| // Copyright 2017 The Dawn Authors |
| // |
| // Licensed under the Apache License, Version 2.0 (the "License"); |
| // you may not use this file except in compliance with the License. |
| // You may obtain a copy of the License at |
| // |
| // http://www.apache.org/licenses/LICENSE-2.0 |
| // |
| // Unless required by applicable law or agreed to in writing, software |
| // distributed under the License is distributed on an "AS IS" BASIS, |
| // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| // See the License for the specific language governing permissions and |
| // limitations under the License. |
| |
| #include "tests/unittests/validation/ValidationTest.h" |
| |
| #include "common/Constants.h" |
| #include "utils/ComboRenderPipelineDescriptor.h" |
| #include "utils/WGPUHelpers.h" |
| |
| #include <cmath> |
| #include <sstream> |
| |
| class RenderPipelineValidationTest : public ValidationTest { |
| protected: |
| void SetUp() override { |
| ValidationTest::SetUp(); |
| |
| vsModule = utils::CreateShaderModule(device, R"( |
| [[stage(vertex)]] fn main() -> [[builtin(position)]] vec4<f32> { |
| return vec4<f32>(0.0, 0.0, 0.0, 1.0); |
| })"); |
| |
| fsModule = utils::CreateShaderModule(device, R"( |
| [[stage(fragment)]] fn main() -> [[location(0)]] vec4<f32> { |
| return vec4<f32>(0.0, 1.0, 0.0, 1.0); |
| })"); |
| |
| fsModuleUint = utils::CreateShaderModule(device, R"( |
| [[stage(fragment)]] fn main() -> [[location(0)]] vec4<u32> { |
| return vec4<u32>(0u, 255u, 0u, 255u); |
| })"); |
| } |
| |
| wgpu::ShaderModule vsModule; |
| wgpu::ShaderModule fsModule; |
| wgpu::ShaderModule fsModuleUint; |
| }; |
| |
| namespace { |
| bool BlendFactorContainsSrcAlpha(const wgpu::BlendFactor& blendFactor) { |
| return blendFactor == wgpu::BlendFactor::SrcAlpha || |
| blendFactor == wgpu::BlendFactor::OneMinusSrcAlpha || |
| blendFactor == wgpu::BlendFactor::SrcAlphaSaturated; |
| } |
| } // namespace |
| |
| // Test cases where creation should succeed |
| TEST_F(RenderPipelineValidationTest, CreationSuccess) { |
| { |
| // New format |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| |
| device.CreateRenderPipeline(&descriptor); |
| } |
| } |
| |
| // Tests that depth bias parameters must not be NaN. |
| TEST_F(RenderPipelineValidationTest, DepthBiasParameterNotBeNaN) { |
| // Control case, depth bias parameters in ComboRenderPipeline default to 0 which is finite |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| descriptor.EnableDepthStencil(); |
| device.CreateRenderPipeline(&descriptor); |
| } |
| |
| // Infinite depth bias clamp is valid |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| wgpu::DepthStencilState* depthStencil = descriptor.EnableDepthStencil(); |
| depthStencil->depthBiasClamp = INFINITY; |
| device.CreateRenderPipeline(&descriptor); |
| } |
| // NAN depth bias clamp is invalid |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| wgpu::DepthStencilState* depthStencil = descriptor.EnableDepthStencil(); |
| depthStencil->depthBiasClamp = NAN; |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| |
| // Infinite depth bias slope is valid |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| wgpu::DepthStencilState* depthStencil = descriptor.EnableDepthStencil(); |
| depthStencil->depthBiasSlopeScale = INFINITY; |
| device.CreateRenderPipeline(&descriptor); |
| } |
| // NAN depth bias slope is invalid |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| wgpu::DepthStencilState* depthStencil = descriptor.EnableDepthStencil(); |
| depthStencil->depthBiasSlopeScale = NAN; |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| } |
| |
| // Tests that at least one color target state is required. |
| TEST_F(RenderPipelineValidationTest, ColorTargetStateRequired) { |
| { |
| // This one succeeds because attachment 0 is the color attachment |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| descriptor.cFragment.targetCount = 1; |
| |
| device.CreateRenderPipeline(&descriptor); |
| } |
| |
| { // Fail because lack of color target states (and depth/stencil state) |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| descriptor.cFragment.targetCount = 0; |
| |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| } |
| |
| // Tests that the color formats must be renderable. |
| TEST_F(RenderPipelineValidationTest, NonRenderableFormat) { |
| { |
| // Succeeds because RGBA8Unorm is renderable |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| descriptor.cTargets[0].format = wgpu::TextureFormat::RGBA8Unorm; |
| |
| device.CreateRenderPipeline(&descriptor); |
| } |
| |
| { |
| // Fails because RG11B10Ufloat is non-renderable |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| descriptor.cTargets[0].format = wgpu::TextureFormat::RG11B10Ufloat; |
| |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| } |
| |
| // Tests that the color formats must be blendable when blending is enabled. |
| // Those are renderable color formats with "float" capabilities in |
| // https://gpuweb.github.io/gpuweb/#plain-color-formats |
| TEST_F(RenderPipelineValidationTest, NonBlendableFormat) { |
| { |
| // Succeeds because RGBA8Unorm is blendable |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| descriptor.cTargets[0].blend = &descriptor.cBlends[0]; |
| descriptor.cTargets[0].format = wgpu::TextureFormat::RGBA8Unorm; |
| |
| device.CreateRenderPipeline(&descriptor); |
| } |
| |
| { |
| // Fails because RGBA32Float is not blendable |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| descriptor.cTargets[0].blend = &descriptor.cBlends[0]; |
| descriptor.cTargets[0].format = wgpu::TextureFormat::RGBA32Float; |
| |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| |
| { |
| // Succeeds because RGBA32Float is not blendable but blending is disabled |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| descriptor.cTargets[0].blend = nullptr; |
| descriptor.cTargets[0].format = wgpu::TextureFormat::RGBA32Float; |
| |
| device.CreateRenderPipeline(&descriptor); |
| } |
| |
| { |
| // Fails because RGBA8Uint is not blendable |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModuleUint; |
| descriptor.cTargets[0].blend = &descriptor.cBlends[0]; |
| descriptor.cTargets[0].format = wgpu::TextureFormat::RGBA8Uint; |
| |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| |
| { |
| // Succeeds because RGBA8Uint is not blendable but blending is disabled |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModuleUint; |
