early-access version 3501

README.md

@@ -1,7 +1,7 @@
 yuzu emulator early access
 =============
 
-This is the source code for early-access 3500.
+This is the source code for early-access 3501.
 
 ## Legal Notice
 

src/CMakeLists.txt

@@ -126,6 +126,17 @@ else()
         add_compile_options("-stdlib=libc++")
     endif()
 
+    # GCC bugs
+    if (CMAKE_CXX_COMPILER_VERSION VERSION_GREATER_EQUAL "12" AND CMAKE_CXX_COMPILER_ID STREQUAL "GNU")
+        # These diagnostics would be great if they worked, but are just completely broken
+        # and produce bogus errors on external libraries like fmt.
+        add_compile_options(
+            -Wno-array-bounds
+            -Wno-stringop-overread
+            -Wno-stringop-overflow
+        )
+    endif()
+
     # Set file offset size to 64 bits.
     #
     # On modern Unixes, this is typically already the case. The lone exception is

src/common/intrusive_red_black_tree.h

@@ -96,10 +96,6 @@ public:
             return m_node == rhs.m_node;
         }
 
-        constexpr bool operator!=(const Iterator& rhs) const {
-            return !(*this == rhs);
-        }
-
         constexpr pointer operator->() const {
             return m_node;
         }
@@ -324,10 +320,6 @@ public:
             return m_impl == rhs.m_impl;
         }
 
-        constexpr bool operator!=(const Iterator& rhs) const {
-            return !(*this == rhs);
-        }
-
         constexpr pointer operator->() const {
             return Traits::GetParent(std::addressof(*m_impl));
         }

src/common/typed_address.h

@@ -116,7 +116,6 @@ public:
 
     // Comparison operators.
     constexpr bool operator==(const TypedAddress&) const = default;
-    constexpr bool operator!=(const TypedAddress&) const = default;
     constexpr auto operator<=>(const TypedAddress&) const = default;
 
     // For convenience, also define comparison operators versus uint64_t.
@@ -124,10 +123,6 @@ public:
         return m_address == rhs;
     }
 
-    constexpr inline bool operator!=(uint64_t rhs) const {
-        return m_address != rhs;
-    }
-
     // Allow getting the address explicitly, for use in accessors.
     constexpr inline uint64_t GetValue() const {
         return m_address;

src/core/internal_network/socket_proxy.h

@@ -16,9 +16,6 @@ namespace Network {
 
 class ProxySocket : public SocketBase {
 public:
-    YUZU_NON_COPYABLE(ProxySocket);
-    YUZU_NON_MOVEABLE(ProxySocket);
-
     explicit ProxySocket(RoomNetwork& room_network_) noexcept;
     ~ProxySocket() override;
 

src/core/internal_network/sockets.h

@@ -36,13 +36,10 @@ public:
 
     SocketBase() = default;
     explicit SocketBase(SOCKET fd_) : fd{fd_} {}
-
     virtual ~SocketBase() = default;
 
-    virtual SocketBase& operator=(const SocketBase&) = delete;
+    YUZU_NON_COPYABLE(SocketBase);
-
+    YUZU_NON_MOVEABLE(SocketBase);
-    // Avoid closing sockets implicitly
-    virtual SocketBase& operator=(SocketBase&&) noexcept = delete;
 
     virtual Errno Initialize(Domain domain, Type type, Protocol protocol) = 0;
 
@@ -109,14 +106,8 @@ public:
 
     ~Socket() override;
 
-    Socket(const Socket&) = delete;
-    Socket& operator=(const Socket&) = delete;
-
     Socket(Socket&& rhs) noexcept;
 
-    // Avoid closing sockets implicitly
-    Socket& operator=(Socket&&) noexcept = delete;
-
     Errno Initialize(Domain domain, Type type, Protocol protocol) override;
 
