Files
llvm/mlir/lib/Interfaces/DataLayoutInterfaces.cpp

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

848 lines
31 KiB
C++
Raw Normal View History

//===- DataLayoutInterfaces.cpp - Data Layout Interface Implementation ----===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
#include "mlir/Interfaces/DataLayoutInterfaces.h"
#include "mlir/IR/BuiltinDialect.h"
#include "mlir/IR/BuiltinOps.h"
#include "mlir/IR/BuiltinTypes.h"
#include "mlir/IR/Operation.h"
#include "llvm/ADT/TypeSwitch.h"
#include "llvm/Support/MathExtras.h"
using namespace mlir;
//===----------------------------------------------------------------------===//
// Default implementations
//===----------------------------------------------------------------------===//
/// Reports that the given type is missing the data layout information and
/// exits.
[[noreturn]] static void reportMissingDataLayout(Type type) {
std::string message;
llvm::raw_string_ostream os(message);
os << "neither the scoping op nor the type class provide data layout "
"information for "
<< type;
llvm::report_fatal_error(Twine(message));
}
/// Returns the bitwidth of the index type if specified in the param list.
/// Assumes 64-bit index otherwise.
static uint64_t getIndexBitwidth(DataLayoutEntryListRef params) {
if (params.empty())
return 64;
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
auto attr = cast<IntegerAttr>(params.front().getValue());
return attr.getValue().getZExtValue();
}
llvm::TypeSize
mlir::detail::getDefaultTypeSize(Type type, const DataLayout &dataLayout,
ArrayRef<DataLayoutEntryInterface> params) {
llvm::TypeSize bits = getDefaultTypeSizeInBits(type, dataLayout, params);
return divideCeil(bits, 8);
}
llvm::TypeSize
mlir::detail::getDefaultTypeSizeInBits(Type type, const DataLayout &dataLayout,
DataLayoutEntryListRef params) {
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
if (isa<IntegerType, FloatType>(type))
return llvm::TypeSize::getFixed(type.getIntOrFloatBitWidth());
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
if (auto ctype = dyn_cast<ComplexType>(type)) {
Type et = ctype.getElementType();
uint64_t innerAlignment =
getDefaultPreferredAlignment(et, dataLayout, params) * 8;
llvm::TypeSize innerSize = getDefaultTypeSizeInBits(et, dataLayout, params);
// Include padding required to align the imaginary value in the complex
// type.
return llvm::alignTo(innerSize, innerAlignment) + innerSize;
}
// Index is an integer of some bitwidth.
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
if (isa<IndexType>(type))
return dataLayout.getTypeSizeInBits(
IntegerType::get(type.getContext(), getIndexBitwidth(params)));
// Sizes of vector types are rounded up to those of types with closest
// power-of-two number of elements in the innermost dimension. We also assume
// there is no bit-packing at the moment element sizes are taken in bytes and
// multiplied with 8 bits.
// TODO: make this extensible.
if (auto vecType = dyn_cast<VectorType>(type)) {
uint64_t baseSize = vecType.getNumElements() / vecType.getShape().back() *
llvm::PowerOf2Ceil(vecType.getShape().back()) *
dataLayout.getTypeSize(vecType.getElementType()) * 8;
return llvm::TypeSize::get(baseSize, vecType.isScalable());
}
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
if (auto typeInterface = dyn_cast<DataLayoutTypeInterface>(type))
return typeInterface.getTypeSizeInBits(dataLayout, params);
reportMissingDataLayout(type);
}
static DataLayoutEntryInterface
findEntryForIntegerType(IntegerType intType,
ArrayRef<DataLayoutEntryInterface> params) {
assert(!params.empty() && "expected non-empty parameter list");
std::map<unsigned, DataLayoutEntryInterface> sortedParams;
for (DataLayoutEntryInterface entry : params) {
sortedParams.insert(std::make_pair(
entry.getKey().get<Type>().getIntOrFloatBitWidth(), entry));
}
auto iter = sortedParams.lower_bound(intType.getWidth());
if (iter == sortedParams.end())
iter = std::prev(iter);
return iter->second;
}
constexpr const static uint64_t kDefaultBitsInByte = 8u;
static uint64_t extractABIAlignment(DataLayoutEntryInterface entry) {
auto values =
cast<DenseIntElementsAttr>(entry.getValue()).getValues<uint64_t>();
return static_cast<uint64_t>(*values.begin()) / kDefaultBitsInByte;
}
static uint64_t
getIntegerTypeABIAlignment(IntegerType intType,
ArrayRef<DataLayoutEntryInterface> params) {
constexpr uint64_t kDefaultSmallIntAlignment = 4u;
constexpr unsigned kSmallIntSize = 64;
if (params.empty()) {
return intType.getWidth() < kSmallIntSize
? llvm::PowerOf2Ceil(
llvm::divideCeil(intType.getWidth(), kDefaultBitsInByte))
: kDefaultSmallIntAlignment;
}
return extractABIAlignment(findEntryForIntegerType(intType, params));
}
static uint64_t
getFloatTypeABIAlignment(FloatType fltType, const DataLayout &dataLayout,
ArrayRef<DataLayoutEntryInterface> params) {
assert(params.size() <= 1 && "at most one data layout entry is expected for "
"the singleton floating-point type");
if (params.empty())
return llvm::PowerOf2Ceil(dataLayout.getTypeSize(fltType).getFixedValue());
return extractABIAlignment(params[0]);
}
uint64_t mlir::detail::getDefaultABIAlignment(
Type type, const DataLayout &dataLayout,
ArrayRef<DataLayoutEntryInterface> params) {
// Natural alignment is the closest power-of-two number above. For scalable
// vectors, aligning them to the same as the base vector is sufficient.