| descriptor.cTargets[0].blend = nullptr; |
| descriptor.cTargets[0].format = wgpu::TextureFormat::RGBA8Uint; |
| |
| device.CreateRenderPipeline(&descriptor); |
| } |
| } |
| |
| // Tests that the format of the color state descriptor must match the output of the fragment shader. |
| TEST_F(RenderPipelineValidationTest, FragmentOutputFormatCompatibility) { |
| std::array<const char*, 3> kScalarTypes = {{"f32", "i32", "u32"}}; |
| std::array<wgpu::TextureFormat, 3> kColorFormats = {{wgpu::TextureFormat::RGBA8Unorm, |
| wgpu::TextureFormat::RGBA8Sint, |
| wgpu::TextureFormat::RGBA8Uint}}; |
| |
| for (size_t i = 0; i < kScalarTypes.size(); ++i) { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| std::ostringstream stream; |
| stream << R"( |
| [[stage(fragment)]] fn main() -> [[location(0)]] vec4<)" |
| << kScalarTypes[i] << R"(> { |
| var result : vec4<)" |
| << kScalarTypes[i] << R"(>; |
| return result; |
| })"; |
| descriptor.cFragment.module = utils::CreateShaderModule(device, stream.str().c_str()); |
| |
| for (size_t j = 0; j < kColorFormats.size(); ++j) { |
| descriptor.cTargets[0].format = kColorFormats[j]; |
| if (i == j) { |
| device.CreateRenderPipeline(&descriptor); |
| } else { |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| } |
| } |
| } |
| |
| // Tests that the component count of the color state target format must be fewer than that of the |
| // fragment shader output. |
| TEST_F(RenderPipelineValidationTest, FragmentOutputComponentCountCompatibility) { |
| std::array<wgpu::TextureFormat, 3> kColorFormats = {wgpu::TextureFormat::R8Unorm, |
| wgpu::TextureFormat::RG8Unorm, |
| wgpu::TextureFormat::RGBA8Unorm}; |
| |
| std::array<wgpu::BlendFactor, 8> kBlendFactors = {wgpu::BlendFactor::Zero, |
| wgpu::BlendFactor::One, |
| wgpu::BlendFactor::SrcAlpha, |
| wgpu::BlendFactor::OneMinusSrcAlpha, |
| wgpu::BlendFactor::Src, |
| wgpu::BlendFactor::DstAlpha, |
| wgpu::BlendFactor::OneMinusDstAlpha, |
| wgpu::BlendFactor::Dst}; |
| |
| for (size_t componentCount = 1; componentCount <= 4; ++componentCount) { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| |
| std::ostringstream stream; |
| stream << R"( |
| [[stage(fragment)]] fn main() -> [[location(0)]] )"; |
| switch (componentCount) { |
| case 1: |
| stream << R"(f32 { |
| return 1.0; |
| })"; |
| break; |
| case 2: |
| stream << R"(vec2<f32> { |
| return vec2<f32>(1.0, 1.0); |
| })"; |
| break; |
| case 3: |
| stream << R"(vec3<f32> { |
| return vec3<f32>(1.0, 1.0, 1.0); |
| })"; |
| break; |
| case 4: |
| stream << R"(vec4<f32> { |
| return vec4<f32>(1.0, 1.0, 1.0, 1.0); |
| })"; |
| break; |
| default: |
| UNREACHABLE(); |
| } |
| descriptor.cFragment.module = utils::CreateShaderModule(device, stream.str().c_str()); |
| |
| for (auto colorFormat : kColorFormats) { |
| descriptor.cTargets[0].format = colorFormat; |
| |
| descriptor.cTargets[0].blend = nullptr; |
| if (componentCount >= utils::GetWGSLRenderableColorTextureComponentCount(colorFormat)) { |
| device.CreateRenderPipeline(&descriptor); |
| } else { |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| |
| descriptor.cTargets[0].blend = &descriptor.cBlends[0]; |
| |
| for (auto colorSrcFactor : kBlendFactors) { |
| descriptor.cBlends[0].color.srcFactor = colorSrcFactor; |
| for (auto colorDstFactor : kBlendFactors) { |
| descriptor.cBlends[0].color.dstFactor = colorDstFactor; |
| for (auto alphaSrcFactor : kBlendFactors) { |
| descriptor.cBlends[0].alpha.srcFactor = alphaSrcFactor; |
| for (auto alphaDstFactor : kBlendFactors) { |
| descriptor.cBlends[0].alpha.dstFactor = alphaDstFactor; |
| |
| bool valid = true; |
| if (componentCount >= |
| utils::GetWGSLRenderableColorTextureComponentCount(colorFormat)) { |
| if (BlendFactorContainsSrcAlpha( |
| descriptor.cTargets[0].blend->color.srcFactor) || |
| BlendFactorContainsSrcAlpha( |
| descriptor.cTargets[0].blend->color.dstFactor)) { |
| valid = componentCount == 4; |
| } |
| } else { |
| valid = false; |
| } |
| |
| if (valid) { |
| device.CreateRenderPipeline(&descriptor); |
| } else { |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| } |
| } |
| } |
| } |
| } |
| } |
| } |
| |
| /// Tests that the sample count of the render pipeline must be valid. |
| TEST_F(RenderPipelineValidationTest, SampleCount) { |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| descriptor.multisample.count = 4; |
| |
| device.CreateRenderPipeline(&descriptor); |
| } |
| |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| descriptor.multisample.count = 3; |
| |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| } |
| |
| // Tests that the sample count of the render pipeline must be equal to the one of every attachments |
| // in the render pass. |
| TEST_F(RenderPipelineValidationTest, SampleCountCompatibilityWithRenderPass) { |
| constexpr uint32_t kMultisampledCount = 4; |
| constexpr wgpu::TextureFormat kColorFormat = wgpu::TextureFormat::RGBA8Unorm; |
| constexpr wgpu::TextureFormat kDepthStencilFormat = wgpu::TextureFormat::Depth24PlusStencil8; |
| |
| wgpu::TextureDescriptor baseTextureDescriptor; |
| baseTextureDescriptor.size.width = 4; |
| baseTextureDescriptor.size.height = 4; |
| baseTextureDescriptor.size.depthOrArrayLayers = 1; |
| baseTextureDescriptor.mipLevelCount = 1; |
| baseTextureDescriptor.dimension = wgpu::TextureDimension::e2D; |
| baseTextureDescriptor.usage = wgpu::TextureUsage::RenderAttachment; |
| |
| utils::ComboRenderPipelineDescriptor nonMultisampledPipelineDescriptor; |
| nonMultisampledPipelineDescriptor.multisample.count = 1; |
| nonMultisampledPipelineDescriptor.vertex.module = vsModule; |
| nonMultisampledPipelineDescriptor.cFragment.module = fsModule; |
| wgpu::RenderPipeline nonMultisampledPipeline = |
| device.CreateRenderPipeline(&nonMultisampledPipelineDescriptor); |
| |
| nonMultisampledPipelineDescriptor.cFragment.targetCount = 0; |
| nonMultisampledPipelineDescriptor.EnableDepthStencil(); |
| wgpu::RenderPipeline nonMultisampledPipelineWithDepthStencilOnly = |
| device.CreateRenderPipeline(&nonMultisampledPipelineDescriptor); |
| |
| utils::ComboRenderPipelineDescriptor multisampledPipelineDescriptor; |
| multisampledPipelineDescriptor.multisample.count = kMultisampledCount; |
| multisampledPipelineDescriptor.vertex.module = vsModule; |
| multisampledPipelineDescriptor.cFragment.module = fsModule; |
| wgpu::RenderPipeline multisampledPipeline = |