     Errno Close() override;

src/shader_recompiler/backend/glsl/emit_glsl_image.cpp

@@ -152,7 +152,8 @@ std::string ImageGatherSubpixelOffset(const IR::TextureInstInfo& info, std::stri
         return fmt::format("{}+vec2(0.001953125)/vec2(textureSize({}, 0))", coords, texture);
     case TextureType::ColorArray2D:
     case TextureType::ColorCube:
-        return fmt::format("vec3({0}.xy+vec2(0.001953125)/vec2(textureSize({1}, 0)),{0}.z)", coords, texture);
+        return fmt::format("vec3({0}.xy+vec2(0.001953125)/vec2(textureSize({1}, 0)),{0}.z)", coords,
+                           texture);
     default:
         return std::string{coords};
     }

src/shader_recompiler/backend/spirv/emit_spirv_image.cpp

@@ -266,30 +266,21 @@ Id ImageGatherSubpixelOffset(EmitContext& ctx, const IR::TextureInstInfo& info,
                              Id coords) {
     // Apply a subpixel offset of 1/512 the texel size of the texture to ensure same rounding on
     // AMD hardware as on Maxwell or other Nvidia architectures.
-    const auto calculate_offset{[&](size_t dim) -> std::array<Id, 2> {
+    const auto calculate_coords{[&](size_t dim) {
         const Id nudge{ctx.Const(0x1p-9f)};
         const Id image_size{ctx.OpImageQuerySizeLod(ctx.U32[dim], texture, ctx.u32_zero_value)};
-        const Id offset_x{ctx.OpFDiv(
+        Id offset{dim == 2 ? ctx.ConstantComposite(ctx.F32[dim], nudge, nudge)
-            ctx.F32[1], nudge,
+                           : ctx.ConstantComposite(ctx.F32[dim], nudge, nudge, ctx.f32_zero_value)};
-            ctx.OpConvertUToF(ctx.F32[1], ctx.OpCompositeExtract(ctx.U32[1], image_size, 0)))};
+        offset = ctx.OpFDiv(ctx.F32[dim], offset, ctx.OpConvertUToF(ctx.F32[dim], image_size));
-        const Id offset_y{ctx.OpFDiv(
+        return ctx.OpFAdd(ctx.F32[dim], coords, offset);
-            ctx.F32[1], nudge,
-            ctx.OpConvertUToF(ctx.F32[1], ctx.OpCompositeExtract(ctx.U32[1], image_size, 1)))};
-        return {ctx.OpFAdd(ctx.F32[1], ctx.OpCompositeExtract(ctx.F32[1], coords, 0), offset_x),
-                ctx.OpFAdd(ctx.F32[1], ctx.OpCompositeExtract(ctx.F32[1], coords, 1), offset_y)};
     }};
     switch (info.type) {
     case TextureType::Color2D:
-    case TextureType::Color2DRect: {
+    case TextureType::Color2DRect:
-        const auto offset{calculate_offset(2)};
+        return calculate_coords(2);
-        return ctx.OpCompositeConstruct(ctx.F32[2], offset[0], offset[1]);
-    }
     case TextureType::ColorArray2D:
-    case TextureType::ColorCube: {
+    case TextureType::ColorCube:
-        const auto offset{calculate_offset(3)};
+        return calculate_coords(3);
-        return ctx.OpCompositeConstruct(ctx.F32[3], offset[0], offset[1],
-                                        ctx.OpCompositeExtract(ctx.F32[1], coords, 2));
-    }
     default:
         return coords;
     }

src/shader_recompiler/profile.h

@@ -52,9 +52,9 @@ struct Profile {
     bool need_declared_frag_colors{};
     /// Prevents fast math optimizations that may cause inaccuracies
     bool need_fastmath_off{};
-    /// Some GPU vendors use a lower fixed point format of 16.8 when calculating pixel coordinates
+    /// Some GPU vendors use a different rounding precision when calculating texture pixel
-    /// in the ImageGather instruction than the Maxwell architecture does. Applying an offset does
+    /// coordinates with the 16.8 format in the ImageGather instruction than the Maxwell
-    /// fix this mismatching rounding behaviour.
+    /// architecture. Applying an offset does fix this mismatching rounding behaviour.
     bool need_gather_subpixel_offset{};
 
     /// OpFClamp is broken and OpFMax + OpFMin should be used instead

src/video_core/renderer_opengl/gl_device.cpp

@@ -169,7 +169,6 @@ Device::Device(Core::Frontend::EmuWindow& emu_window) {
     has_draw_texture = GLAD_GL_NV_draw_texture;
     warp_size_potentially_larger_than_guest = !is_nvidia && !is_intel;
     need_fastmath_off = is_nvidia;
-    need_gather_subpixel_offset = is_amd;
     can_report_memory = GLAD_GL_NVX_gpu_memory_info;
 
