dawn-cmake/src/dawn_native/ShaderModule.cpp
Corentin Wallez a594f8fdb4 WebGPU error handling 1: Return error objects on errors.
Dawn used to return "nullptr" when an error happened while creating an
object. To match WebGPU we want to return valid pointers but to "error"
objects.

This commit implements WebGPU error handling for all "descriptorized"
objects and changes the nullptr error checks into "ValidateObject"
checks. This method is used both to check that the object isn't an
error, but also that all objects in a function call are from the same
device.

New validation is added to objects with methods (apart from Device) so
they check they aren't error objects.

A large number of ASSERTs were added to check that frontend objects
aren't accessed when they are errors, so that missing validation would
hit the asserts instead of crashing randomly.

The bind group validation tests were modified to test the behavior with
both nullptrs and error objects.

Future commits will change the CommandBufferBuilder to not be a builder
anymore but an "Encoder" instead with special-cased error handling. Then
once all objects are descriptorized, the notion of builders will be
removed from the code.

BUG=dawn:8

Change-Id: I8647712d5de3deb0e99e3bc58f34496f67710edd
Reviewed-on: https://dawn-review.googlesource.com/c/4360
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
2019-02-13 13:09:18 +00:00

288 lines
11 KiB
C++

// 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 "dawn_native/ShaderModule.h"
#include "dawn_native/BindGroupLayout.h"
#include "dawn_native/Device.h"
#include "dawn_native/Pipeline.h"
#include "dawn_native/PipelineLayout.h"
#include <spirv-cross/spirv_cross.hpp>
#include <spirv-tools/libspirv.hpp>
namespace dawn_native {
MaybeError ValidateShaderModuleDescriptor(DeviceBase*,
const ShaderModuleDescriptor* descriptor) {
if (descriptor->nextInChain != nullptr) {
return DAWN_VALIDATION_ERROR("nextInChain must be nullptr");
}
spvtools::SpirvTools spirvTools(SPV_ENV_VULKAN_1_1);
std::ostringstream errorStream;
errorStream << "SPIRV Validation failure:" << std::endl;
spirvTools.SetMessageConsumer([&errorStream](spv_message_level_t level, const char*,
const spv_position_t& position,
const char* message) {
switch (level) {
case SPV_MSG_FATAL:
case SPV_MSG_INTERNAL_ERROR:
case SPV_MSG_ERROR:
errorStream << "error: line " << position.index << ": " << message << std::endl;
break;
case SPV_MSG_WARNING:
errorStream << "warning: line " << position.index << ": " << message
<< std::endl;
break;
case SPV_MSG_INFO:
errorStream << "info: line " << position.index << ": " << message << std::endl;
break;
default:
break;
}
});
if (!spirvTools.Validate(descriptor->code, descriptor->codeSize)) {
return DAWN_VALIDATION_ERROR(errorStream.str().c_str());
}
return {};
}
// ShaderModuleBase
ShaderModuleBase::ShaderModuleBase(DeviceBase* device, const ShaderModuleDescriptor*)
: ObjectBase(device) {
}
ShaderModuleBase::ShaderModuleBase(DeviceBase* device, ObjectBase::ErrorTag tag)
: ObjectBase(device, tag) {
}
// static
ShaderModuleBase* ShaderModuleBase::MakeError(DeviceBase* device) {
return new ShaderModuleBase(device, ObjectBase::kError);
}
void ShaderModuleBase::ExtractSpirvInfo(const spirv_cross::Compiler& compiler) {
ASSERT(!IsError());
DeviceBase* device = GetDevice();
// TODO(cwallez@chromium.org): make errors here builder-level
// currently errors here do not prevent the shadermodule from being used
const auto& resources = compiler.get_shader_resources();
switch (compiler.get_execution_model()) {
case spv::ExecutionModelVertex:
mExecutionModel = dawn::ShaderStage::Vertex;
break;
case spv::ExecutionModelFragment:
mExecutionModel = dawn::ShaderStage::Fragment;
break;
case spv::ExecutionModelGLCompute:
mExecutionModel = dawn::ShaderStage::Compute;
break;
default:
UNREACHABLE();
}
// Extract push constants
mPushConstants.mask.reset();
mPushConstants.sizes.fill(0);
mPushConstants.types.fill(PushConstantType::Int);
if (resources.push_constant_buffers.size() > 0) {
auto interfaceBlock = resources.push_constant_buffers[0];
const auto& blockType = compiler.get_type(interfaceBlock.type_id);
ASSERT(blockType.basetype == spirv_cross::SPIRType::Struct);
for (uint32_t i = 0; i < blockType.member_types.size(); i++) {
ASSERT(compiler.get_member_decoration_bitset(blockType.self, i)
.get(spv::DecorationOffset));
uint32_t offset =
compiler.get_member_decoration(blockType.self, i, spv::DecorationOffset);
ASSERT(offset % 4 == 0);
offset /= 4;
auto memberType = compiler.get_type(blockType.member_types[i]);
PushConstantType constantType;
if (memberType.basetype == spirv_cross::SPIRType::Int) {
constantType = PushConstantType::Int;
} else if (memberType.basetype == spirv_cross::SPIRType::UInt) {