| device.CreateRenderPipeline(&multisampledPipelineDescriptor); |
| |
| multisampledPipelineDescriptor.cFragment.targetCount = 0; |
| multisampledPipelineDescriptor.EnableDepthStencil(); |
| wgpu::RenderPipeline multisampledPipelineWithDepthStencilOnly = |
| device.CreateRenderPipeline(&multisampledPipelineDescriptor); |
| |
| // It is not allowed to use multisampled render pass and non-multisampled render pipeline. |
| { |
| wgpu::TextureDescriptor textureDescriptor = baseTextureDescriptor; |
| textureDescriptor.format = kColorFormat; |
| textureDescriptor.sampleCount = kMultisampledCount; |
| wgpu::Texture multisampledColorTexture = device.CreateTexture(&textureDescriptor); |
| utils::ComboRenderPassDescriptor renderPassDescriptor( |
| {multisampledColorTexture.CreateView()}); |
| |
| wgpu::CommandEncoder encoder = device.CreateCommandEncoder(); |
| wgpu::RenderPassEncoder renderPass = encoder.BeginRenderPass(&renderPassDescriptor); |
| renderPass.SetPipeline(nonMultisampledPipeline); |
| renderPass.EndPass(); |
| |
| ASSERT_DEVICE_ERROR(encoder.Finish()); |
| } |
| |
| { |
| wgpu::TextureDescriptor textureDescriptor = baseTextureDescriptor; |
| textureDescriptor.sampleCount = kMultisampledCount; |
| textureDescriptor.format = kDepthStencilFormat; |
| wgpu::Texture multisampledDepthStencilTexture = device.CreateTexture(&textureDescriptor); |
| utils::ComboRenderPassDescriptor renderPassDescriptor( |
| {}, multisampledDepthStencilTexture.CreateView()); |
| |
| wgpu::CommandEncoder encoder = device.CreateCommandEncoder(); |
| wgpu::RenderPassEncoder renderPass = encoder.BeginRenderPass(&renderPassDescriptor); |
| renderPass.SetPipeline(nonMultisampledPipelineWithDepthStencilOnly); |
| renderPass.EndPass(); |
| |
| ASSERT_DEVICE_ERROR(encoder.Finish()); |
| } |
| |
| // It is allowed to use multisampled render pass and multisampled render pipeline. |
| { |
| wgpu::TextureDescriptor textureDescriptor = baseTextureDescriptor; |
| textureDescriptor.format = kColorFormat; |
| textureDescriptor.sampleCount = kMultisampledCount; |
| wgpu::Texture multisampledColorTexture = device.CreateTexture(&textureDescriptor); |
| utils::ComboRenderPassDescriptor renderPassDescriptor( |
| {multisampledColorTexture.CreateView()}); |
| |
| wgpu::CommandEncoder encoder = device.CreateCommandEncoder(); |
| wgpu::RenderPassEncoder renderPass = encoder.BeginRenderPass(&renderPassDescriptor); |
| renderPass.SetPipeline(multisampledPipeline); |
| renderPass.EndPass(); |
| |
| encoder.Finish(); |
| } |
| |
| { |
| wgpu::TextureDescriptor textureDescriptor = baseTextureDescriptor; |
| textureDescriptor.sampleCount = kMultisampledCount; |
| textureDescriptor.format = kDepthStencilFormat; |
| wgpu::Texture multisampledDepthStencilTexture = device.CreateTexture(&textureDescriptor); |
| utils::ComboRenderPassDescriptor renderPassDescriptor( |
| {}, multisampledDepthStencilTexture.CreateView()); |
| |
| wgpu::CommandEncoder encoder = device.CreateCommandEncoder(); |
| wgpu::RenderPassEncoder renderPass = encoder.BeginRenderPass(&renderPassDescriptor); |
| renderPass.SetPipeline(multisampledPipelineWithDepthStencilOnly); |
| renderPass.EndPass(); |
| |
| encoder.Finish(); |
| } |
| |
| // It is not allowed to use non-multisampled render pass and multisampled render pipeline. |
| { |
| wgpu::TextureDescriptor textureDescriptor = baseTextureDescriptor; |
| textureDescriptor.format = kColorFormat; |
| textureDescriptor.sampleCount = 1; |
| wgpu::Texture nonMultisampledColorTexture = device.CreateTexture(&textureDescriptor); |
| utils::ComboRenderPassDescriptor nonMultisampledRenderPassDescriptor( |
| {nonMultisampledColorTexture.CreateView()}); |
| |
| wgpu::CommandEncoder encoder = device.CreateCommandEncoder(); |
| wgpu::RenderPassEncoder renderPass = |
| encoder.BeginRenderPass(&nonMultisampledRenderPassDescriptor); |
| renderPass.SetPipeline(multisampledPipeline); |
| renderPass.EndPass(); |
| |
| ASSERT_DEVICE_ERROR(encoder.Finish()); |
| } |
| |
| { |
| wgpu::TextureDescriptor textureDescriptor = baseTextureDescriptor; |
| textureDescriptor.sampleCount = 1; |
| textureDescriptor.format = kDepthStencilFormat; |
| wgpu::Texture nonMultisampledDepthStencilTexture = device.CreateTexture(&textureDescriptor); |
| utils::ComboRenderPassDescriptor renderPassDescriptor( |
| {}, nonMultisampledDepthStencilTexture.CreateView()); |
| |
| wgpu::CommandEncoder encoder = device.CreateCommandEncoder(); |
| wgpu::RenderPassEncoder renderPass = encoder.BeginRenderPass(&renderPassDescriptor); |
| renderPass.SetPipeline(multisampledPipelineWithDepthStencilOnly); |
| renderPass.EndPass(); |
| |
| ASSERT_DEVICE_ERROR(encoder.Finish()); |
| } |
| } |
| |
| // Tests that the vertex only pipeline must be used with a depth-stencil attachment only render pass |
| TEST_F(RenderPipelineValidationTest, VertexOnlyPipelineRequireDepthStencilAttachment) { |
| constexpr wgpu::TextureFormat kColorFormat = wgpu::TextureFormat::RGBA8Unorm; |
| constexpr wgpu::TextureFormat kDepthStencilFormat = wgpu::TextureFormat::Depth24PlusStencil8; |
| |
| wgpu::TextureDescriptor baseTextureDescriptor; |
| baseTextureDescriptor.size = {4, 4}; |
| baseTextureDescriptor.mipLevelCount = 1; |
| baseTextureDescriptor.dimension = wgpu::TextureDimension::e2D; |
| baseTextureDescriptor.usage = wgpu::TextureUsage::RenderAttachment; |
| |
| wgpu::TextureDescriptor colorTextureDescriptor = baseTextureDescriptor; |
| colorTextureDescriptor.format = kColorFormat; |
| colorTextureDescriptor.sampleCount = 1; |
| wgpu::Texture colorTexture = device.CreateTexture(&colorTextureDescriptor); |
| |
| wgpu::TextureDescriptor depthStencilTextureDescriptor = baseTextureDescriptor; |
| depthStencilTextureDescriptor.sampleCount = 1; |
| depthStencilTextureDescriptor.format = kDepthStencilFormat; |
| wgpu::Texture depthStencilTexture = device.CreateTexture(&depthStencilTextureDescriptor); |
| utils::ComboRenderPassDescriptor renderPassDescriptor({}, depthStencilTexture.CreateView()); |
| |
| utils::ComboRenderPipelineDescriptor renderPipelineDescriptor; |
| renderPipelineDescriptor.multisample.count = 1; |
| renderPipelineDescriptor.vertex.module = vsModule; |
| |
| renderPipelineDescriptor.fragment = nullptr; |
| |
| renderPipelineDescriptor.EnableDepthStencil(kDepthStencilFormat); |
| |
| wgpu::RenderPipeline vertexOnlyPipeline = |
| device.CreateRenderPipeline(&renderPipelineDescriptor); |
| |
| // Vertex-only render pipeline can work with depth stencil attachment and no color target |
