TranslatorHLSL.cpp

Branch :

Show log
Commit
Author : Shahbaz Youssefi
Date : 2021-08-23 11:05:23
Hash : 800e82c6
Message : Translator: Validate precisions When declaring a variable, a struct field, function parameter etc, there's a precision necessarily applied to the entity being declared. AST Validation is added to enforce this. Intermediate nodes derive their precision from these entities automatically. Consistency of intermediate nodes is not validated. This is because AST transformations replace a node with a transformed one, and that may not have the same precision. Take the following code: mediump float x = ...; mediump float y = ...; ... x + y ... and assume is transformed as such: highp float driver_uniform; ... (x * driver_uniform) + y ... The addition was originally done in mediump, but would seemingly need to be done in highp after transformation. There are a number of options here: - Make sure that when nodes are replaced, the precision is unaffected. This can be intrusive, requiring temp variables. - Bubble up the new precision - Accept the discrepancy ANGLE opts for the last option, which actually respects the original shader's intended precision for operations, even if some transformation needs to temporarily evaluate an expression at a higher precision. Bug: angleproject:4889 Bug: angleproject:6132 Change-Id: Ibcde3a230de159157783b1c6d5ef1cd63ceb4d8f Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3114027 Reviewed-by: Tim Van Patten <timvp@google.com> Reviewed-by: Jamie Madill <jmadill@chromium.org> Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>

src/compiler/translator/TranslatorHLSL.cpp

//
// Copyright 2002 The ANGLE Project Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
//

#include "compiler/translator/TranslatorHLSL.h"

#include "compiler/translator/OutputHLSL.h"
#include "compiler/translator/tree_ops/RemoveDynamicIndexing.h"
#include "compiler/translator/tree_ops/RewriteTexelFetchOffset.h"
#include "compiler/translator/tree_ops/SimplifyLoopConditions.h"
#include "compiler/translator/tree_ops/SplitSequenceOperator.h"
#include "compiler/translator/tree_ops/d3d/AddDefaultReturnStatements.h"
#include "compiler/translator/tree_ops/d3d/ArrayReturnValueToOutParameter.h"
#include "compiler/translator/tree_ops/d3d/BreakVariableAliasingInInnerLoops.h"
#include "compiler/translator/tree_ops/d3d/ExpandIntegerPowExpressions.h"
#include "compiler/translator/tree_ops/d3d/RecordUniformBlocksWithLargeArrayMember.h"
#include "compiler/translator/tree_ops/d3d/RewriteAtomicFunctionExpressions.h"
#include "compiler/translator/tree_ops/d3d/RewriteElseBlocks.h"
#include "compiler/translator/tree_ops/d3d/RewriteExpressionsWithShaderStorageBlock.h"
#include "compiler/translator/tree_ops/d3d/RewriteUnaryMinusOperatorInt.h"
#include "compiler/translator/tree_ops/d3d/SeparateArrayConstructorStatements.h"
#include "compiler/translator/tree_ops/d3d/SeparateArrayInitialization.h"
#include "compiler/translator/tree_ops/d3d/SeparateExpressionsReturningArrays.h"
#include "compiler/translator/tree_ops/d3d/UnfoldShortCircuitToIf.h"
#include "compiler/translator/tree_ops/d3d/WrapSwitchStatementsInBlocks.h"
#include "compiler/translator/tree_util/IntermNodePatternMatcher.h"

namespace sh
{

TranslatorHLSL::TranslatorHLSL(sh::GLenum type, ShShaderSpec spec, ShShaderOutput output)
    : TCompiler(type, spec, output)
{}

bool TranslatorHLSL::translate(TIntermBlock *root,
                               ShCompileOptions compileOptions,
                               PerformanceDiagnostics *perfDiagnostics)
{
    // A few transformations leave the tree in an inconsistent state.  For example, when unfolding
    // the short-circuit in the following function:
    //
    //     mediump float f(float a) { return a > 0 ? 0.0 : 1.0; }
    //
    // a temp variable is created to hold the result of the expression.  Currently the precision of
    // the return value of the function is not propagated to its return expressions.  Additionally,
    // an expression such as
    //
    //     cond ? gl_NumWorkGroups.x : gl_NumWorkGroups.y
    //
    // does not have a precision as the built-in does not specify a precision.
    //
    // Precision is not applicable to HLSL so fixing these issues are deferred.
    mValidateASTOptions.validatePrecision = false;

    const ShBuiltInResources &resources = getResources();
    int numRenderTargets                = resources.EXT_draw_buffers ? resources.MaxDrawBuffers : 1;
    int maxDualSourceDrawBuffers =
        resources.EXT_blend_func_extended ? resources.MaxDualSourceDrawBuffers : 0;

    if (!sh::AddDefaultReturnStatements(this, root))
    {
        return false;
    }