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
if (isa<VectorType>(type))
return llvm::PowerOf2Ceil(dataLayout.getTypeSize(type).getKnownMinValue());
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
if (auto fltType = dyn_cast<FloatType>(type))
return getFloatTypeABIAlignment(fltType, dataLayout, params);
// Index is an integer of some bitwidth.
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
if (isa<IndexType>(type))
return dataLayout.getTypeABIAlignment(
IntegerType::get(type.getContext(), getIndexBitwidth(params)));
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
if (auto intType = dyn_cast<IntegerType>(type))
return getIntegerTypeABIAlignment(intType, params);
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
if (auto ctype = dyn_cast<ComplexType>(type))
return getDefaultABIAlignment(ctype.getElementType(), dataLayout, params);
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
if (auto typeInterface = dyn_cast<DataLayoutTypeInterface>(type))
return typeInterface.getABIAlignment(dataLayout, params);
reportMissingDataLayout(type);
}
static uint64_t extractPreferredAlignment(DataLayoutEntryInterface entry) {
auto values =
cast<DenseIntElementsAttr>(entry.getValue()).getValues<uint64_t>();
return *std::next(values.begin(), values.size() - 1) / kDefaultBitsInByte;
}
static uint64_t
getIntegerTypePreferredAlignment(IntegerType intType,
const DataLayout &dataLayout,
ArrayRef<DataLayoutEntryInterface> params) {
if (params.empty())
return llvm::PowerOf2Ceil(dataLayout.getTypeSize(intType).getFixedValue());
return extractPreferredAlignment(findEntryForIntegerType(intType, params));
}
static uint64_t
getFloatTypePreferredAlignment(FloatType fltType, const DataLayout &dataLayout,
ArrayRef<DataLayoutEntryInterface> params) {
assert(params.size() <= 1 && "at most one data layout entry is expected for "
"the singleton floating-point type");
if (params.empty())
return dataLayout.getTypeABIAlignment(fltType);
return extractPreferredAlignment(params[0]);
}
uint64_t mlir::detail::getDefaultPreferredAlignment(
Type type, const DataLayout &dataLayout,
ArrayRef<DataLayoutEntryInterface> params) {
// Preferred alignment is same as natural for floats and vectors.
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
if (isa<VectorType>(type))
return dataLayout.getTypeABIAlignment(type);
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
if (auto fltType = dyn_cast<FloatType>(type))
return getFloatTypePreferredAlignment(fltType, dataLayout, params);
// Preferred alignment is the closest power-of-two number above for integers
// (ABI alignment may be smaller).