| { |
| utils::ComboRenderPassDescriptor renderPassDescriptor({}, depthStencilTexture.CreateView()); |
| |
| wgpu::CommandEncoder encoder = device.CreateCommandEncoder(); |
| wgpu::RenderPassEncoder renderPass = encoder.BeginRenderPass(&renderPassDescriptor); |
| renderPass.SetPipeline(vertexOnlyPipeline); |
| renderPass.EndPass(); |
| |
| encoder.Finish(); |
| } |
| |
| // Vertex-only render pipeline must have a depth stencil attachment |
| { |
| utils::ComboRenderPassDescriptor renderPassDescriptor({}); |
| |
| wgpu::CommandEncoder encoder = device.CreateCommandEncoder(); |
| wgpu::RenderPassEncoder renderPass = encoder.BeginRenderPass(&renderPassDescriptor); |
| renderPass.SetPipeline(vertexOnlyPipeline); |
| renderPass.EndPass(); |
| |
| ASSERT_DEVICE_ERROR(encoder.Finish()); |
| } |
| |
| // Vertex-only render pipeline can not work with color target |
| { |
| utils::ComboRenderPassDescriptor renderPassDescriptor({colorTexture.CreateView()}, |
| depthStencilTexture.CreateView()); |
| |
| wgpu::CommandEncoder encoder = device.CreateCommandEncoder(); |
| wgpu::RenderPassEncoder renderPass = encoder.BeginRenderPass(&renderPassDescriptor); |
| renderPass.SetPipeline(vertexOnlyPipeline); |
| renderPass.EndPass(); |
| |
| ASSERT_DEVICE_ERROR(encoder.Finish()); |
| } |
| |
| // Vertex-only render pipeline can not work with color target, and must have a depth stencil |
| // attachment |
| { |
| utils::ComboRenderPassDescriptor renderPassDescriptor({colorTexture.CreateView()}); |
| |
| wgpu::CommandEncoder encoder = device.CreateCommandEncoder(); |
| wgpu::RenderPassEncoder renderPass = encoder.BeginRenderPass(&renderPassDescriptor); |
| renderPass.SetPipeline(vertexOnlyPipeline); |
| renderPass.EndPass(); |
| |
| ASSERT_DEVICE_ERROR(encoder.Finish()); |
| } |
| } |
| |
| // Tests that the sample count of the render pipeline must be valid |
| // when the alphaToCoverage mode is enabled. |
| TEST_F(RenderPipelineValidationTest, AlphaToCoverageAndSampleCount) { |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| descriptor.multisample.count = 4; |
| descriptor.multisample.alphaToCoverageEnabled = true; |
| |
| device.CreateRenderPipeline(&descriptor); |
| } |
| |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| descriptor.multisample.count = 1; |
| descriptor.multisample.alphaToCoverageEnabled = true; |
| |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| } |
| |
| // Tests that the texture component type in shader must match the bind group layout. |
| TEST_F(RenderPipelineValidationTest, TextureComponentTypeCompatibility) { |
| constexpr uint32_t kNumTextureComponentType = 3u; |
| std::array<const char*, kNumTextureComponentType> kScalarTypes = {{"f32", "i32", "u32"}}; |
| std::array<wgpu::TextureSampleType, kNumTextureComponentType> kTextureComponentTypes = {{ |
| wgpu::TextureSampleType::Float, |
| wgpu::TextureSampleType::Sint, |
| wgpu::TextureSampleType::Uint, |
| }}; |
| |
| for (size_t i = 0; i < kNumTextureComponentType; ++i) { |
| for (size_t j = 0; j < kNumTextureComponentType; ++j) { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| |
| std::ostringstream stream; |
| stream << R"( |
| [[group(0), binding(0)]] var myTexture : texture_2d<)" |
| << kScalarTypes[i] << R"(>; |
| |
| [[stage(fragment)]] fn main() { |
| textureDimensions(myTexture); |
| })"; |
| descriptor.cFragment.module = utils::CreateShaderModule(device, stream.str().c_str()); |
| descriptor.cTargets[0].writeMask = wgpu::ColorWriteMask::None; |
| |
| wgpu::BindGroupLayout bgl = utils::MakeBindGroupLayout( |
| device, {{0, wgpu::ShaderStage::Fragment, kTextureComponentTypes[j]}}); |
| descriptor.layout = utils::MakeBasicPipelineLayout(device, &bgl); |
| |
| if (i == j) { |
| device.CreateRenderPipeline(&descriptor); |
| } else { |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| } |
| } |
| } |
| |
| // Tests that the texture view dimension in shader must match the bind group layout. |
| TEST_F(RenderPipelineValidationTest, TextureViewDimensionCompatibility) { |
| constexpr uint32_t kNumTextureViewDimensions = 6u; |
| std::array<const char*, kNumTextureViewDimensions> kTextureKeywords = {{ |
| "texture_1d", |
| "texture_2d", |
| "texture_2d_array", |
| "texture_cube", |
| "texture_cube_array", |
| "texture_3d", |
| }}; |
| |
| std::array<wgpu::TextureViewDimension, kNumTextureViewDimensions> kTextureViewDimensions = {{ |
| wgpu::TextureViewDimension::e1D, |
| wgpu::TextureViewDimension::e2D, |
| wgpu::TextureViewDimension::e2DArray, |
| wgpu::TextureViewDimension::Cube, |
| wgpu::TextureViewDimension::CubeArray, |
| wgpu::TextureViewDimension::e3D, |
| }}; |
| |
| for (size_t i = 0; i < kNumTextureViewDimensions; ++i) { |
| for (size_t j = 0; j < kNumTextureViewDimensions; ++j) { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| |
| std::ostringstream stream; |
| stream << R"( |
| [[group(0), binding(0)]] var myTexture : )" |
| << kTextureKeywords[i] << R"(<f32>; |
| [[stage(fragment)]] fn main() { |
| textureDimensions(myTexture); |
| })"; |
| descriptor.cFragment.module = utils::CreateShaderModule(device, stream.str().c_str()); |
| descriptor.cTargets[0].writeMask = wgpu::ColorWriteMask::None; |
| |
| wgpu::BindGroupLayout bgl = utils::MakeBindGroupLayout( |
| device, {{0, wgpu::ShaderStage::Fragment, wgpu::TextureSampleType::Float, |
| kTextureViewDimensions[j]}}); |
| descriptor.layout = utils::MakeBasicPipelineLayout(device, &bgl); |
| |
| if (i == j) { |
| device.CreateRenderPipeline(&descriptor); |
| } else { |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| } |
| } |
| } |
| |
| // Test that declaring a storage buffer in the vertex shader without setting pipeline layout won't |
| // cause crash. |
| TEST_F(RenderPipelineValidationTest, StorageBufferInVertexShaderNoLayout) { |
| wgpu::ShaderModule vsModuleWithStorageBuffer = utils::CreateShaderModule(device, R"( |
| [[block]] struct Dst { |
| data : array<u32, 100>; |
| }; |
| [[group(0), binding(0)]] var<storage, read_write> dst : Dst; |
| [[stage(vertex)]] fn main([[builtin(vertex_index)]] VertexIndex : u32) -> [[builtin(position)]] vec4<f32> { |
| dst.data[VertexIndex] = 0x1234u; |
| return vec4<f32>(); |
| })"); |
| |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.layout = nullptr; |
| descriptor.vertex.module = vsModuleWithStorageBuffer; |
| descriptor.cFragment.module = fsModule; |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| |