    // Note that SimplifyLoopConditions needs to be run before any other AST transformations that
    // may need to generate new statements from loop conditions or loop expressions.
    // Note that SeparateDeclarations has already been run in TCompiler::compileTreeImpl().
    if (!SimplifyLoopConditions(
            this, root,
            IntermNodePatternMatcher::kExpressionReturningArray |
                IntermNodePatternMatcher::kUnfoldedShortCircuitExpression |
                IntermNodePatternMatcher::kDynamicIndexingOfVectorOrMatrixInLValue,
            &getSymbolTable()))
    {
        return false;
    }

    if (!SplitSequenceOperator(
            this, root,
            IntermNodePatternMatcher::kExpressionReturningArray |
                IntermNodePatternMatcher::kUnfoldedShortCircuitExpression |
                IntermNodePatternMatcher::kDynamicIndexingOfVectorOrMatrixInLValue,
            &getSymbolTable()))
    {
        return false;
    }

    // Note that SeparateDeclarations needs to be run before UnfoldShortCircuitToIf.
    if (!UnfoldShortCircuitToIf(this, root, &getSymbolTable()))
    {
        return false;
    }

    if (!SeparateArrayConstructorStatements(this, root))
    {
        return false;
    }

    if (!SeparateExpressionsReturningArrays(this, root, &getSymbolTable()))
    {
        return false;
    }

    // Note that SeparateDeclarations needs to be run before SeparateArrayInitialization.
    if (!SeparateArrayInitialization(this, root))
    {
        return false;
    }

    // HLSL doesn't support arrays as return values, we'll need to make functions that have an array
    // as a return value to use an out parameter to transfer the array data instead.
    if (!ArrayReturnValueToOutParameter(this, root, &getSymbolTable()))
    {
        return false;
    }

    if (!shouldRunLoopAndIndexingValidation(compileOptions))
    {
        // HLSL doesn't support dynamic indexing of vectors and matrices.
        if (!RemoveDynamicIndexingOfNonSSBOVectorOrMatrix(this, root, &getSymbolTable(),
                                                          perfDiagnostics))
        {
            return false;
        }
    }

    // Work around D3D9 bug that would manifest in vertex shaders with selection blocks which
    // use a vertex attribute as a condition, and some related computation in the else block.
    if (getOutputType() == SH_HLSL_3_0_OUTPUT && getShaderType() == GL_VERTEX_SHADER)
    {
        if (!sh::RewriteElseBlocks(this, root, &getSymbolTable()))
        {
            return false;
        }
    }

    // Work around an HLSL compiler frontend aliasing optimization bug.
    // TODO(cwallez) The date is 2016-08-25, Microsoft said the bug would be fixed
    // in the next release of d3dcompiler.dll, it would be nice to detect the DLL
    // version and only apply the workaround if it is too old.
    if (!sh::BreakVariableAliasingInInnerLoops(this, root))
    {
        return false;
    }

    // WrapSwitchStatementsInBlocks should be called after any AST transformations that might
    // introduce variable declarations inside the main scope of any switch statement. It cannot
    // result in no-op cases at the end of switch statements, because unreferenced variables
    // have already been pruned.
    if (!WrapSwitchStatementsInBlocks(this, root))
    {
        return false;
    }

    if ((compileOptions & SH_EXPAND_SELECT_HLSL_INTEGER_POW_EXPRESSIONS) != 0)
    {
        if (!sh::ExpandIntegerPowExpressions(this, root, &getSymbolTable()))
        {
            return false;
        }
    }

    if ((compileOptions & SH_REWRITE_TEXELFETCHOFFSET_TO_TEXELFETCH) != 0)
    {
        if (!sh::RewriteTexelFetchOffset(this, root, getSymbolTable(), getShaderVersion()))
        {
            return false;
        }
    }

    if (((compileOptions & SH_REWRITE_INTEGER_UNARY_MINUS_OPERATOR) != 0) &&
        getShaderType() == GL_VERTEX_SHADER)
    {
        if (!sh::RewriteUnaryMinusOperatorInt(this, root))
        {
            return false;
        }
    }

    if (getShaderVersion() >= 310)
    {
        // Due to ssbo also can be used as the argument of atomic memory functions, we should put
        // RewriteExpressionsWithShaderStorageBlock before RewriteAtomicFunctionExpressions.
        if (!sh::RewriteExpressionsWithShaderStorageBlock(this, root, &getSymbolTable()))
        {
            return false;
        }
        if (!sh::RewriteAtomicFunctionExpressions(this, root, &getSymbolTable(),
                                                  getShaderVersion()))
        {
            return false;
        }
    }

