diff --git a/README.md b/README.md index 6593af197..5329d425c 100755 --- a/README.md +++ b/README.md @@ -1,7 +1,7 @@ yuzu emulator early access ============= -This is the source code for early-access 1286. +This is the source code for early-access 1287. ## Legal Notice diff --git a/src/common/logging/backend.cpp b/src/common/logging/backend.cpp index 631f64d05..2d4d2e9e7 100755 --- a/src/common/logging/backend.cpp +++ b/src/common/logging/backend.cpp @@ -145,10 +145,18 @@ void ColorConsoleBackend::Write(const Entry& entry) { PrintColoredMessage(entry); } -// _SH_DENYWR allows read only access to the file for other programs. -// It is #defined to 0 on other platforms -FileBackend::FileBackend(const std::string& filename) - : file(filename, "w", _SH_DENYWR), bytes_written(0) {} +FileBackend::FileBackend(const std::string& filename) : bytes_written(0) { + if (Common::FS::Exists(filename + ".old.txt")) { + Common::FS::Delete(filename + ".old.txt"); + } + if (Common::FS::Exists(filename)) { + Common::FS::Rename(filename, filename + ".old.txt"); + } + + // _SH_DENYWR allows read only access to the file for other programs. + // It is #defined to 0 on other platforms + file = Common::FS::IOFile(filename, "w", _SH_DENYWR); +} void FileBackend::Write(const Entry& entry) { // prevent logs from going over the maximum size (in case its spamming and the user doesn't diff --git a/src/core/hle/service/nvflinger/buffer_queue.cpp b/src/core/hle/service/nvflinger/buffer_queue.cpp index 0e6bde9f5..c68905e19 100755 --- a/src/core/hle/service/nvflinger/buffer_queue.cpp +++ b/src/core/hle/service/nvflinger/buffer_queue.cpp @@ -26,10 +26,10 @@ void BufferQueue::SetPreallocatedBuffer(u32 slot, const IGBPBuffer& igbp_buffer) LOG_WARNING(Service, "Adding graphics buffer {}", slot); { - std::unique_lock lock{queue_mutex}; + std::unique_lock lock{free_buffers_mutex}; free_buffers.push_back(slot); } - condition.notify_one(); + free_buffers_condition.notify_one(); buffers[slot] = { .slot = slot, @@ -48,8 +48,8 @@ std::optional> BufferQueue::Dequeue u32 height) { // Wait for first request before trying to dequeue { - std::unique_lock lock{queue_mutex}; - condition.wait(lock, [this] { return !free_buffers.empty() || !is_connect; }); + std::unique_lock lock{free_buffers_mutex}; + free_buffers_condition.wait(lock, [this] { return !free_buffers.empty() || !is_connect; }); } if (!is_connect) { @@ -58,7 +58,7 @@ std::optional> BufferQueue::Dequeue return std::nullopt; } - std::unique_lock lock{queue_mutex}; + std::unique_lock lock{free_buffers_mutex}; auto f_itr = free_buffers.begin(); auto slot = buffers.size(); @@ -100,6 +100,7 @@ void BufferQueue::QueueBuffer(u32 slot, BufferTransformFlags transform, buffers[slot].crop_rect = crop_rect; buffers[slot].swap_interval = swap_interval; buffers[slot].multi_fence = multi_fence; + std::unique_lock lock{queue_sequence_mutex}; queue_sequence.push_back(slot); } @@ -113,15 +114,16 @@ void BufferQueue::CancelBuffer(u32 slot, const Service::Nvidia::MultiFence& mult buffers[slot].swap_interval = 0; { - std::unique_lock lock{queue_mutex}; + std::unique_lock lock{free_buffers_mutex}; free_buffers.push_back(slot); } - condition.notify_one(); + free_buffers_condition.notify_one(); buffer_wait_event.writable->Signal(); } std::optional> BufferQueue::AcquireBuffer() { + std::unique_lock lock{queue_sequence_mutex}; std::size_t buffer_slot = buffers.size(); // Iterate to find a queued buffer matching the requested slot. while (buffer_slot == buffers.size() && !queue_sequence.empty()) { @@ -147,25 +149,29 @@ void BufferQueue::ReleaseBuffer(u32 slot) { buffers[slot].status = Buffer::Status::Free; { - std::unique_lock lock{queue_mutex}; + std::unique_lock lock{free_buffers_mutex}; free_buffers.push_back(slot); } - condition.notify_one(); + free_buffers_condition.notify_one(); buffer_wait_event.writable->Signal(); } void BufferQueue::Connect() { + std::unique_lock lock{queue_sequence_mutex}; queue_sequence.clear(); is_connect = true; } void BufferQueue::Disconnect() { buffers.fill({}); - queue_sequence.clear(); + { + std::unique_lock lock{queue_sequence_mutex}; + queue_sequence.clear(); + } buffer_wait_event.writable->Signal(); is_connect = false; - condition.notify_one(); + free_buffers_condition.notify_one(); } u32 BufferQueue::Query(QueryType type) { diff --git a/src/core/hle/service/nvflinger/buffer_queue.h b/src/core/hle/service/nvflinger/buffer_queue.h index a2f60d9eb..ad7469277 100755 --- a/src/core/hle/service/nvflinger/buffer_queue.h +++ b/src/core/hle/service/nvflinger/buffer_queue.h @@ -129,8 +129,10 @@ private: std::list queue_sequence; Kernel::EventPair buffer_wait_event; - std::mutex queue_mutex; - std::condition_variable condition; + std::mutex free_buffers_mutex; + std::condition_variable free_buffers_condition; + + std::mutex queue_sequence_mutex; }; } // namespace Service::NVFlinger diff --git a/src/video_core/compatible_formats.cpp b/src/video_core/compatible_formats.cpp index 1619d8664..acf2668dc 100755 --- a/src/video_core/compatible_formats.cpp +++ b/src/video_core/compatible_formats.cpp @@ -10,9 +10,7 @@ #include "video_core/surface.h" namespace VideoCore::Surface { - namespace { - using Table = std::array, MaxPixelFormat>; // Compatibility table taken from Table 3.X.2 in: @@ -233,10 +231,13 @@ constexpr Table MakeCopyTable() { EnableRange(copy, COPY_CLASS_64_BITS); return copy; } - } // Anonymous namespace -bool IsViewCompatible(PixelFormat format_a, PixelFormat format_b) { +bool IsViewCompatible(PixelFormat format_a, PixelFormat format_b, bool broken_views) { + if (broken_views) { + // If format views are broken, only accept formats that are identical. + return format_a == format_b; + } static constexpr Table TABLE = MakeViewTable(); return IsSupported(TABLE, format_a, format_b); } diff --git a/src/video_core/compatible_formats.h b/src/video_core/compatible_formats.h index b5eb03bea..9a0522988 100755 --- a/src/video_core/compatible_formats.h +++ b/src/video_core/compatible_formats.h @@ -8,7 +8,7 @@ namespace VideoCore::Surface { -bool IsViewCompatible(PixelFormat format_a, PixelFormat format_b); +bool IsViewCompatible(PixelFormat format_a, PixelFormat format_b, bool broken_views); bool IsCopyCompatible(PixelFormat format_a, PixelFormat format_b); diff --git a/src/video_core/renderer_opengl/gl_device.cpp b/src/video_core/renderer_opengl/gl_device.cpp index b24179d59..81b71edfb 100755 --- a/src/video_core/renderer_opengl/gl_device.cpp +++ b/src/video_core/renderer_opengl/gl_device.cpp @@ -208,6 +208,7 @@ Device::Device() const bool is_nvidia = vendor == "NVIDIA Corporation"; const bool is_amd = vendor == "ATI Technologies Inc."; + const bool is_intel = vendor == "Intel"; bool disable_fast_buffer_sub_data = false; if (is_nvidia && version == "4.6.0 NVIDIA 443.24") { @@ -231,6 +232,7 @@ Device::Device() has_variable_aoffi = TestVariableAoffi(); has_component_indexing_bug = is_amd; has_precise_bug = TestPreciseBug(); + has_broken_texture_view_formats = is_amd || is_intel; has_nv_viewport_array2 = GLAD_GL_NV_viewport_array2; has_vertex_buffer_unified_memory = GLAD_GL_NV_vertex_buffer_unified_memory; has_debugging_tool_attached = IsDebugToolAttached(extensions); @@ -248,6 +250,8 @@ Device::Device() LOG_INFO(Render_OpenGL, "Renderer_VariableAOFFI: {}", has_variable_aoffi); LOG_INFO(Render_OpenGL, "Renderer_ComponentIndexingBug: {}", has_component_indexing_bug); LOG_INFO(Render_OpenGL, "Renderer_PreciseBug: {}", has_precise_bug); + LOG_INFO(Render_OpenGL, "Renderer_BrokenTextureViewFormats: {}", + has_broken_texture_view_formats); if (Settings::values.use_assembly_shaders.GetValue() && !use_assembly_shaders) { LOG_ERROR(Render_OpenGL, "Assembly shaders enabled but not supported"); diff --git a/src/video_core/renderer_opengl/gl_device.h b/src/video_core/renderer_opengl/gl_device.h index 13e66846c..3e79d1e37 100755 --- a/src/video_core/renderer_opengl/gl_device.h +++ b/src/video_core/renderer_opengl/gl_device.h @@ -96,6 +96,10 @@ public: return has_precise_bug; } + bool HasBrokenTextureViewFormats() const { + return has_broken_texture_view_formats; + } + bool HasFastBufferSubData() const { return has_fast_buffer_sub_data; } @@ -137,6 +141,7 @@ private: bool has_variable_aoffi{}; bool has_component_indexing_bug{}; bool has_precise_bug{}; + bool has_broken_texture_view_formats{}; bool has_fast_buffer_sub_data{}; bool has_nv_viewport_array2{}; bool has_debugging_tool_attached{}; diff --git a/src/video_core/renderer_opengl/gl_texture_cache.cpp b/src/video_core/renderer_opengl/gl_texture_cache.cpp index 4c690418c..546cb6d00 100755 --- a/src/video_core/renderer_opengl/gl_texture_cache.cpp +++ b/src/video_core/renderer_opengl/gl_texture_cache.cpp @@ -61,100 +61,99 @@ struct FormatTuple { GLenum internal_format; GLenum format = GL_NONE; GLenum type = GL_NONE; - GLenum store_format = internal_format; }; constexpr std::array FORMAT_TABLE = {{ - {GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV}, // A8B8G8R8_UNORM - {GL_RGBA8_SNORM, GL_RGBA, GL_BYTE}, // A8B8G8R8_SNORM - {GL_RGBA8I, GL_RGBA_INTEGER, GL_BYTE}, // A8B8G8R8_SINT - {GL_RGBA8UI, GL_RGBA_INTEGER, GL_UNSIGNED_BYTE}, // A8B8G8R8_UINT - {GL_RGB565, GL_RGB, GL_UNSIGNED_SHORT_5_6_5}, // R5G6B5_UNORM - {GL_RGB565, GL_RGB, GL_UNSIGNED_SHORT_5_6_5_REV}, // B5G6R5_UNORM - {GL_RGB5_A1, GL_BGRA, GL_UNSIGNED_SHORT_1_5_5_5_REV}, // A1R5G5B5_UNORM - {GL_RGB10_A2, GL_RGBA, GL_UNSIGNED_INT_2_10_10_10_REV}, // A2B10G10R10_UNORM - {GL_RGB10_A2UI, GL_RGBA_INTEGER, GL_UNSIGNED_INT_2_10_10_10_REV}, // A2B10G10R10_UINT - {GL_RGB5_A1, GL_RGBA, GL_UNSIGNED_SHORT_1_5_5_5_REV}, // A1B5G5R5_UNORM - {GL_R8, GL_RED, GL_UNSIGNED_BYTE}, // R8_UNORM - {GL_R8_SNORM, GL_RED, GL_BYTE}, // R8_SNORM - {GL_R8I, GL_RED_INTEGER, GL_BYTE}, // R8_SINT - {GL_R8UI, GL_RED_INTEGER, GL_UNSIGNED_BYTE}, // R8_UINT - {GL_RGBA16F, GL_RGBA, GL_HALF_FLOAT}, // R16G16B16A16_FLOAT - {GL_RGBA16, GL_RGBA, GL_UNSIGNED_SHORT}, // R16G16B16A16_UNORM - {GL_RGBA16_SNORM, GL_RGBA, GL_SHORT}, // R16G16B16A16_SNORM - {GL_RGBA16I, GL_RGBA_INTEGER, GL_SHORT}, // R16G16B16A16_SINT - {GL_RGBA16UI, GL_RGBA_INTEGER, GL_UNSIGNED_SHORT}, // R16G16B16A16_UINT - {GL_R11F_G11F_B10F, GL_RGB, GL_UNSIGNED_INT_10F_11F_11F_REV}, // B10G11R11_FLOAT - {GL_RGBA32UI, GL_RGBA_INTEGER, GL_UNSIGNED_INT}, // R32G32B32A32_UINT - {GL_COMPRESSED_RGBA_S3TC_DXT1_EXT}, // BC1_RGBA_UNORM - {GL_COMPRESSED_RGBA_S3TC_DXT3_EXT}, // BC2_UNORM - {GL_COMPRESSED_RGBA_S3TC_DXT5_EXT}, // BC3_UNORM - {GL_COMPRESSED_RED_RGTC1}, // BC4_UNORM - {GL_COMPRESSED_SIGNED_RED_RGTC1}, // BC4_SNORM - {GL_COMPRESSED_RG_RGTC2}, // BC5_UNORM - {GL_COMPRESSED_SIGNED_RG_RGTC2}, // BC5_SNORM - {GL_COMPRESSED_RGBA_BPTC_UNORM}, // BC7_UNORM - {GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT}, // BC6H_UFLOAT - {GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT}, // BC6H_SFLOAT - {GL_COMPRESSED_RGBA_ASTC_4x4_KHR}, // ASTC_2D_4X4_UNORM - {GL_RGBA8, GL_BGRA, GL_UNSIGNED_BYTE}, // B8G8R8A8_UNORM - {GL_RGBA32F, GL_RGBA, GL_FLOAT}, // R32G32B32A32_FLOAT - {GL_RGBA32I, GL_RGBA_INTEGER, GL_INT}, // R32G32B32A32_SINT - {GL_RG32F, GL_RG, GL_FLOAT}, // R32G32_FLOAT - {GL_RG32I, GL_RG_INTEGER, GL_INT}, // R32G32_SINT - {GL_R32F, GL_RED, GL_FLOAT}, // R32_FLOAT - {GL_R16F, GL_RED, GL_HALF_FLOAT}, // R16_FLOAT - {GL_R16, GL_RED, GL_UNSIGNED_SHORT}, // R16_UNORM - {GL_R16_SNORM, GL_RED, GL_SHORT}, // R16_SNORM - {GL_R16UI, GL_RED_INTEGER, GL_UNSIGNED_SHORT}, // R16_UINT - {GL_R16I, GL_RED_INTEGER, GL_SHORT}, // R16_SINT - {GL_RG16, GL_RG, GL_UNSIGNED_SHORT}, // R16G16_UNORM - {GL_RG16F, GL_RG, GL_HALF_FLOAT}, // R16G16_FLOAT - {GL_RG16UI, GL_RG_INTEGER, GL_UNSIGNED_SHORT}, // R16G16_UINT - {GL_RG16I, GL_RG_INTEGER, GL_SHORT}, // R16G16_SINT - {GL_RG16_SNORM, GL_RG, GL_SHORT}, // R16G16_SNORM - {GL_RGB32F, GL_RGB, GL_FLOAT}, // R32G32B32_FLOAT - {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, GL_RGBA8}, // A8B8G8R8_SRGB - {GL_RG8, GL_RG, GL_UNSIGNED_BYTE}, // R8G8_UNORM - {GL_RG8_SNORM, GL_RG, GL_BYTE}, // R8G8_SNORM - {GL_RG8I, GL_RG_INTEGER, GL_BYTE}, // R8G8_SINT - {GL_RG8UI, GL_RG_INTEGER, GL_UNSIGNED_BYTE}, // R8G8_UINT - {GL_RG32UI, GL_RG_INTEGER, GL_UNSIGNED_INT}, // R32G32_UINT - {GL_RGB16F, GL_RGBA, GL_HALF_FLOAT}, // R16G16B16X16_FLOAT - {GL_R32UI, GL_RED_INTEGER, GL_UNSIGNED_INT}, // R32_UINT - {GL_R32I, GL_RED_INTEGER, GL_INT}, // R32_SINT - {GL_COMPRESSED_RGBA_ASTC_8x8_KHR}, // ASTC_2D_8X8_UNORM - {GL_COMPRESSED_RGBA_ASTC_8x5_KHR}, // ASTC_2D_8X5_UNORM - {GL_COMPRESSED_RGBA_ASTC_5x4_KHR}, // ASTC_2D_5X4_UNORM - {GL_SRGB8_ALPHA8, GL_BGRA, GL_UNSIGNED_BYTE, GL_RGBA8}, // B8G8R8A8_UNORM - {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT}, // BC1_RGBA_SRGB - {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT}, // BC2_SRGB - {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT}, // BC3_SRGB - {GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM}, // BC7_SRGB - {GL_RGBA4, GL_RGBA, GL_UNSIGNED_SHORT_4_4_4_4_REV}, // A4B4G4R4_UNORM - {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_4x4_KHR}, // ASTC_2D_4X4_SRGB - {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x8_KHR}, // ASTC_2D_8X8_SRGB - {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x5_KHR}, // ASTC_2D_8X5_SRGB - {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_5x4_KHR}, // ASTC_2D_5X4_SRGB - {GL_COMPRESSED_RGBA_ASTC_5x5_KHR}, // ASTC_2D_5X5_UNORM - {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_5x5_KHR}, // ASTC_2D_5X5_SRGB - {GL_COMPRESSED_RGBA_ASTC_10x8_KHR}, // ASTC_2D_10X8_UNORM - {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_10x8_KHR}, // ASTC_2D_10X8_SRGB - {GL_COMPRESSED_RGBA_ASTC_6x6_KHR}, // ASTC_2D_6X6_UNORM - {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_6x6_KHR}, // ASTC_2D_6X6_SRGB - {GL_COMPRESSED_RGBA_ASTC_10x10_KHR}, // ASTC_2D_10X10_UNORM - {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_10x10_KHR}, // ASTC_2D_10X10_SRGB - {GL_COMPRESSED_RGBA_ASTC_12x12_KHR}, // ASTC_2D_12X12_UNORM - {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_12x12_KHR}, // ASTC_2D_12X12_SRGB - {GL_COMPRESSED_RGBA_ASTC_8x6_KHR}, // ASTC_2D_8X6_UNORM - {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x6_KHR}, // ASTC_2D_8X6_SRGB - {GL_COMPRESSED_RGBA_ASTC_6x5_KHR}, // ASTC_2D_6X5_UNORM - {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_6x5_KHR}, // ASTC_2D_6X5_SRGB - {GL_RGB9_E5, GL_RGB, GL_UNSIGNED_INT_5_9_9_9_REV}, // E5B9G9R9_FLOAT - {GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT}, // D32_FLOAT - {GL_DEPTH_COMPONENT16, GL_DEPTH_COMPONENT, GL_UNSIGNED_SHORT}, // D16_UNORM - {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8}, // D24_UNORM_S8_UINT - {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8}, // S8_UINT_D24_UNORM + {GL_RGBA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV}, // A8B8G8R8_UNORM + {GL_RGBA8_SNORM, GL_RGBA, GL_BYTE}, // A8B8G8R8_SNORM + {GL_RGBA8I, GL_RGBA_INTEGER, GL_BYTE}, // A8B8G8R8_SINT + {GL_RGBA8UI, GL_RGBA_INTEGER, GL_UNSIGNED_BYTE}, // A8B8G8R8_UINT + {GL_RGB565, GL_RGB, GL_UNSIGNED_SHORT_5_6_5}, // R5G6B5_UNORM + {GL_RGB565, GL_RGB, GL_UNSIGNED_SHORT_5_6_5_REV}, // B5G6R5_UNORM + {GL_RGB5_A1, GL_BGRA, GL_UNSIGNED_SHORT_1_5_5_5_REV}, // A1R5G5B5_UNORM + {GL_RGB10_A2, GL_RGBA, GL_UNSIGNED_INT_2_10_10_10_REV}, // A2B10G10R10_UNORM + {GL_RGB10_A2UI, GL_RGBA_INTEGER, GL_UNSIGNED_INT_2_10_10_10_REV}, // A2B10G10R10_UINT + {GL_RGB5_A1, GL_RGBA, GL_UNSIGNED_SHORT_1_5_5_5_REV}, // A1B5G5R5_UNORM + {GL_R8, GL_RED, GL_UNSIGNED_BYTE}, // R8_UNORM + {GL_R8_SNORM, GL_RED, GL_BYTE}, // R8_SNORM + {GL_R8I, GL_RED_INTEGER, GL_BYTE}, // R8_SINT + {GL_R8UI, GL_RED_INTEGER, GL_UNSIGNED_BYTE}, // R8_UINT + {GL_RGBA16F, GL_RGBA, GL_HALF_FLOAT}, // R16G16B16A16_FLOAT + {GL_RGBA16, GL_RGBA, GL_UNSIGNED_SHORT}, // R16G16B16A16_UNORM + {GL_RGBA16_SNORM, GL_RGBA, GL_SHORT}, // R16G16B16A16_SNORM + {GL_RGBA16I, GL_RGBA_INTEGER, GL_SHORT}, // R16G16B16A16_SINT + {GL_RGBA16UI, GL_RGBA_INTEGER, GL_UNSIGNED_SHORT}, // R16G16B16A16_UINT + {GL_R11F_G11F_B10F, GL_RGB, GL_UNSIGNED_INT_10F_11F_11F_REV}, // B10G11R11_FLOAT + {GL_RGBA32UI, GL_RGBA_INTEGER, GL_UNSIGNED_INT}, // R32G32B32A32_UINT + {GL_COMPRESSED_RGBA_S3TC_DXT1_EXT}, // BC1_RGBA_UNORM + {GL_COMPRESSED_RGBA_S3TC_DXT3_EXT}, // BC2_UNORM + {GL_COMPRESSED_RGBA_S3TC_DXT5_EXT}, // BC3_UNORM + {GL_COMPRESSED_RED_RGTC1}, // BC4_UNORM + {GL_COMPRESSED_SIGNED_RED_RGTC1}, // BC4_SNORM + {GL_COMPRESSED_RG_RGTC2}, // BC5_UNORM + {GL_COMPRESSED_SIGNED_RG_RGTC2}, // BC5_SNORM + {GL_COMPRESSED_RGBA_BPTC_UNORM}, // BC7_UNORM + {GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT}, // BC6H_UFLOAT + {GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT}, // BC6H_SFLOAT + {GL_COMPRESSED_RGBA_ASTC_4x4_KHR}, // ASTC_2D_4X4_UNORM + {GL_RGBA8, GL_BGRA, GL_UNSIGNED_BYTE}, // B8G8R8A8_UNORM + {GL_RGBA32F, GL_RGBA, GL_FLOAT}, // R32G32B32A32_FLOAT + {GL_RGBA32I, GL_RGBA_INTEGER, GL_INT}, // R32G32B32A32_SINT + {GL_RG32F, GL_RG, GL_FLOAT}, // R32G32_FLOAT + {GL_RG32I, GL_RG_INTEGER, GL_INT}, // R32G32_SINT + {GL_R32F, GL_RED, GL_FLOAT}, // R32_FLOAT + {GL_R16F, GL_RED, GL_HALF_FLOAT}, // R16_FLOAT + {GL_R16, GL_RED, GL_UNSIGNED_SHORT}, // R16_UNORM + {GL_R16_SNORM, GL_RED, GL_SHORT}, // R16_SNORM + {GL_R16UI, GL_RED_INTEGER, GL_UNSIGNED_SHORT}, // R16_UINT + {GL_R16I, GL_RED_INTEGER, GL_SHORT}, // R16_SINT + {GL_RG16, GL_RG, GL_UNSIGNED_SHORT}, // R16G16_UNORM + {GL_RG16F, GL_RG, GL_HALF_FLOAT}, // R16G16_FLOAT + {GL_RG16UI, GL_RG_INTEGER, GL_UNSIGNED_SHORT}, // R16G16_UINT + {GL_RG16I, GL_RG_INTEGER, GL_SHORT}, // R16G16_SINT + {GL_RG16_SNORM, GL_RG, GL_SHORT}, // R16G16_SNORM + {GL_RGB32F, GL_RGB, GL_FLOAT}, // R32G32B32_FLOAT + {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV}, // A8B8G8R8_SRGB + {GL_RG8, GL_RG, GL_UNSIGNED_BYTE}, // R8G8_UNORM + {GL_RG8_SNORM, GL_RG, GL_BYTE}, // R8G8_SNORM + {GL_RG8I, GL_RG_INTEGER, GL_BYTE}, // R8G8_SINT + {GL_RG8UI, GL_RG_INTEGER, GL_UNSIGNED_BYTE}, // R8G8_UINT + {GL_RG32UI, GL_RG_INTEGER, GL_UNSIGNED_INT}, // R32G32_UINT + {GL_RGB16F, GL_RGBA, GL_HALF_FLOAT}, // R16G16B16X16_FLOAT + {GL_R32UI, GL_RED_INTEGER, GL_UNSIGNED_INT}, // R32_UINT + {GL_R32I, GL_RED_INTEGER, GL_INT}, // R32_SINT + {GL_COMPRESSED_RGBA_ASTC_8x8_KHR}, // ASTC_2D_8X8_UNORM + {GL_COMPRESSED_RGBA_ASTC_8x5_KHR}, // ASTC_2D_8X5_UNORM + {GL_COMPRESSED_RGBA_ASTC_5x4_KHR}, // ASTC_2D_5X4_UNORM + {GL_SRGB8_ALPHA8, GL_BGRA, GL_UNSIGNED_BYTE}, // B8G8R8A8_UNORM + {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT}, // BC1_RGBA_SRGB + {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT}, // BC2_SRGB + {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT}, // BC3_SRGB + {GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM}, // BC7_SRGB + {GL_RGBA4, GL_RGBA, GL_UNSIGNED_SHORT_4_4_4_4_REV}, // A4B4G4R4_UNORM + {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_4x4_KHR}, // ASTC_2D_4X4_SRGB + {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x8_KHR}, // ASTC_2D_8X8_SRGB + {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x5_KHR}, // ASTC_2D_8X5_SRGB + {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_5x4_KHR}, // ASTC_2D_5X4_SRGB + {GL_COMPRESSED_RGBA_ASTC_5x5_KHR}, // ASTC_2D_5X5_UNORM + {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_5x5_KHR}, // ASTC_2D_5X5_SRGB + {GL_COMPRESSED_RGBA_ASTC_10x8_KHR}, // ASTC_2D_10X8_UNORM + {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_10x8_KHR}, // ASTC_2D_10X8_SRGB + {GL_COMPRESSED_RGBA_ASTC_6x6_KHR}, // ASTC_2D_6X6_UNORM + {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_6x6_KHR}, // ASTC_2D_6X6_SRGB + {GL_COMPRESSED_RGBA_ASTC_10x10_KHR}, // ASTC_2D_10X10_UNORM + {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_10x10_KHR}, // ASTC_2D_10X10_SRGB + {GL_COMPRESSED_RGBA_ASTC_12x12_KHR}, // ASTC_2D_12X12_UNORM + {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_12x12_KHR}, // ASTC_2D_12X12_SRGB + {GL_COMPRESSED_RGBA_ASTC_8x6_KHR}, // ASTC_2D_8X6_UNORM + {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_8x6_KHR}, // ASTC_2D_8X6_SRGB + {GL_COMPRESSED_RGBA_ASTC_6x5_KHR}, // ASTC_2D_6X5_UNORM + {GL_COMPRESSED_SRGB8_ALPHA8_ASTC_6x5_KHR}, // ASTC_2D_6X5_SRGB + {GL_RGB9_E5, GL_RGB, GL_UNSIGNED_INT_5_9_9_9_REV}, // E5B9G9R9_FLOAT + {GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT}, // D32_FLOAT + {GL_DEPTH_COMPONENT16, GL_DEPTH_COMPONENT, GL_UNSIGNED_SHORT}, // D16_UNORM + {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8}, // D24_UNORM_S8_UINT + {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8}, // S8_UINT_D24_UNORM {GL_DEPTH32F_STENCIL8, GL_DEPTH_STENCIL, GL_FLOAT_32_UNSIGNED_INT_24_8_REV}, // D32_FLOAT_S8_UINT }}; @@ -431,6 +430,8 @@ TextureCacheRuntime::TextureCacheRuntime(const Device& device_, ProgramManager& format_properties[i].emplace(format, properties); } } + has_broken_texture_view_formats = device.HasBrokenTextureViewFormats(); + null_image_1d_array.Create(GL_TEXTURE_1D_ARRAY); null_image_cube_array.Create(GL_TEXTURE_CUBE_MAP_ARRAY); null_image_3d.Create(GL_TEXTURE_3D); @@ -651,13 +652,11 @@ Image::Image(TextureCacheRuntime& runtime, const VideoCommon::ImageInfo& info_, if (IsConverted(runtime.device, info.format, info.type)) { flags |= ImageFlagBits::Converted; gl_internal_format = IsPixelFormatSRGB(info.format) ? GL_SRGB8_ALPHA8 : GL_RGBA8; - gl_store_format = GL_RGBA8; gl_format = GL_RGBA; gl_type = GL_UNSIGNED_INT_8_8_8_8_REV; } else { const auto& tuple = GetFormatTuple(info.format); gl_internal_format = tuple.internal_format; - gl_store_format = tuple.store_format; gl_format = tuple.format; gl_type = tuple.type; } @@ -677,23 +676,23 @@ Image::Image(TextureCacheRuntime& runtime, const VideoCommon::ImageInfo& info_, } switch (target) { case GL_TEXTURE_1D_ARRAY: - glTextureStorage2D(handle, num_levels, gl_store_format, width, num_layers); + glTextureStorage2D(handle, num_levels, gl_internal_format, width, num_layers); break; case GL_TEXTURE_2D_ARRAY: - glTextureStorage3D(handle, num_levels, gl_store_format, width, height, num_layers); + glTextureStorage3D(handle, num_levels, gl_internal_format, width, height, num_layers); break; case GL_TEXTURE_2D_MULTISAMPLE_ARRAY: { // TODO: Where should 'fixedsamplelocations' come from? const auto [samples_x, samples_y] = SamplesLog2(info.num_samples); - glTextureStorage3DMultisample(handle, num_samples, gl_store_format, width >> samples_x, + glTextureStorage3DMultisample(handle, num_samples, gl_internal_format, width >> samples_x, height >> samples_y, num_layers, GL_FALSE); break; } case GL_TEXTURE_RECTANGLE: - glTextureStorage2D(handle, num_levels, gl_store_format, width, height); + glTextureStorage2D(handle, num_levels, gl_internal_format, width, height); break; case GL_TEXTURE_3D: - glTextureStorage3D(handle, num_levels, gl_store_format, width, height, depth); + glTextureStorage3D(handle, num_levels, gl_internal_format, width, height, depth); break; case GL_TEXTURE_BUFFER: buffer.Create(); diff --git a/src/video_core/renderer_opengl/gl_texture_cache.h b/src/video_core/renderer_opengl/gl_texture_cache.h index 04193e31e..15b7c3676 100755 --- a/src/video_core/renderer_opengl/gl_texture_cache.h +++ b/src/video_core/renderer_opengl/gl_texture_cache.h @@ -96,6 +96,10 @@ public: FormatProperties FormatInfo(VideoCommon::ImageType type, GLenum internal_format) const; + bool HasBrokenTextureViewFormats() const noexcept { + return has_broken_texture_view_formats; + } + private: struct StagingBuffers { explicit StagingBuffers(GLenum storage_flags_, GLenum map_flags_); @@ -120,6 +124,7 @@ private: UtilShaders util_shaders; std::array, 3> format_properties; + bool has_broken_texture_view_formats = false; StagingBuffers upload_buffers{GL_MAP_WRITE_BIT, GL_MAP_WRITE_BIT | GL_MAP_FLUSH_EXPLICIT_BIT}; StagingBuffers download_buffers{GL_MAP_READ_BIT, GL_MAP_READ_BIT}; @@ -165,7 +170,6 @@ private: OGLTextureView store_view; OGLBuffer buffer; GLenum gl_internal_format = GL_NONE; - GLenum gl_store_format = GL_NONE; GLenum gl_format = GL_NONE; GLenum gl_type = GL_NONE; }; diff --git a/src/video_core/renderer_vulkan/vk_texture_cache.h b/src/video_core/renderer_vulkan/vk_texture_cache.h index e68a3db43..92a7aad8b 100755 --- a/src/video_core/renderer_vulkan/vk_texture_cache.h +++ b/src/video_core/renderer_vulkan/vk_texture_cache.h @@ -104,6 +104,11 @@ struct TextureCacheRuntime { } void InsertUploadMemoryBarrier() {} + + bool HasBrokenTextureViewFormats() const noexcept { + // No known Vulkan driver has broken image views + return false; + } }; class Image : public VideoCommon::ImageBase { diff --git a/src/video_core/texture_cache/image_base.cpp b/src/video_core/texture_cache/image_base.cpp index 448a05fcc..959b3f115 100755 --- a/src/video_core/texture_cache/image_base.cpp +++ b/src/video_core/texture_cache/image_base.cpp @@ -120,7 +120,9 @@ void AddImageAlias(ImageBase& lhs, ImageBase& rhs, ImageId lhs_id, ImageId rhs_i if (lhs.info.type == ImageType::Linear) { base = SubresourceBase{.level = 0, .layer = 0}; } else { - base = FindSubresource(rhs.info, lhs, rhs.gpu_addr, OPTIONS); + // We are passing relaxed formats as an option, having broken views or not won't matter + static constexpr bool broken_views = false; + base = FindSubresource(rhs.info, lhs, rhs.gpu_addr, OPTIONS, broken_views); } if (!base) { LOG_ERROR(HW_GPU, "Image alias should have been flipped"); diff --git a/src/video_core/texture_cache/image_view_base.cpp b/src/video_core/texture_cache/image_view_base.cpp index 076a4bcfd..18f72e508 100755 --- a/src/video_core/texture_cache/image_view_base.cpp +++ b/src/video_core/texture_cache/image_view_base.cpp @@ -24,7 +24,7 @@ ImageViewBase::ImageViewBase(const ImageViewInfo& info, const ImageInfo& image_i .height = std::max(image_info.size.height >> range.base.level, 1u), .depth = std::max(image_info.size.depth >> range.base.level, 1u), } { - ASSERT_MSG(VideoCore::Surface::IsViewCompatible(image_info.format, info.format), + ASSERT_MSG(VideoCore::Surface::IsViewCompatible(image_info.format, info.format, false), "Image view format {} is incompatible with image format {}", info.format, image_info.format); const bool is_async = Settings::values.use_asynchronous_gpu_emulation.GetValue(); diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h index 968059842..ad86c50b4 100755 --- a/src/video_core/texture_cache/texture_cache.h +++ b/src/video_core/texture_cache/texture_cache.h @@ -883,6 +883,7 @@ ImageId TextureCache

::FindImage(const ImageInfo& info, GPUVAddr gpu_addr, if (!cpu_addr) { return ImageId{}; } + const bool broken_views = runtime.HasBrokenTextureViewFormats(); ImageId image_id; const auto lambda = [&](ImageId existing_image_id, ImageBase& existing_image) { if (info.type == ImageType::Linear || existing_image.info.type == ImageType::Linear) { @@ -892,11 +893,11 @@ ImageId TextureCache

::FindImage(const ImageInfo& info, GPUVAddr gpu_addr, if (existing_image.gpu_addr == gpu_addr && existing.type == info.type && existing.pitch == info.pitch && IsPitchLinearSameSize(existing, info, strict_size) && - IsViewCompatible(existing.format, info.format)) { + IsViewCompatible(existing.format, info.format, broken_views)) { image_id = existing_image_id; return true; } - } else if (IsSubresource(info, existing_image, gpu_addr, options)) { + } else if (IsSubresource(info, existing_image, gpu_addr, options, broken_views)) { image_id = existing_image_id; return true; } @@ -926,6 +927,7 @@ template ImageId TextureCache

::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VAddr cpu_addr) { ImageInfo new_info = info; const size_t size_bytes = CalculateGuestSizeInBytes(new_info); + const bool broken_views = runtime.HasBrokenTextureViewFormats(); std::vector overlap_ids; std::vector left_aliased_ids; std::vector right_aliased_ids; @@ -940,7 +942,9 @@ ImageId TextureCache