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
if (auto intType = dyn_cast<IntegerType>(type))
return getIntegerTypePreferredAlignment(intType, dataLayout, params);
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
if (isa<IndexType>(type)) {
return dataLayout.getTypePreferredAlignment(
IntegerType::get(type.getContext(), getIndexBitwidth(params)));
}
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
if (auto ctype = dyn_cast<ComplexType>(type))
return getDefaultPreferredAlignment(ctype.getElementType(), dataLayout,
params);
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
if (auto typeInterface = dyn_cast<DataLayoutTypeInterface>(type))
return typeInterface.getPreferredAlignment(dataLayout, params);
reportMissingDataLayout(type);
}
std::optional<uint64_t> mlir::detail::getDefaultIndexBitwidth(
Type type, const DataLayout &dataLayout,
ArrayRef<DataLayoutEntryInterface> params) {
if (isa<IndexType>(type))
return getIndexBitwidth(params);
if (auto typeInterface = dyn_cast<DataLayoutTypeInterface>(type))
if (std::optional<uint64_t> indexBitwidth =
typeInterface.getIndexBitwidth(dataLayout, params))
return *indexBitwidth;
// Return std::nullopt for all other types, which are assumed to be non
// pointer-like types.
return std::nullopt;
}
// Returns the endianness if specified in the given entry. If the entry is empty
// the default endianness represented by an empty attribute is returned.
Attribute mlir::detail::getDefaultEndianness(DataLayoutEntryInterface entry) {
if (entry == DataLayoutEntryInterface())
return Attribute();
return entry.getValue();
}
// Returns the memory space used for alloca operations if specified in the
// given entry. If the entry is empty the default memory space represented by
// an empty attribute is returned.
Attribute
mlir::detail::getDefaultAllocaMemorySpace(DataLayoutEntryInterface entry) {
if (entry == DataLayoutEntryInterface()) {
return Attribute();
}
return entry.getValue();
}
// Returns the memory space used for the program memory space. if
// specified in the given entry. If the entry is empty the default
// memory space represented by an empty attribute is returned.
Attribute
mlir::detail::getDefaultProgramMemorySpace(DataLayoutEntryInterface entry) {
if (entry == DataLayoutEntryInterface()) {
return Attribute();
}
return entry.getValue();
}
// Returns the memory space used for global the global memory space. if
// specified in the given entry. If the entry is empty the default memory
// space represented by an empty attribute is returned.
Attribute
mlir::detail::getDefaultGlobalMemorySpace(DataLayoutEntryInterface entry) {
if (entry == DataLayoutEntryInterface()) {
return Attribute();
}
return entry.getValue();
}
// Returns the stack alignment if specified in the given entry. If the entry is
// empty the default alignment zero is returned.
uint64_t
mlir::detail::getDefaultStackAlignment(DataLayoutEntryInterface entry) {
if (entry == DataLayoutEntryInterface())
return 0;
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
auto value = cast<IntegerAttr>(entry.getValue());
return value.getValue().getZExtValue();
}
std::optional<Attribute>
mlir::detail::getDevicePropertyValue(DataLayoutEntryInterface entry) {
if (entry == DataLayoutEntryInterface())
return std::nullopt;
return entry.getValue();
}
DataLayoutEntryList
mlir::detail::filterEntriesForType(DataLayoutEntryListRef entries,
TypeID typeID) {
return llvm::to_vector<4>(llvm::make_filter_range(
entries, [typeID](DataLayoutEntryInterface entry) {
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This patch updates all remaining uses of the deprecated functionality in mlir/. This was done with clang-tidy as described below and further modifications to GPUBase.td and OpenMPOpsInterfaces.td. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc ``` Differential Revision: https://reviews.llvm.org/D151542
2023-05-26 10:17:47 +02:00
auto type = llvm::dyn_cast_if_present<Type>(entry.getKey());
return type && type.getTypeID() == typeID;
}));
}
DataLayoutEntryInterface
mlir::detail::filterEntryForIdentifier(DataLayoutEntryListRef entries,
StringAttr id) {
const auto *it = llvm::find_if(entries, [id](DataLayoutEntryInterface entry) {
if (!entry.getKey().is<StringAttr>())
return false;
return entry.getKey().get<StringAttr>() == id;
});
return it == entries.end() ? DataLayoutEntryInterface() : *it;
}
static DataLayoutSpecInterface getSpec(Operation *operation) {
return llvm::TypeSwitch<Operation *, DataLayoutSpecInterface>(operation)
.Case<ModuleOp, DataLayoutOpInterface>(
[&](auto op) { return op.getDataLayoutSpec(); })
.Default([](Operation *) {
llvm_unreachable("expected an op with data layout spec");
return DataLayoutSpecInterface();
});
}
static TargetSystemSpecInterface getTargetSystemSpec(Operation *operation) {
if (operation) {
ModuleOp moduleOp = dyn_cast<ModuleOp>(operation);
if (!moduleOp)
moduleOp = operation->getParentOfType<ModuleOp>();
return moduleOp.getTargetSystemSpec();
}
return TargetSystemSpecInterface();
}
/// Populates `opsWithLayout` with the list of proper ancestors of `leaf` that
/// are either modules or implement the `DataLayoutOpInterface`.