constantType = PushConstantType::UInt;
} else {
ASSERT(memberType.basetype == spirv_cross::SPIRType::Float);
constantType = PushConstantType::Float;
}
// TODO(cwallez@chromium.org): check for overflows and make the logic better take
// into account things like the array of types with padding.
uint32_t size = memberType.vecsize * memberType.columns;
// Handle unidimensional arrays
if (!memberType.array.empty()) {
size *= memberType.array[0];
}
if (offset + size > kMaxPushConstants) {
device->HandleError("Push constant block too big in the SPIRV");
return;
}
mPushConstants.mask.set(offset);
mPushConstants.names[offset] =
interfaceBlock.name + "." + compiler.get_member_name(blockType.self, i);
mPushConstants.sizes[offset] = size;
mPushConstants.types[offset] = constantType;
}
}
// Fill in bindingInfo with the SPIRV bindings
auto ExtractResourcesBinding = [this](const std::vector<spirv_cross::Resource>& resources,
const spirv_cross::Compiler& compiler,
dawn::BindingType bindingType) {
for (const auto& resource : resources) {
ASSERT(compiler.get_decoration_bitset(resource.id).get(spv::DecorationBinding));
ASSERT(
compiler.get_decoration_bitset(resource.id).get(spv::DecorationDescriptorSet));
uint32_t binding = compiler.get_decoration(resource.id, spv::DecorationBinding);
uint32_t set = compiler.get_decoration(resource.id, spv::DecorationDescriptorSet);
if (binding >= kMaxBindingsPerGroup || set >= kMaxBindGroups) {
GetDevice()->HandleError("Binding over limits in the SPIRV");
continue;
}
auto& info = mBindingInfo[set][binding];
info.used = true;
info.id = resource.id;
info.base_type_id = resource.base_type_id;
info.type = bindingType;
}
};
ExtractResourcesBinding(resources.uniform_buffers, compiler,
dawn::BindingType::UniformBuffer);
ExtractResourcesBinding(resources.separate_images, compiler,
dawn::BindingType::SampledTexture);
ExtractResourcesBinding(resources.separate_samplers, compiler, dawn::BindingType::Sampler);
ExtractResourcesBinding(resources.storage_buffers, compiler,
dawn::BindingType::StorageBuffer);
// Extract the vertex attributes
if (mExecutionModel == dawn::ShaderStage::Vertex) {
for (const auto& attrib : resources.stage_inputs) {
ASSERT(compiler.get_decoration_bitset(attrib.id).get(spv::DecorationLocation));
uint32_t location = compiler.get_decoration(attrib.id, spv::DecorationLocation);
if (location >= kMaxVertexAttributes) {
device->HandleError("Attribute location over limits in the SPIRV");
return;
}
mUsedVertexAttributes.set(location);
}
// Without a location qualifier on vertex outputs, spirv_cross::CompilerMSL gives them
// all the location 0, causing a compile error.
for (const auto& attrib : resources.stage_outputs) {
if (!compiler.get_decoration_bitset(attrib.id).get(spv::DecorationLocation)) {
device->HandleError("Need location qualifier on vertex output");
return;
}
}
}
if (mExecutionModel == dawn::ShaderStage::Fragment) {
// Without a location qualifier on vertex inputs, spirv_cross::CompilerMSL gives them
// all the location 0, causing a compile error.
for (const auto& attrib : resources.stage_inputs) {
if (!compiler.get_decoration_bitset(attrib.id).get(spv::DecorationLocation)) {
device->HandleError("Need location qualifier on fragment input");
return;
}
}
}
}
const ShaderModuleBase::PushConstantInfo& ShaderModuleBase::GetPushConstants() const {
ASSERT(!IsError());
return mPushConstants;
}
const ShaderModuleBase::ModuleBindingInfo& ShaderModuleBase::GetBindingInfo() const {
ASSERT(!IsError());
return mBindingInfo;
}
const std::bitset<kMaxVertexAttributes>& ShaderModuleBase::GetUsedVertexAttributes() const {
ASSERT(!IsError());
return mUsedVertexAttributes;
}
dawn::ShaderStage ShaderModuleBase::GetExecutionModel() const {
ASSERT(!IsError());
return mExecutionModel;
}
bool ShaderModuleBase::IsCompatibleWithPipelineLayout(const PipelineLayoutBase* layout) {
ASSERT(!IsError());
for (uint32_t group : IterateBitSet(layout->GetBindGroupLayoutsMask())) {
if (!IsCompatibleWithBindGroupLayout(group, layout->GetBindGroupLayout(group))) {
return false;
}
}
for (uint32_t group : IterateBitSet(~layout->GetBindGroupLayoutsMask())) {
for (size_t i = 0; i < kMaxBindingsPerGroup; ++i) {
if (mBindingInfo[group][i].used) {
return false;
}
}
}
return true;
}
bool ShaderModuleBase::IsCompatibleWithBindGroupLayout(size_t group,
const BindGroupLayoutBase* layout) {
ASSERT(!IsError());
const auto& layoutInfo = layout->GetBindingInfo();
for (size_t i = 0; i < kMaxBindingsPerGroup; ++i) {
const auto& moduleInfo = mBindingInfo[group][i];
if (!moduleInfo.used) {
continue;
}
if (moduleInfo.type != layoutInfo.types[i]) {
return false;
}
if ((layoutInfo.visibilities[i] & StageBit(mExecutionModel)) == 0) {
return false;
}
}
return true;
}
} // namespace dawn_native