| // Tests that strip primitive topologies require an index format |
| TEST_F(RenderPipelineValidationTest, StripIndexFormatRequired) { |
| constexpr uint32_t kNumStripType = 2u; |
| constexpr uint32_t kNumListType = 3u; |
| constexpr uint32_t kNumIndexFormat = 3u; |
| |
| std::array<wgpu::PrimitiveTopology, kNumStripType> kStripTopologyTypes = { |
| {wgpu::PrimitiveTopology::LineStrip, wgpu::PrimitiveTopology::TriangleStrip}}; |
| |
| std::array<wgpu::PrimitiveTopology, kNumListType> kListTopologyTypes = { |
| {wgpu::PrimitiveTopology::PointList, wgpu::PrimitiveTopology::LineList, |
| wgpu::PrimitiveTopology::TriangleList}}; |
| |
| std::array<wgpu::IndexFormat, kNumIndexFormat> kIndexFormatTypes = { |
| {wgpu::IndexFormat::Undefined, wgpu::IndexFormat::Uint16, wgpu::IndexFormat::Uint32}}; |
| |
| for (wgpu::PrimitiveTopology primitiveTopology : kStripTopologyTypes) { |
| for (wgpu::IndexFormat indexFormat : kIndexFormatTypes) { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| descriptor.primitive.topology = primitiveTopology; |
| descriptor.primitive.stripIndexFormat = indexFormat; |
| |
| if (indexFormat == wgpu::IndexFormat::Undefined) { |
| // Fail because the index format is undefined and the primitive |
| // topology is a strip type. |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } else { |
| // Succeeds because the index format is given. |
| device.CreateRenderPipeline(&descriptor); |
| } |
| } |
| } |
| |
| for (wgpu::PrimitiveTopology primitiveTopology : kListTopologyTypes) { |
| for (wgpu::IndexFormat indexFormat : kIndexFormatTypes) { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| descriptor.primitive.topology = primitiveTopology; |
| descriptor.primitive.stripIndexFormat = indexFormat; |
| |
| if (indexFormat == wgpu::IndexFormat::Undefined) { |
| // Succeeds even when the index format is undefined because the |
| // primitive topology isn't a strip type. |
| device.CreateRenderPipeline(&descriptor); |
| } else { |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| } |
| } |
| } |
| |
| // Test that specifying a clampDepth value results in an error if the feature is not enabled. |
| TEST_F(RenderPipelineValidationTest, ClampDepthWithoutFeature) { |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| wgpu::PrimitiveDepthClampingState clampingState; |
| clampingState.clampDepth = true; |
| descriptor.primitive.nextInChain = &clampingState; |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| wgpu::PrimitiveDepthClampingState clampingState; |
| clampingState.clampDepth = false; |
| descriptor.primitive.nextInChain = &clampingState; |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| } |
| |
| // Test that depthStencil.depthCompare must not be undefiend. |
| TEST_F(RenderPipelineValidationTest, DepthCompareUndefinedIsError) { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| descriptor.EnableDepthStencil(wgpu::TextureFormat::Depth32Float); |
| |
| // Control case: Always is valid. |
| descriptor.cDepthStencil.depthCompare = wgpu::CompareFunction::Always; |
| device.CreateRenderPipeline(&descriptor); |
| |
| // Error case: Undefined is invalid. |
| descriptor.cDepthStencil.depthCompare = wgpu::CompareFunction::Undefined; |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| |
| // Test that the entryPoint names must be present for the correct stage in the shader module. |
| TEST_F(RenderPipelineValidationTest, EntryPointNameValidation) { |
| wgpu::ShaderModule module = utils::CreateShaderModule(device, R"( |
| [[stage(vertex)]] fn vertex_main() -> [[builtin(position)]] vec4<f32> { |
| return vec4<f32>(0.0, 0.0, 0.0, 1.0); |
| } |
| |
| [[stage(fragment)]] fn fragment_main() -> [[location(0)]] vec4<f32> { |
| return vec4<f32>(1.0, 0.0, 0.0, 1.0); |
| } |
| )"); |
| |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = module; |
| descriptor.vertex.entryPoint = "vertex_main"; |
| descriptor.cFragment.module = module; |
| descriptor.cFragment.entryPoint = "fragment_main"; |
| |
| // Success case. |
| device.CreateRenderPipeline(&descriptor); |
| |
| // Test for the vertex stage entryPoint name. |
| { |
| // The entryPoint name doesn't exist in the module. |
| descriptor.vertex.entryPoint = "main"; |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| |
| // The entryPoint name exists, but not for the correct stage. |
| descriptor.vertex.entryPoint = "fragment_main"; |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| |
| descriptor.vertex.entryPoint = "vertex_main"; |
| |
| // Test for the fragment stage entryPoint name. |
| { |
| // The entryPoint name doesn't exist in the module. |
| descriptor.cFragment.entryPoint = "main"; |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| |
| // The entryPoint name exists, but not for the correct stage. |
| descriptor.cFragment.entryPoint = "vertex_main"; |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| } |
| |
| // Test that vertex attrib validation is for the correct entryPoint |
| TEST_F(RenderPipelineValidationTest, VertexAttribCorrectEntryPoint) { |
| wgpu::ShaderModule module = utils::CreateShaderModule(device, R"( |
| [[stage(vertex)]] fn vertex0([[location(0)]] attrib0 : vec4<f32>) |
| -> [[builtin(position)]] vec4<f32> { |
| return attrib0; |
| } |
| [[stage(vertex)]] fn vertex1([[location(1)]] attrib1 : vec4<f32>) |
| -> [[builtin(position)]] vec4<f32> { |
| return attrib1; |
| } |
| )"); |
| |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = module; |
| descriptor.cFragment.module = fsModule; |
| |
| descriptor.vertex.bufferCount = 1; |
| descriptor.cBuffers[0].attributeCount = 1; |
| descriptor.cBuffers[0].arrayStride = 16; |
| descriptor.cAttributes[0].format = wgpu::VertexFormat::Float32x4; |
| descriptor.cAttributes[0].offset = 0; |
| |
| // Success cases, the attribute used by the entryPoint is declared in the pipeline. |
| descriptor.vertex.entryPoint = "vertex0"; |
| descriptor.cAttributes[0].shaderLocation = 0; |
| device.CreateRenderPipeline(&descriptor); |
| |
| descriptor.vertex.entryPoint = "vertex1"; |
| descriptor.cAttributes[0].shaderLocation = 1; |
| device.CreateRenderPipeline(&descriptor); |
| |
| // Error cases, the attribute used by the entryPoint isn't declared in the pipeline. |
| descriptor.vertex.entryPoint = "vertex1"; |
| descriptor.cAttributes[0].shaderLocation = 0; |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| |
| descriptor.vertex.entryPoint = "vertex0"; |
