 +426,28 @@ void EmitContext::DefineInputs() {
         }
         for (s32 i = 0; i < runtime_info.fs_info.num_inputs; i++) {
             const auto& input = runtime_info.fs_info.inputs[i];
-            const u32 semantic = input.param_index;
-            ASSERT(semantic < IR::NumParams);
             if (input.IsDefault()) {
-                input_params[semantic] = {
+                input_params[i] = {
-                    MakeDefaultValue(*this, input.default_value), input_f32, F32[1], 4, false, true,
+                    .id = MakeDefaultValue(*this, input.default_value),
+                    .pointer_type = input_f32,
+                    .component_type = F32[1],
+                    .num_components = 4,
+                    .is_integer = false,
+                    .is_loaded = true,
                 };
                 continue;
             }
-            const IR::Attribute param{IR::Attribute::Param0 + input.param_index};
+            const IR::Attribute param{IR::Attribute::Param0 + i};
             const u32 num_components = info.loads.NumComponents(param);
             const Id type{F32[num_components]};
             Id attr_id{};
             if (profile.needs_manual_interpolation && !input.is_flat) {
-                attr_id = DefineInput(TypeArray(type, ConstU32(3U)), semantic);
+                attr_id = DefineInput(TypeArray(type, ConstU32(3U)), input.param_index);
                 Decorate(attr_id, spv::Decoration::PerVertexKHR);
-                Name(attr_id, fmt::format("fs_in_attr{}_p", semantic));
+                Name(attr_id, fmt::format("fs_in_attr{}_p", i));
             } else {
-                attr_id = DefineInput(type, semantic);
+                attr_id = DefineInput(type, input.param_index);
-                Name(attr_id, fmt::format("fs_in_attr{}", semantic));
+                Name(attr_id, fmt::format("fs_in_attr{}", i));
 
                 if (input.is_flat) {
                     Decorate(attr_id, spv::Decoration::Flat);
@@ -453,7 +455,7 @@ void EmitContext::DefineInputs() {
                     Decorate(attr_id, spv::Decoration::NoPerspective);
                 }
             }
-            input_params[semantic] =
+            input_params[i] =
                 GetAttributeInfo(AmdGpu::NumberFormat::Float, attr_id, num_components, false);
         }
         break;

src/shader_recompiler/frontend/translate/vector_interpolation.cpp

@@ -22,15 +22,17 @@ void Translator::EmitVectorInterpolation(const GcnInst& inst) {
 // VINTRP
 
 void Translator::V_INTERP_P2_F32(const GcnInst& inst) {
-    const auto& attr = runtime_info.fs_info.inputs.at(inst.control.vintrp.attr);
+    const u32 attr_index = inst.control.vintrp.attr;
-    info.interp_qualifiers[attr.param_index] = vgpr_to_interp[inst.src[0].code];
+    const auto& attr = runtime_info.fs_info.inputs.at(attr_index);
-    const IR::Attribute attrib{IR::Attribute::Param0 + attr.param_index};
+    info.interp_qualifiers[attr_index] = vgpr_to_interp[inst.src[0].code];
+    const IR::Attribute attrib{IR::Attribute::Param0 + attr_index};
     SetDst(inst.dst[0], ir.GetAttribute(attrib, inst.control.vintrp.chan));
 }
 
 void Translator::V_INTERP_MOV_F32(const GcnInst& inst) {
-    const auto& attr = runtime_info.fs_info.inputs.at(inst.control.vintrp.attr);
+    const u32 attr_index = inst.control.vintrp.attr;
-    const IR::Attribute attrib{IR::Attribute::Param0 + attr.param_index};
+    const auto& attr = runtime_info.fs_info.inputs.at(attr_index);
+    const IR::Attribute attrib{IR::Attribute::Param0 + attr_index};
     SetDst(inst.dst[0], ir.GetAttribute(attrib, inst.control.vintrp.chan));
 }
 