static void
collectParentLayouts(Operation *leaf,
SmallVectorImpl<DataLayoutSpecInterface> &specs,
SmallVectorImpl<Location> *opLocations = nullptr) {
if (!leaf)
return;
for (Operation *parent = leaf->getParentOp(); parent != nullptr;
parent = parent->getParentOp()) {
llvm::TypeSwitch<Operation *>(parent)
.Case<ModuleOp>([&](ModuleOp op) {
// Skip top-level module op unless it has a layout. Top-level module
// without layout is most likely the one implicitly added by the
// parser and it doesn't have location. Top-level null specification
// would have had the same effect as not having a specification at all
// (using type defaults).
if (!op->getParentOp() && !op.getDataLayoutSpec())
return;
specs.push_back(op.getDataLayoutSpec());
if (opLocations)
opLocations->push_back(op.getLoc());
})
.Case<DataLayoutOpInterface>([&](DataLayoutOpInterface op) {
specs.push_back(op.getDataLayoutSpec());
if (opLocations)
opLocations->push_back(op.getLoc());
});
}
}
/// Returns a layout spec that is a combination of the layout specs attached
/// to the given operation and all its ancestors.
static DataLayoutSpecInterface getCombinedDataLayout(Operation *leaf) {
if (!leaf)
return {};
assert((isa<ModuleOp, DataLayoutOpInterface>(leaf)) &&
"expected an op with data layout spec");
SmallVector<DataLayoutOpInterface> opsWithLayout;
SmallVector<DataLayoutSpecInterface> specs;
collectParentLayouts(leaf, specs);
// Fast track if there are no ancestors.
if (specs.empty())
return getSpec(leaf);
// Create the list of non-null specs (null/missing specs can be safely
// ignored) from the outermost to the innermost.
auto nonNullSpecs = llvm::to_vector<2>(llvm::make_filter_range(
llvm::reverse(specs),
[](DataLayoutSpecInterface iface) { return iface != nullptr; }));
// Combine the specs using the innermost as anchor.
if (DataLayoutSpecInterface current = getSpec(leaf))
return current.combineWith(nonNullSpecs);
if (nonNullSpecs.empty())
return {};
return nonNullSpecs.back().combineWith(
llvm::ArrayRef(nonNullSpecs).drop_back());
}
LogicalResult mlir::detail::verifyDataLayoutOp(Operation *op) {
DataLayoutSpecInterface spec = getSpec(op);
// The layout specification may be missing and it's fine.
if (!spec)
return success();
if (failed(spec.verifySpec(op->getLoc())))
return failure();
if (!getCombinedDataLayout(op)) {
InFlightDiagnostic diag =
op->emitError()
<< "data layout does not combine with layouts of enclosing ops";
SmallVector<DataLayoutSpecInterface> specs;
SmallVector<Location> opLocations;
collectParentLayouts(op, specs, &opLocations);
for (Location loc : opLocations)
diag.attachNote(loc) << "enclosing op with data layout";
return diag;
}
return success();
}
llvm::TypeSize mlir::detail::divideCeil(llvm::TypeSize numerator,
uint64_t denominator) {
uint64_t divided =
llvm::divideCeil(numerator.getKnownMinValue(), denominator);
return llvm::TypeSize::get(divided, numerator.isScalable());
}
//===----------------------------------------------------------------------===//
// DataLayout
//===----------------------------------------------------------------------===//
template <typename OpTy>
void checkMissingLayout(DataLayoutSpecInterface originalLayout, OpTy op) {
if (!originalLayout) {
assert((!op || !op.getDataLayoutSpec()) &&
"could not compute layout information for an op (failed to "
"combine attributes?)");
}
}
mlir::DataLayout::DataLayout() : DataLayout(ModuleOp()) {}
mlir::DataLayout::DataLayout(DataLayoutOpInterface op)
: originalLayout(getCombinedDataLayout(op)),
originalTargetSystemDesc(getTargetSystemSpec(op)), scope(op),
allocaMemorySpace(std::nullopt), programMemorySpace(std::nullopt),
globalMemorySpace(std::nullopt), stackAlignment(std::nullopt) {
#if LLVM_ENABLE_ABI_BREAKING_CHECKS
checkMissingLayout(originalLayout, op);
collectParentLayouts(op, layoutStack);
#endif
}
mlir::DataLayout::DataLayout(ModuleOp op)
: originalLayout(getCombinedDataLayout(op)),
originalTargetSystemDesc(getTargetSystemSpec(op)), scope(op),
allocaMemorySpace(std::nullopt), programMemorySpace(std::nullopt),
globalMemorySpace(std::nullopt), stackAlignment(std::nullopt) {
#if LLVM_ENABLE_ABI_BREAKING_CHECKS
checkMissingLayout(originalLayout, op);
collectParentLayouts(op, layoutStack);
#endif
}
mlir::DataLayout mlir::DataLayout::closest(Operation *op) {
// Search the closest parent either being a module operation or implementing
// the data layout interface.