| descriptor.cAttributes[0].shaderLocation = 1; |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| |
| // Test that fragment output validation is for the correct entryPoint |
| TEST_F(RenderPipelineValidationTest, FragmentOutputCorrectEntryPoint) { |
| wgpu::ShaderModule module = utils::CreateShaderModule(device, R"( |
| [[stage(fragment)]] fn fragmentFloat() -> [[location(0)]] vec4<f32> { |
| return vec4<f32>(0.0, 0.0, 0.0, 0.0); |
| } |
| [[stage(fragment)]] fn fragmentUint() -> [[location(0)]] vec4<u32> { |
| return vec4<u32>(0u, 0u, 0u, 0u); |
| } |
| )"); |
| |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = module; |
| |
| // Success case, the component type matches between the pipeline and the entryPoint |
| descriptor.cFragment.entryPoint = "fragmentFloat"; |
| descriptor.cTargets[0].format = wgpu::TextureFormat::RGBA32Float; |
| device.CreateRenderPipeline(&descriptor); |
| |
| descriptor.cFragment.entryPoint = "fragmentUint"; |
| descriptor.cTargets[0].format = wgpu::TextureFormat::RGBA32Uint; |
| device.CreateRenderPipeline(&descriptor); |
| |
| // Error case, the component type doesn't match between the pipeline and the entryPoint |
| descriptor.cFragment.entryPoint = "fragmentUint"; |
| descriptor.cTargets[0].format = wgpu::TextureFormat::RGBA32Float; |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| |
| descriptor.cFragment.entryPoint = "fragmentFloat"; |
| descriptor.cTargets[0].format = wgpu::TextureFormat::RGBA32Uint; |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| |
| // Test that unwritten fragment outputs must have a write mask of 0. |
| TEST_F(RenderPipelineValidationTest, UnwrittenFragmentOutputsMask0) { |
| wgpu::ShaderModule vsModule = utils::CreateShaderModule(device, R"( |
| [[stage(vertex)]] fn main() -> [[builtin(position)]] vec4<f32> { |
| return vec4<f32>(); |
| } |
| )"); |
| |
| wgpu::ShaderModule fsModuleWriteNone = utils::CreateShaderModule(device, R"( |
| [[stage(fragment)]] fn main() {} |
| )"); |
| |
| wgpu::ShaderModule fsModuleWrite0 = utils::CreateShaderModule(device, R"( |
| [[stage(fragment)]] fn main() -> [[location(0)]] vec4<f32> { |
| return vec4<f32>(); |
| } |
| )"); |
| |
| wgpu::ShaderModule fsModuleWrite1 = utils::CreateShaderModule(device, R"( |
| [[stage(fragment)]] fn main() -> [[location(1)]] vec4<f32> { |
| return vec4<f32>(); |
| } |
| )"); |
| |
| wgpu::ShaderModule fsModuleWriteBoth = utils::CreateShaderModule(device, R"( |
| struct FragmentOut { |
| [[location(0)]] target0 : vec4<f32>; |
| [[location(1)]] target1 : vec4<f32>; |
| }; |
| [[stage(fragment)]] fn main() -> FragmentOut { |
| var out : FragmentOut; |
| return out; |
| } |
| )"); |
| |
| // Control case: write to target 0 |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| |
| descriptor.cFragment.targetCount = 1; |
| descriptor.cFragment.module = fsModuleWrite0; |
| device.CreateRenderPipeline(&descriptor); |
| } |
| |
| // Control case: write to target 0 and target 1 |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| |
| descriptor.cFragment.targetCount = 2; |
| descriptor.cFragment.module = fsModuleWriteBoth; |
| device.CreateRenderPipeline(&descriptor); |
| } |
| |
| // Write only target 1 (not in pipeline fragment state). |
| // Errors because target 0 does not have a write mask of 0. |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| |
| descriptor.cFragment.targetCount = 1; |
| descriptor.cTargets[0].writeMask = wgpu::ColorWriteMask::All; |
| descriptor.cFragment.module = fsModuleWrite1; |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| |
| // Write only target 1 (not in pipeline fragment state). |
| // OK because target 0 has a write mask of 0. |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| |
| descriptor.cFragment.targetCount = 1; |
| descriptor.cTargets[0].writeMask = wgpu::ColorWriteMask::None; |
| descriptor.cFragment.module = fsModuleWrite1; |
| device.CreateRenderPipeline(&descriptor); |
| } |
| |
| // Write only target 0 with two color targets. |
| // Errors because target 1 does not have a write mask of 0. |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| |
| descriptor.cFragment.targetCount = 2; |
| descriptor.cTargets[0].writeMask = wgpu::ColorWriteMask::Red; |
| descriptor.cTargets[1].writeMask = wgpu::ColorWriteMask::Alpha; |
| descriptor.cFragment.module = fsModuleWrite0; |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| |
| // Write only target 0 with two color targets. |
| // OK because target 1 has a write mask of 0. |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| |
| descriptor.cFragment.targetCount = 2; |
| descriptor.cTargets[0].writeMask = wgpu::ColorWriteMask::All; |
| descriptor.cTargets[1].writeMask = wgpu::ColorWriteMask::None; |
| descriptor.cFragment.module = fsModuleWrite0; |
| device.CreateRenderPipeline(&descriptor); |
| } |
| |
| // Write nothing with two color targets. |
| // Errors because both target 0 and 1 have nonzero write masks. |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| |
| descriptor.cFragment.targetCount = 2; |
| descriptor.cTargets[0].writeMask = wgpu::ColorWriteMask::Red; |
| descriptor.cTargets[1].writeMask = wgpu::ColorWriteMask::Green; |
| descriptor.cFragment.module = fsModuleWriteNone; |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| |
| // Write nothing with two color targets. |
| // OK because target 0 and 1 have write masks of 0. |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| |
| descriptor.cFragment.targetCount = 2; |
| descriptor.cTargets[0].writeMask = wgpu::ColorWriteMask::None; |
| descriptor.cTargets[1].writeMask = wgpu::ColorWriteMask::None; |
| descriptor.cFragment.module = fsModuleWriteNone; |
| device.CreateRenderPipeline(&descriptor); |
| } |
| } |
| |
| // Test that fragment output validation is for the correct entryPoint |
| TEST_F(RenderPipelineValidationTest, BindingsFromCorrectEntryPoint) { |
| wgpu::ShaderModule module = utils::CreateShaderModule(device, R"( |
| [[block]] struct Uniforms { |
| data : vec4<f32>; |
| }; |
| [[group(0), binding(0)]] var<uniform> var0 : Uniforms; |
| [[group(0), binding(1)]] var<uniform> var1 : Uniforms; |
| |
| [[stage(vertex)]] fn vertex0() -> [[builtin(position)]] vec4<f32> { |
| return var0.data; |
| } |
| [[stage(vertex)]] fn vertex1() -> [[builtin(position)]] vec4<f32> { |
| return var1.data; |
| } |
| )"); |
| |
| wgpu::BindGroupLayout bgl0 = utils::MakeBindGroupLayout( |