::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA } return; } - const auto solution = ResolveOverlap(new_info, gpu_addr, cpu_addr, overlap, true); + static constexpr bool strict_size = true; + const std::optional solution = + ResolveOverlap(new_info, gpu_addr, cpu_addr, overlap, strict_size, broken_views); if (solution) { gpu_addr = solution->gpu_addr; cpu_addr = solution->cpu_addr; @@ -950,9 +954,10 @@ ImageId TextureCache

::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA } static constexpr auto options = RelaxedOptions::Size | RelaxedOptions::Format; const ImageBase new_image_base(new_info, gpu_addr, cpu_addr); - if (IsSubresource(new_info, overlap, gpu_addr, options)) { + if (IsSubresource(new_info, overlap, gpu_addr, options, broken_views)) { left_aliased_ids.push_back(overlap_id); - } else if (IsSubresource(overlap.info, new_image_base, overlap.gpu_addr, options)) { + } else if (IsSubresource(overlap.info, new_image_base, overlap.gpu_addr, options, + broken_views)) { right_aliased_ids.push_back(overlap_id); } }); diff --git a/src/video_core/texture_cache/util.cpp b/src/video_core/texture_cache/util.cpp index 9ed1fc007..279932778 100755 --- a/src/video_core/texture_cache/util.cpp +++ b/src/video_core/texture_cache/util.cpp @@ -1069,13 +1069,13 @@ bool IsPitchLinearSameSize(const ImageInfo& lhs, const ImageInfo& rhs, bool stri std::optional ResolveOverlap(const ImageInfo& new_info, GPUVAddr gpu_addr, VAddr cpu_addr, const ImageBase& overlap, - bool strict_size) { + bool strict_size, bool broken_views) { ASSERT(new_info.type != ImageType::Linear); ASSERT(overlap.info.type != ImageType::Linear); if (!IsLayerStrideCompatible(new_info, overlap.info)) { return std::nullopt; } - if (!IsViewCompatible(overlap.info.format, new_info.format)) { + if (!IsViewCompatible(overlap.info.format, new_info.format, broken_views)) { return std::nullopt; } if (gpu_addr == overlap.gpu_addr) { @@ -1118,14 +1118,15 @@ bool IsLayerStrideCompatible(const ImageInfo& lhs, const ImageInfo& rhs) { } std::optional FindSubresource(const ImageInfo& candidate, const ImageBase& image, - GPUVAddr candidate_addr, RelaxedOptions options) { + GPUVAddr candidate_addr, RelaxedOptions options, + bool broken_views) { const std::optional base = image.TryFindBase(candidate_addr); if (!base) { return std::nullopt; } const ImageInfo& existing = image.info; if (False(options & RelaxedOptions::Format)) { - if (!IsViewCompatible(existing.format, candidate.format)) { + if (!IsViewCompatible(existing.format, candidate.format, broken_views)) { return std::nullopt; } } @@ -1162,8 +1163,8 @@ std::optional FindSubresource(const ImageInfo& candidate, const } bool IsSubresource(const ImageInfo& candidate, const ImageBase& image, GPUVAddr candidate_addr, - RelaxedOptions options) { - return FindSubresource(candidate, image, candidate_addr, options).has_value(); + RelaxedOptions options, bool broken_views) { + return FindSubresource(candidate, image, candidate_addr, options, broken_views).has_value(); } void DeduceBlitImages(ImageInfo& dst_info, ImageInfo& src_info, const ImageBase* dst, diff --git a/src/video_core/texture_cache/util.h b/src/video_core/texture_cache/util.h index dbbbd33cd..52a9207d6 100755 --- a/src/video_core/texture_cache/util.h +++ b/src/video_core/texture_cache/util.h @@ -87,17 +87,19 @@ void SwizzleImage(Tegra::MemoryManager& gpu_memory, GPUVAddr gpu_addr, const Ima [[nodiscard]] std::optional ResolveOverlap(const ImageInfo& new_info, GPUVAddr gpu_addr, VAddr cpu_addr, const ImageBase& overlap, - bool strict_size); + bool strict_size, bool broken_views); [[nodiscard]] bool IsLayerStrideCompatible(const ImageInfo& lhs, const ImageInfo& rhs); [[nodiscard]] std::optional FindSubresource(const ImageInfo& candidate, const ImageBase& image, GPUVAddr candidate_addr, - RelaxedOptions options); + RelaxedOptions options, + bool broken_views); [[nodiscard]] bool IsSubresource(const ImageInfo& candidate, const ImageBase& image, - GPUVAddr candidate_addr, RelaxedOptions options); + GPUVAddr candidate_addr, RelaxedOptions options, + bool broken_views); void DeduceBlitImages(ImageInfo& dst_info, ImageInfo& src_info, const ImageBase* dst, const ImageBase* src);