while (op) {
if (auto module = dyn_cast<ModuleOp>(op))
return DataLayout(module);
if (auto iface = dyn_cast<DataLayoutOpInterface>(op))
return DataLayout(iface);
op = op->getParentOp();
}
return DataLayout();
}
void mlir::DataLayout::checkValid() const {
#if LLVM_ENABLE_ABI_BREAKING_CHECKS
SmallVector<DataLayoutSpecInterface> specs;
collectParentLayouts(scope, specs);
assert(specs.size() == layoutStack.size() &&
"data layout object used, but no longer valid due to the change in "
"number of nested layouts");
for (auto pair : llvm::zip(specs, layoutStack)) {
Attribute newLayout = std::get<0>(pair);
Attribute origLayout = std::get<1>(pair);
assert(newLayout == origLayout &&
"data layout object used, but no longer valid "
"due to the change in layout attributes");
}
#endif
assert(((!scope && !this->originalLayout) ||
(scope && this->originalLayout == getCombinedDataLayout(scope))) &&
"data layout object used, but no longer valid due to the change in "
"layout spec");
}
/// Looks up the value for the given type key in the given cache. If there is no
/// such value in the cache, compute it using the given callback and put it in
/// the cache before returning.
template <typename T>
static T cachedLookup(Type t, DenseMap<Type, T> &cache,
function_ref<T(Type)> compute) {
auto it = cache.find(t);
if (it != cache.end())
return it->second;
auto result = cache.try_emplace(t, compute(t));
return result.first->second;
}
llvm::TypeSize mlir::DataLayout::getTypeSize(Type t) const {
checkValid();
return cachedLookup<llvm::TypeSize>(t, sizes, [&](Type ty) {
DataLayoutEntryList list;
if (originalLayout)
list = originalLayout.getSpecForType(ty.getTypeID());
if (auto iface = dyn_cast_or_null<DataLayoutOpInterface>(scope))
return iface.getTypeSize(ty, *this, list);
return detail::getDefaultTypeSize(ty, *this, list);
});
}
llvm::TypeSize mlir::DataLayout::getTypeSizeInBits(Type t) const {
checkValid();
return cachedLookup<llvm::TypeSize>(t, bitsizes, [&](Type ty) {
DataLayoutEntryList list;
if (originalLayout)
list = originalLayout.getSpecForType(ty.getTypeID());
if (auto iface = dyn_cast_or_null<DataLayoutOpInterface>(scope))
return iface.getTypeSizeInBits(ty, *this, list);
return detail::getDefaultTypeSizeInBits(ty, *this, list);
});
}
uint64_t mlir::DataLayout::getTypeABIAlignment(Type t) const {
checkValid();
return cachedLookup<uint64_t>(t, abiAlignments, [&](Type ty) {
DataLayoutEntryList list;
if (originalLayout)
list = originalLayout.getSpecForType(ty.getTypeID());
if (auto iface = dyn_cast_or_null<DataLayoutOpInterface>(scope))
return iface.getTypeABIAlignment(ty, *this, list);
return detail::getDefaultABIAlignment(ty, *this, list);
});
}
uint64_t mlir::DataLayout::getTypePreferredAlignment(Type t) const {
checkValid();
return cachedLookup<uint64_t>(t, preferredAlignments, [&](Type ty) {
DataLayoutEntryList list;
if (originalLayout)
list = originalLayout.getSpecForType(ty.getTypeID());
if (auto iface = dyn_cast_or_null<DataLayoutOpInterface>(scope))
return iface.getTypePreferredAlignment(ty, *this, list);
return detail::getDefaultPreferredAlignment(ty, *this, list);
});
}
std::optional<uint64_t> mlir::DataLayout::getTypeIndexBitwidth(Type t) const {
checkValid();
return cachedLookup<std::optional<uint64_t>>(t, indexBitwidths, [&](Type ty) {
DataLayoutEntryList list;
if (originalLayout)