    mUniformBlockOptimizedMap.clear();
    mSlowCompilingUniformBlockSet.clear();
    // In order to get the exact maximum of slots are available for shader resources, which would
    // been bound with StructuredBuffer, we only translate uniform block with a large array member
    // into StructuredBuffer when shader version is 300.
    if (getShaderVersion() == 300 &&
        (compileOptions & SH_ALLOW_TRANSLATE_UNIFORM_BLOCK_TO_STRUCTUREDBUFFER) != 0)
    {
        if (!sh::RecordUniformBlocksWithLargeArrayMember(root, mUniformBlockOptimizedMap,
                                                         mSlowCompilingUniformBlockSet))
        {
            return false;
        }
    }

    sh::OutputHLSL outputHLSL(getShaderType(), getShaderSpec(), getShaderVersion(),
                              getExtensionBehavior(), getSourcePath(), getOutputType(),
                              numRenderTargets, maxDualSourceDrawBuffers, getUniforms(),
                              compileOptions, getComputeShaderLocalSize(), &getSymbolTable(),
                              perfDiagnostics, mUniformBlockOptimizedMap, mShaderStorageBlocks);

    outputHLSL.output(root, getInfoSink().obj);

    mShaderStorageBlockRegisterMap      = outputHLSL.getShaderStorageBlockRegisterMap();
    mUniformBlockRegisterMap            = outputHLSL.getUniformBlockRegisterMap();
    mUniformBlockUseStructuredBufferMap = outputHLSL.getUniformBlockUseStructuredBufferMap();
    mUniformRegisterMap                 = outputHLSL.getUniformRegisterMap();
    mReadonlyImage2DRegisterIndex       = outputHLSL.getReadonlyImage2DRegisterIndex();
    mImage2DRegisterIndex               = outputHLSL.getImage2DRegisterIndex();
    mUsedImage2DFunctionNames           = outputHLSL.getUsedImage2DFunctionNames();

    return true;
}

bool TranslatorHLSL::shouldFlattenPragmaStdglInvariantAll()
{
    // Not necessary when translating to HLSL.
    return false;
}

bool TranslatorHLSL::hasShaderStorageBlock(const std::string &uniformBlockName) const
{
    return (mShaderStorageBlockRegisterMap.count(uniformBlockName) > 0);
}

unsigned int TranslatorHLSL::getShaderStorageBlockRegister(
    const std::string &shaderStorageBlockName) const
{
    ASSERT(hasShaderStorageBlock(shaderStorageBlockName));
    return mShaderStorageBlockRegisterMap.find(shaderStorageBlockName)->second;
}

bool TranslatorHLSL::hasUniformBlock(const std::string &uniformBlockName) const
{
    return (mUniformBlockRegisterMap.count(uniformBlockName) > 0);
}

unsigned int TranslatorHLSL::getUniformBlockRegister(const std::string &uniformBlockName) const
{
    ASSERT(hasUniformBlock(uniformBlockName));
    return mUniformBlockRegisterMap.find(uniformBlockName)->second;
}

const std::map<std::string, unsigned int> *TranslatorHLSL::getUniformRegisterMap() const
{
    return &mUniformRegisterMap;
}

const std::set<std::string> *TranslatorHLSL::getSlowCompilingUniformBlockSet() const
{
    return &mSlowCompilingUniformBlockSet;
}

unsigned int TranslatorHLSL::getReadonlyImage2DRegisterIndex() const
{
    return mReadonlyImage2DRegisterIndex;
}

unsigned int TranslatorHLSL::getImage2DRegisterIndex() const
{
    return mImage2DRegisterIndex;
}

const std::set<std::string> *TranslatorHLSL::getUsedImage2DFunctionNames() const
{
    return &mUsedImage2DFunctionNames;
}

bool TranslatorHLSL::shouldUniformBlockUseStructuredBuffer(
    const std::string &uniformBlockName) const
{
    auto uniformBlockIter = mUniformBlockUseStructuredBufferMap.find(uniformBlockName);
    return uniformBlockIter != mUniformBlockUseStructuredBufferMap.end() &&
           uniformBlockIter->second;
}

}  // namespace sh

kc3-lang/angle/src/compiler/translator/TranslatorHLSL.cpp

Commit

kc3-lang/angle /src/compiler/translator/TranslatorHLSL.cpp