| device, {{0, wgpu::ShaderStage::Vertex, wgpu::BufferBindingType::Uniform}}); |
| wgpu::PipelineLayout layout0 = utils::MakeBasicPipelineLayout(device, &bgl0); |
| |
| wgpu::BindGroupLayout bgl1 = utils::MakeBindGroupLayout( |
| device, {{1, wgpu::ShaderStage::Vertex, wgpu::BufferBindingType::Uniform}}); |
| wgpu::PipelineLayout layout1 = utils::MakeBasicPipelineLayout(device, &bgl1); |
| |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = module; |
| descriptor.cFragment.module = fsModule; |
| |
| // Success case, the BGL matches the bindings used by the entryPoint |
| descriptor.vertex.entryPoint = "vertex0"; |
| descriptor.layout = layout0; |
| device.CreateRenderPipeline(&descriptor); |
| |
| descriptor.vertex.entryPoint = "vertex1"; |
| descriptor.layout = layout1; |
| device.CreateRenderPipeline(&descriptor); |
| |
| // Error case, the BGL doesn't match the bindings used by the entryPoint |
| descriptor.vertex.entryPoint = "vertex1"; |
| descriptor.layout = layout0; |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| |
| descriptor.vertex.entryPoint = "vertex0"; |
| descriptor.layout = layout1; |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| |
| class DepthClampingValidationTest : public RenderPipelineValidationTest { |
| protected: |
| WGPUDevice CreateTestDevice() override { |
| dawn_native::DawnDeviceDescriptor descriptor; |
| descriptor.requiredFeatures = {"depth-clamping"}; |
| return adapter.CreateDevice(&descriptor); |
| } |
| }; |
| |
| // Tests that specifying a clampDepth value succeeds if the feature is enabled. |
| TEST_F(DepthClampingValidationTest, Success) { |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| wgpu::PrimitiveDepthClampingState clampingState; |
| clampingState.clampDepth = true; |
| descriptor.primitive.nextInChain = &clampingState; |
| device.CreateRenderPipeline(&descriptor); |
| } |
| { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vsModule; |
| descriptor.cFragment.module = fsModule; |
| wgpu::PrimitiveDepthClampingState clampingState; |
| clampingState.clampDepth = false; |
| descriptor.primitive.nextInChain = &clampingState; |
| device.CreateRenderPipeline(&descriptor); |
| } |
| } |
| |
| class InterStageVariableMatchingValidationTest : public RenderPipelineValidationTest { |
| protected: |
| void CheckCreatingRenderPipeline(wgpu::ShaderModule vertexModule, |
| wgpu::ShaderModule fragmentModule, |
| bool shouldSucceed) { |
| utils::ComboRenderPipelineDescriptor descriptor; |
| descriptor.vertex.module = vertexModule; |
| descriptor.cFragment.module = fragmentModule; |
| if (shouldSucceed) { |
| device.CreateRenderPipeline(&descriptor); |
| } else { |
| ASSERT_DEVICE_ERROR(device.CreateRenderPipeline(&descriptor)); |
| } |
| } |
| }; |
| |
| // Tests that creating render pipeline should fail when there is a vertex output that doesn't have |
| // its corresponding fragment input at the same location, and there is a fragment input that |
| // doesn't have its corresponding vertex output at the same location. |
| TEST_F(InterStageVariableMatchingValidationTest, MissingDeclarationAtSameLocation) { |
| wgpu::ShaderModule vertexModuleOutputAtLocation0 = utils::CreateShaderModule(device, R"( |
| struct A { |
| [[location(0)]] vout: f32; |
| [[builtin(position)]] pos: vec4<f32>; |
| }; |
| [[stage(vertex)]] fn main() -> A { |
| var vertexOut: A; |
| vertexOut.pos = vec4<f32>(0.0, 0.0, 0.0, 1.0); |
| return vertexOut; |
| })"); |
| wgpu::ShaderModule fragmentModuleAtLocation0 = utils::CreateShaderModule(device, R"( |
| struct B { |
| [[location(0)]] fin: f32; |
| }; |
| [[stage(fragment)]] fn main(fragmentIn: B) -> [[location(0)]] vec4<f32> { |
| return vec4<f32>(fragmentIn.fin, 0.0, 0.0, 1.0); |
| })"); |
| wgpu::ShaderModule fragmentModuleInputAtLocation1 = utils::CreateShaderModule(device, R"( |
| struct A { |
| [[location(1)]] vout: f32; |
| }; |
| [[stage(fragment)]] fn main(vertexOut: A) -> [[location(0)]] vec4<f32> { |
| return vec4<f32>(vertexOut.vout, 0.0, 0.0, 1.0); |
| })"); |
| wgpu::ShaderModule vertexModuleOutputAtLocation1 = utils::CreateShaderModule(device, R"( |
| struct B { |
| [[location(1)]] fin: f32; |
| [[builtin(position)]] pos: vec4<f32>; |
| }; |
| [[stage(vertex)]] fn main() -> B { |
| var fragmentIn: B; |
| fragmentIn.pos = vec4<f32>(0.0, 0.0, 0.0, 1.0); |
| return fragmentIn; |
| })"); |
| |
| { |
| CheckCreatingRenderPipeline(vertexModuleOutputAtLocation0, fsModule, false); |
| CheckCreatingRenderPipeline(vsModule, fragmentModuleAtLocation0, false); |
| CheckCreatingRenderPipeline(vertexModuleOutputAtLocation0, fragmentModuleInputAtLocation1, |
| false); |
| CheckCreatingRenderPipeline(vertexModuleOutputAtLocation1, fragmentModuleAtLocation0, |
| false); |
| } |
| |
| { |
| CheckCreatingRenderPipeline(vertexModuleOutputAtLocation0, fragmentModuleAtLocation0, true); |
| CheckCreatingRenderPipeline(vertexModuleOutputAtLocation1, fragmentModuleInputAtLocation1, |
| true); |
| } |
| } |
| |
| // Tests that creating render pipeline should fail when the type of a vertex stage output variable |
| // doesn't match the type of the fragment stage input variable at the same location. |
| TEST_F(InterStageVariableMatchingValidationTest, DifferentTypeAtSameLocation) { |
| constexpr std::array<const char*, 12> kTypes = {{"f32", "vec2<f32>", "vec3<f32>", "vec4<f32>", |
| "i32", "vec2<i32>", "vec3<i32>", "vec4<i32>", |
| "u32", "vec2<u32>", "vec3<u32>", "vec4<u32>"}}; |
| |
| std::array<wgpu::ShaderModule, 12> vertexModules; |
| std::array<wgpu::ShaderModule, 12> fragmentModules; |
| for (uint32_t i = 0; i < kTypes.size(); ++i) { |
| std::string interfaceDeclaration; |
| { |
| std::ostringstream sstream; |
| sstream << "struct A { [[location(0)]] a: " << kTypes[i] << ";" << std::endl; |
| interfaceDeclaration = sstream.str(); |
| } |
| { |
| std::ostringstream vertexStream; |
| vertexStream << interfaceDeclaration << R"( |
| [[builtin(position)]] pos: vec4<f32>; |
| }; |
| [[stage(vertex)]] fn main() -> A { |
| var vertexOut: A; |
| vertexOut.pos = vec4<f32>(0.0, 0.0, 0.0, 1.0); |
| return vertexOut; |
| })"; |
| vertexModules[i] = utils::CreateShaderModule(device, vertexStream.str().c_str()); |
| } |
| { |
| std::ostringstream fragmentStream; |
| fragmentStream << interfaceDeclaration << R"( |
| }; |
| [[stage(fragment)]] fn main(fragmentIn: A) -> [[location(0)]] vec4<f32> { |
| return vec4<f32>(0.0, 0.0, 0.0, 1.0); |
| })"; |