list = originalLayout.getSpecForType(ty.getTypeID());
if (auto iface = dyn_cast_or_null<DataLayoutOpInterface>(scope))
return iface.getIndexBitwidth(ty, *this, list);
return detail::getDefaultIndexBitwidth(ty, *this, list);
});
}
mlir::Attribute mlir::DataLayout::getEndianness() const {
checkValid();
if (endianness)
return *endianness;
DataLayoutEntryInterface entry;
if (originalLayout)
entry = originalLayout.getSpecForIdentifier(
originalLayout.getEndiannessIdentifier(originalLayout.getContext()));
if (auto iface = dyn_cast_or_null<DataLayoutOpInterface>(scope))
endianness = iface.getEndianness(entry);
else
endianness = detail::getDefaultEndianness(entry);
return *endianness;
}
mlir::Attribute mlir::DataLayout::getAllocaMemorySpace() const {
checkValid();
if (allocaMemorySpace)
return *allocaMemorySpace;
DataLayoutEntryInterface entry;
if (originalLayout)
entry = originalLayout.getSpecForIdentifier(
originalLayout.getAllocaMemorySpaceIdentifier(
originalLayout.getContext()));
if (auto iface = dyn_cast_or_null<DataLayoutOpInterface>(scope))
allocaMemorySpace = iface.getAllocaMemorySpace(entry);
else
allocaMemorySpace = detail::getDefaultAllocaMemorySpace(entry);
return *allocaMemorySpace;
}
mlir::Attribute mlir::DataLayout::getProgramMemorySpace() const {
checkValid();
if (programMemorySpace)
return *programMemorySpace;
DataLayoutEntryInterface entry;
if (originalLayout)
entry = originalLayout.getSpecForIdentifier(
originalLayout.getProgramMemorySpaceIdentifier(
originalLayout.getContext()));
if (auto iface = dyn_cast_or_null<DataLayoutOpInterface>(scope))
programMemorySpace = iface.getProgramMemorySpace(entry);
else
programMemorySpace = detail::getDefaultProgramMemorySpace(entry);
return *programMemorySpace;
}
mlir::Attribute mlir::DataLayout::getGlobalMemorySpace() const {
checkValid();
if (globalMemorySpace)
return *globalMemorySpace;
DataLayoutEntryInterface entry;
if (originalLayout)
entry = originalLayout.getSpecForIdentifier(
originalLayout.getGlobalMemorySpaceIdentifier(
originalLayout.getContext()));
if (auto iface = dyn_cast_or_null<DataLayoutOpInterface>(scope))
globalMemorySpace = iface.getGlobalMemorySpace(entry);
else
globalMemorySpace = detail::getDefaultGlobalMemorySpace(entry);
return *globalMemorySpace;
}
uint64_t mlir::DataLayout::getStackAlignment() const {
checkValid();
if (stackAlignment)
return *stackAlignment;
DataLayoutEntryInterface entry;
if (originalLayout)
entry = originalLayout.getSpecForIdentifier(
originalLayout.getStackAlignmentIdentifier(
originalLayout.getContext()));
if (auto iface = dyn_cast_or_null<DataLayoutOpInterface>(scope))
stackAlignment = iface.getStackAlignment(entry);
else
stackAlignment = detail::getDefaultStackAlignment(entry);
return *stackAlignment;
}
std::optional<Attribute> mlir::DataLayout::getDevicePropertyValue(
TargetSystemSpecInterface::DeviceID deviceID,
StringAttr propertyName) const {
checkValid();
DataLayoutEntryInterface entry;
if (originalTargetSystemDesc) {
if (std::optional<TargetDeviceSpecInterface> device =
originalTargetSystemDesc.getDeviceSpecForDeviceID(deviceID))
entry = device->getSpecForIdentifier(propertyName);
}
// Currently I am not caching the results because we do not return
// default values of these properties. Instead if the property is
// missing, we return std::nullopt so that the users can resort to
// the default value however they want.