src/shader_recompiler/ir/passes/shared_memory_to_storage_pass.cpp

@@ -34,11 +34,11 @@ void SharedMemoryToStoragePass(IR::Program& program, const RuntimeInfo& runtime_
     if (program.info.stage != Stage::Compute) {
         return;
     }
-    // Only perform the transform if the host shared memory is insufficient
+    // Only perform the transform if there is shared memory and either host shared memory is
-    // or the device does not support VK_KHR_workgroup_memory_explicit_layout
+    // insufficient or the device does not support VK_KHR_workgroup_memory_explicit_layout
     const u32 shared_memory_size = runtime_info.cs_info.shared_memory_size;
-    if (shared_memory_size <= profile.max_shared_memory_size &&
+    if (shared_memory_size == 0 || (shared_memory_size <= profile.max_shared_memory_size &&
-        profile.supports_workgroup_explicit_memory_layout) {
+                                    profile.supports_workgroup_explicit_memory_layout)) {
         return;
     }
     // Add buffer binding for shared memory storage buffer.

src/video_core/renderer_vulkan/vk_instance.cpp

@@ -445,7 +445,25 @@ bool Instance::CreateDevice() {
                 workgroup_memory_explicit_layout_features.workgroupMemoryExplicitLayout16BitAccess,
         },
 #ifdef __APPLE__
-        portability_features,
+        vk::PhysicalDevicePortabilitySubsetFeaturesKHR{
+            .constantAlphaColorBlendFactors = portability_features.constantAlphaColorBlendFactors,
+            .events = portability_features.events,
+            .imageViewFormatReinterpretation = portability_features.imageViewFormatReinterpretation,
+            .imageViewFormatSwizzle = portability_features.imageViewFormatSwizzle,
+            .imageView2DOn3DImage = portability_features.imageView2DOn3DImage,
+            .multisampleArrayImage = portability_features.multisampleArrayImage,
+            .mutableComparisonSamplers = portability_features.mutableComparisonSamplers,
+            .pointPolygons = portability_features.pointPolygons,
+            .samplerMipLodBias = portability_features.samplerMipLodBias,
+            .separateStencilMaskRef = portability_features.separateStencilMaskRef,
+            .shaderSampleRateInterpolationFunctions =
+                portability_features.shaderSampleRateInterpolationFunctions,
+            .tessellationIsolines = portability_features.tessellationIsolines,
+            .tessellationPointMode = portability_features.tessellationPointMode,
+            .triangleFans = portability_features.triangleFans,
+            .vertexAttributeAccessBeyondStride =
+                portability_features.vertexAttributeAccessBeyondStride,
+        },
 #endif
     };
 

src/video_core/texture_cache/image.cpp

@@ -319,7 +319,8 @@ void Image::CopyImage(const Image& image) {
     auto cmdbuf = scheduler->CommandBuffer();
 
     boost::container::small_vector<vk::ImageCopy, 14> image_copy{};
-    for (u32 m = 0; m < image.info.resources.levels; ++m) {
+    const u32 num_mips = std::min(image.info.resources.levels, info.resources.levels);
+    for (u32 m = 0; m < num_mips; ++m) {
         const auto mip_w = std::max(image.info.size.width >> m, 1u);
         const auto mip_h = std::max(image.info.size.height >> m, 1u);
         const auto mip_d = std::max(image.info.size.depth >> m, 1u);

src/video_core/texture_cache/texture_cache.cpp

@@ -461,9 +461,9 @@ ImageView& TextureCache::FindDepthTarget(BaseDesc& desc) {
     const ImageId image_id = FindImage(desc);
     Image& image = slot_images[image_id];
     image.flags |= ImageFlagBits::GpuModified;
-    image.flags &= ~ImageFlagBits::Dirty;
     image.usage.depth_target = 1u;
     image.usage.stencil = image.info.HasStencil();
+    UpdateImage(image_id);
 
     // Register meta data for this depth buffer
     if (!(image.flags & ImageFlagBits::MetaRegistered)) {