     // At the moment of writing this, only Nvidia's driver optimizes BufferSubData on exclusive

src/video_core/renderer_opengl/gl_device.h

@@ -160,10 +160,6 @@ public:
         return need_fastmath_off;
     }
 
-    bool NeedsGatherSubpixelOffset() const {
-        return need_gather_subpixel_offset;
-    }
-
     bool HasCbufFtouBug() const {
         return has_cbuf_ftou_bug;
     }
@@ -180,6 +176,10 @@ public:
         return vendor_name == "ATI Technologies Inc.";
     }
 
+    bool IsIntel() const {
+        return vendor_name == "Intel";
+    }
+
     bool CanReportMemoryUsage() const {
         return can_report_memory;
     }
@@ -229,7 +229,6 @@ private:
     bool has_draw_texture{};
     bool warp_size_potentially_larger_than_guest{};
     bool need_fastmath_off{};
-    bool need_gather_subpixel_offset{};
     bool has_cbuf_ftou_bug{};
     bool has_bool_ref_bug{};
     bool can_report_memory{};

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -218,7 +218,7 @@ ShaderCache::ShaderCache(RasterizerOpenGL& rasterizer_, Core::Frontend::EmuWindo
           .lower_left_origin_mode = true,
           .need_declared_frag_colors = true,
           .need_fastmath_off = device.NeedsFastmathOff(),
-          .need_gather_subpixel_offset = device.NeedsGatherSubpixelOffset(),
+          .need_gather_subpixel_offset = device.IsAmd() || device.IsIntel(),
 
           .has_broken_spirv_clamp = true,
           .has_broken_unsigned_image_offsets = true,

src/video_core/renderer_vulkan/vk_pipeline_cache.cpp

@@ -329,7 +329,9 @@ PipelineCache::PipelineCache(RasterizerVulkan& rasterizer_, const Device& device
 
         .lower_left_origin_mode = false,
         .need_declared_frag_colors = false,
-        .need_gather_subpixel_offset = device.NeedsGatherSubpixelOffset(),
+        .need_gather_subpixel_offset = driver_id == VK_DRIVER_ID_AMD_PROPRIETARY ||
+                                       driver_id == VK_DRIVER_ID_INTEL_PROPRIETARY_WINDOWS ||
+                                       driver_id == VK_DRIVER_ID_INTEL_OPEN_SOURCE_MESA,
 
         .has_broken_spirv_clamp = driver_id == VK_DRIVER_ID_INTEL_PROPRIETARY_WINDOWS,
         .has_broken_spirv_position_input = driver_id == VK_DRIVER_ID_QUALCOMM_PROPRIETARY,

src/video_core/vulkan_common/vulkan_device.cpp

@@ -431,7 +431,6 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
                         "AMD GCN4 and earlier have broken VK_IMAGE_CREATE_CUBE_COMPATIBLE_BIT");
             has_broken_cube_compatibility = true;
         }
-        need_gather_subpixel_offset = true;
     }
     if (extensions.sampler_filter_minmax && is_amd) {
         // Disable ext_sampler_filter_minmax on AMD GCN4 and lower as it is broken.

src/video_core/vulkan_common/vulkan_device.h

@@ -554,10 +554,6 @@ public:
         return features.robustness2.nullDescriptor;
     }
 
-    bool NeedsGatherSubpixelOffset() const {
-        return need_gather_subpixel_offset;
-    }
-
     u32 GetMaxVertexInputAttributes() const {
         return properties.properties.limits.maxVertexInputAttributes;
     }
@@ -668,7 +664,6 @@ private:
     bool must_emulate_bgr565{};             ///< Emulates BGR565 by swizzling RGB565 format.
     bool dynamic_state3_blending{};         ///< Has all blending features of dynamic_state3.
     bool dynamic_state3_enables{};          ///< Has all enables features of dynamic_state3.
-    bool need_gather_subpixel_offset{};     ///< Needs offset at ImageGather for correct rounding.
     u64 device_access_memory{};             ///< Total size of device local memory in bytes.
     u32 sets_per_pool{};                    ///< Sets per Description Pool
 

src/web_service/verify_login.cpp

@@ -21,7 +21,7 @@ bool VerifyLogin(const std::string& host, const std::string& username, const std
         return username.empty();
     }
 
-    return username == *iter;
+    return *iter == username;
 }
 
 } // namespace WebService