if (auto iface = dyn_cast_or_null<DataLayoutOpInterface>(scope))
return iface.getDevicePropertyValue(entry);
else
return detail::getDevicePropertyValue(entry);
}
//===----------------------------------------------------------------------===//
// DataLayoutSpecInterface
//===----------------------------------------------------------------------===//
void DataLayoutSpecInterface::bucketEntriesByType(
DenseMap<TypeID, DataLayoutEntryList> &types,
DenseMap<StringAttr, DataLayoutEntryInterface> &ids) {
for (DataLayoutEntryInterface entry : getEntries()) {
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This patch updates all remaining uses of the deprecated functionality in mlir/. This was done with clang-tidy as described below and further modifications to GPUBase.td and OpenMPOpsInterfaces.td. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc ``` Differential Revision: https://reviews.llvm.org/D151542
2023-05-26 10:17:47 +02:00
if (auto type = llvm::dyn_cast_if_present<Type>(entry.getKey()))
types[type.getTypeID()].push_back(entry);
else
ids[entry.getKey().get<StringAttr>()] = entry;
}
}
LogicalResult mlir::detail::verifyDataLayoutSpec(DataLayoutSpecInterface spec,
Location loc) {
// First, verify individual entries.
for (DataLayoutEntryInterface entry : spec.getEntries())
if (failed(entry.verifyEntry(loc)))
return failure();
// Second, dispatch verifications of entry groups to types or dialects they
// are associated with.
DenseMap<TypeID, DataLayoutEntryList> types;
DenseMap<StringAttr, DataLayoutEntryInterface> ids;
spec.bucketEntriesByType(types, ids);
for (const auto &kvp : types) {
auto sampleType = kvp.second.front().getKey().get<Type>();
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
if (isa<IndexType>(sampleType)) {
assert(kvp.second.size() == 1 &&
"expected one data layout entry for non-parametric 'index' type");
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
if (!isa<IntegerAttr>(kvp.second.front().getValue()))
return emitError(loc)
<< "expected integer attribute in the data layout entry for "
<< sampleType;
continue;
}
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
if (isa<IntegerType, FloatType>(sampleType)) {
for (DataLayoutEntryInterface entry : kvp.second) {
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
auto value = dyn_cast<DenseIntElementsAttr>(entry.getValue());
if (!value || !value.getElementType().isSignlessInteger(64)) {
emitError(loc) << "expected a dense i64 elements attribute in the "
"data layout entry "
<< entry;
return failure();
}
auto elements = llvm::to_vector<2>(value.getValues<uint64_t>());
unsigned numElements = elements.size();
if (numElements < 1 || numElements > 2) {
emitError(loc) << "expected 1 or 2 elements in the data layout entry "
<< entry;
return failure();
}
uint64_t abi = elements[0];
uint64_t preferred = numElements == 2 ? elements[1] : abi;
if (preferred < abi) {
emitError(loc)
<< "preferred alignment is expected to be greater than or equal "
"to the abi alignment in data layout entry "
<< entry;
return failure();
}
}
continue;
}
if (isa<BuiltinDialect>(&sampleType.getDialect()))
return emitError(loc) << "unexpected data layout for a built-in type";
[mlir] Move casting calls from methods to function calls The MLIR classes Type/Attribute/Operation/Op/Value support cast/dyn_cast/isa/dyn_cast_or_null functionality through llvm's doCast functionality in addition to defining methods with the same name. This change begins the migration of uses of the method to the corresponding function call as has been decided as more consistent. Note that there still exist classes that only define methods directly, such as AffineExpr, and this does not include work currently to support a functional cast/isa call. Caveats include: - This clang-tidy script probably has more problems. - This only touches C++ code, so nothing that is being generated. Context: - https://mlir.llvm.org/deprecation/ at "Use the free function variants for dyn_cast/cast/isa/…" - Original discussion at https://discourse.llvm.org/t/preferred-casting-style-going-forward/68443 Implementation: This first patch was created with the following steps. The intention is to only do automated changes at first, so I waste less time if it's reverted, and so the first mass change is more clear as an example to other teams that will need to follow similar steps. Steps are described per line, as comments are removed by git: 0. Retrieve the change from the following to build clang-tidy with an additional check: https://github.com/llvm/llvm-project/compare/main...tpopp:llvm-project:tidy-cast-check 1. Build clang-tidy 2. Run clang-tidy over your entire codebase while disabling all checks and enabling the one relevant one. Run on all header files also. 3. Delete .inc files that were also modified, so the next build rebuilds them to a pure state. 