| fragmentModules[i] = utils::CreateShaderModule(device, fragmentStream.str().c_str()); |
| } |
| } |
| |
| for (uint32_t vertexModuleIndex = 0; vertexModuleIndex < kTypes.size(); ++vertexModuleIndex) { |
| wgpu::ShaderModule vertexModule = vertexModules[vertexModuleIndex]; |
| for (uint32_t fragmentModuleIndex = 0; fragmentModuleIndex < kTypes.size(); |
| ++fragmentModuleIndex) { |
| wgpu::ShaderModule fragmentModule = fragmentModules[fragmentModuleIndex]; |
| bool shouldSuccess = vertexModuleIndex == fragmentModuleIndex; |
| CheckCreatingRenderPipeline(vertexModule, fragmentModule, shouldSuccess); |
| } |
| } |
| } |
| |
| // Tests that creating render pipeline should fail when the interpolation attribute of a vertex |
| // stage output variable doesn't match the type of the fragment stage input variable at the same |
| // location. |
| TEST_F(InterStageVariableMatchingValidationTest, DifferentInterpolationAttributeAtSameLocation) { |
| enum class InterpolationType : uint8_t { |
| None = 0, |
| Perspective, |
| Linear, |
| Flat, |
| Count, |
| }; |
| enum class InterpolationSampling : uint8_t { |
| None = 0, |
| Center, |
| Centroid, |
| Sample, |
| Count, |
| }; |
| constexpr std::array<const char*, static_cast<size_t>(InterpolationType::Count)> |
| kInterpolationTypeString = {{"", "perspective", "linear", "flat"}}; |
| constexpr std::array<const char*, static_cast<size_t>(InterpolationSampling::Count)> |
| kInterpolationSamplingString = {{"", "center", "centroid", "sample"}}; |
| |
| struct InterpolationAttribute { |
| InterpolationType interpolationType; |
| InterpolationSampling interpolationSampling; |
| }; |
| |
| // Interpolation sampling is not used with flat interpolation. |
| constexpr std::array<InterpolationAttribute, 10> validInterpolationAttributes = {{ |
| {InterpolationType::None, InterpolationSampling::None}, |
| {InterpolationType::Flat, InterpolationSampling::None}, |
| {InterpolationType::Linear, InterpolationSampling::None}, |
| {InterpolationType::Linear, InterpolationSampling::Center}, |
| {InterpolationType::Linear, InterpolationSampling::Centroid}, |
| {InterpolationType::Linear, InterpolationSampling::Sample}, |
| {InterpolationType::Perspective, InterpolationSampling::None}, |
| {InterpolationType::Perspective, InterpolationSampling::Center}, |
| {InterpolationType::Perspective, InterpolationSampling::Centroid}, |
| {InterpolationType::Perspective, InterpolationSampling::Sample}, |
| }}; |
| |
| std::vector<wgpu::ShaderModule> vertexModules(validInterpolationAttributes.size()); |
| std::vector<wgpu::ShaderModule> fragmentModules(validInterpolationAttributes.size()); |
| for (uint32_t i = 0; i < validInterpolationAttributes.size(); ++i) { |
| std::string interfaceDeclaration; |
| { |
| const auto& interpolationAttribute = validInterpolationAttributes[i]; |
| std::ostringstream sstream; |
| sstream << "struct A { [[location(0)"; |
| if (interpolationAttribute.interpolationType != InterpolationType::None) { |
| sstream << ", interpolate(" |
| << kInterpolationTypeString[static_cast<uint8_t>( |
| interpolationAttribute.interpolationType)]; |
| if (interpolationAttribute.interpolationSampling != InterpolationSampling::None) { |
| sstream << ", " |
| << kInterpolationSamplingString[static_cast<uint8_t>( |
| interpolationAttribute.interpolationSampling)]; |
| } |
| sstream << ")"; |
| } |
| sstream << " ]] a : vec4<f32>;" << std::endl; |
| interfaceDeclaration = sstream.str(); |
| } |
| { |
| std::ostringstream vertexStream; |
| vertexStream << interfaceDeclaration << R"( |
| [[builtin(position)]] pos: vec4<f32>; |
| }; |
| [[stage(vertex)]] fn main() -> A { |
| var vertexOut: A; |
| vertexOut.pos = vec4<f32>(0.0, 0.0, 0.0, 1.0); |
| return vertexOut; |
| })"; |
| vertexModules[i] = utils::CreateShaderModule(device, vertexStream.str().c_str()); |
| } |
| { |
| std::ostringstream fragmentStream; |
| fragmentStream << interfaceDeclaration << R"( |
| }; |
| [[stage(fragment)]] fn main(fragmentIn: A) -> [[location(0)]] vec4<f32> { |
| return fragmentIn.a; |
| })"; |
| fragmentModules[i] = utils::CreateShaderModule(device, fragmentStream.str().c_str()); |
| } |
| } |
| |
| auto GetAppliedInterpolationAttribute = [](const InterpolationAttribute& attribute) { |
| InterpolationAttribute appliedAttribute = {attribute.interpolationType, |
| attribute.interpolationSampling}; |
| switch (attribute.interpolationType) { |
| // If the interpolation attribute is not specified, then |
| // [[interpolate(perspective, center)]] or [[interpolate(perspective)]] is assumed. |
| case InterpolationType::None: |
| appliedAttribute.interpolationType = InterpolationType::Perspective; |
| appliedAttribute.interpolationSampling = InterpolationSampling::Center; |
| break; |
| |
| // If the interpolation type is perspective or linear, and the interpolation |
| // sampling is not specified, then 'center' is assumed. |
| case InterpolationType::Perspective: |
| case InterpolationType::Linear: |
| if (appliedAttribute.interpolationSampling == InterpolationSampling::None) { |
| appliedAttribute.interpolationSampling = InterpolationSampling::Center; |
| } |
| break; |
| |
| case InterpolationType::Flat: |
| break; |
| default: |
| UNREACHABLE(); |
| } |
| return appliedAttribute; |
| }; |
| |
| auto InterpolationAttributeMatch = [GetAppliedInterpolationAttribute]( |
| const InterpolationAttribute& attribute1, |
| const InterpolationAttribute& attribute2) { |
| InterpolationAttribute appliedAttribute1 = GetAppliedInterpolationAttribute(attribute1); |
| InterpolationAttribute appliedAttribute2 = GetAppliedInterpolationAttribute(attribute2); |
| |
| return appliedAttribute1.interpolationType == appliedAttribute2.interpolationType && |
| appliedAttribute1.interpolationSampling == appliedAttribute2.interpolationSampling; |
| }; |
| |
| for (uint32_t vertexModuleIndex = 0; vertexModuleIndex < validInterpolationAttributes.size(); |
| ++vertexModuleIndex) { |
| wgpu::ShaderModule vertexModule = vertexModules[vertexModuleIndex]; |
| for (uint32_t fragmentModuleIndex = 0; |
| fragmentModuleIndex < validInterpolationAttributes.size(); ++fragmentModuleIndex) { |
| wgpu::ShaderModule fragmentModule = fragmentModules[fragmentModuleIndex]; |
| bool shouldSuccess = |
| InterpolationAttributeMatch(validInterpolationAttributes[vertexModuleIndex], |
| validInterpolationAttributes[fragmentModuleIndex]); |
| CheckCreatingRenderPipeline(vertexModule, fragmentModule, shouldSuccess); |
| } |
| } |
| } |