4. Some changes have been deleted for the following reasons: - Some files had a variable also named cast - Some files had not included a header file that defines the cast functions - Some files are definitions of the classes that have the casting methods, so the code still refers to the method instead of the function without adding a prefix or removing the method declaration at the same time. ``` ninja -C $BUILD_DIR clang-tidy run-clang-tidy -clang-tidy-binary=$BUILD_DIR/bin/clang-tidy -checks='-*,misc-cast-functions'\ -header-filter=mlir/ mlir/* -fix rm -rf $BUILD_DIR/tools/mlir/**/*.inc git restore mlir/lib/IR mlir/lib/Dialect/DLTI/DLTI.cpp\ mlir/lib/Dialect/Complex/IR/ComplexDialect.cpp\ mlir/lib/**/IR/\ mlir/lib/Dialect/SparseTensor/Transforms/SparseVectorization.cpp\ mlir/lib/Dialect/Vector/Transforms/LowerVectorMultiReduction.cpp\ mlir/test/lib/Dialect/Test/TestTypes.cpp\ mlir/test/lib/Dialect/Transform/TestTransformDialectExtension.cpp\ mlir/test/lib/Dialect/Test/TestAttributes.cpp\ mlir/unittests/TableGen/EnumsGenTest.cpp\ mlir/test/python/lib/PythonTestCAPI.cpp\ mlir/include/mlir/IR/ ``` Differential Revision: https://reviews.llvm.org/D150123
2023-05-08 16:33:54 +02:00
auto dlType = dyn_cast<DataLayoutTypeInterface>(sampleType);
if (!dlType)
return emitError(loc)
<< "data layout specified for a type that does not support it";
if (failed(dlType.verifyEntries(kvp.second, loc)))
return failure();
}
for (const auto &kvp : ids) {
StringAttr identifier = kvp.second.getKey().get<StringAttr>();
Dialect *dialect = identifier.getReferencedDialect();
// Ignore attributes that belong to an unknown dialect, the dialect may
// actually implement the relevant interface but we don't know about that.
if (!dialect)
continue;
const auto *iface = dyn_cast<DataLayoutDialectInterface>(dialect);
if (!iface) {
return emitError(loc)
<< "the '" << dialect->getNamespace()
<< "' dialect does not support identifier data layout entries";
}
if (failed(iface->verifyEntry(kvp.second, loc)))
return failure();
}
return success();
}
LogicalResult
mlir::detail::verifyTargetSystemSpec(TargetSystemSpecInterface spec,
Location loc) {
DenseMap<StringAttr, DataLayoutEntryInterface> deviceDescKeys;
DenseSet<TargetSystemSpecInterface::DeviceID> deviceIDs;
for (const auto &entry : spec.getEntries()) {
auto targetDeviceSpec =
dyn_cast<TargetDeviceSpecInterface>(entry.getValue());
if (!targetDeviceSpec)
return failure();
// First, verify individual target device desc specs.
if (failed(targetDeviceSpec.verifyEntry(loc)))
return failure();
// Check that device IDs are unique across all entries.
auto deviceID =
llvm::dyn_cast<TargetSystemSpecInterface::DeviceID>(entry.getKey());
if (!deviceID)
return failure();
if (!deviceIDs.insert(deviceID).second) {
return failure();
}
// collect all the keys used by all the target device specs.
for (DataLayoutEntryInterface entry : targetDeviceSpec.getEntries()) {
if (auto type = llvm::dyn_cast_if_present<Type>(entry.getKey())) {
// targetDeviceSpec does not support Type as a key.
return failure();
} else {
deviceDescKeys[entry.getKey().get<StringAttr>()] = entry;
}
}
}
for (const auto &[keyName, keyVal] : deviceDescKeys) {
Dialect *dialect = keyName.getReferencedDialect();
// Ignore attributes that belong to an unknown dialect, the dialect may
// actually implement the relevant interface but we don't know about that.
if (!dialect)
return failure();
const auto *iface = dyn_cast<DataLayoutDialectInterface>(dialect);
if (!iface) {
return emitError(loc)
<< "the '" << dialect->getNamespace()
<< "' dialect does not support identifier data layout entries";
}
if (failed(iface->verifyEntry(keyVal, loc)))
return failure();
}
return success();
}
#include "mlir/Interfaces/DataLayoutAttrInterface.cpp.inc"
#include "mlir/Interfaces/DataLayoutOpInterface.cpp.inc"
#include "mlir/Interfaces/DataLayoutTypeInterface.cpp.inc"