Google Git Google Git
Sign in
llvm/llvm-project/refs/heads/users/arsenm/codegen/remove-targetoptions-loopalignment/./flang/lib/Lower/ConvertExprToHLFIR.cpp
blob: 24c990141be552c8321cdd3547a7bd91c83a9aae [file] [edit]
//===-- ConvertExprToHLFIR.cpp --------------------------------------------===//
//
// 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
//
//===----------------------------------------------------------------------===//
//
// Coding style: https://mlir.llvm.org/getting_started/DeveloperGuide/
//
//===----------------------------------------------------------------------===//
#include "flang/Lower/ConvertExprToHLFIR.h"
#include "flang/Evaluate/shape.h"
#include "flang/Evaluate/tools.h"
#include "flang/Lower/AbstractConverter.h"
#include "flang/Lower/Allocatable.h"
#include "flang/Lower/CallInterface.h"
#include "flang/Lower/ConvertArrayConstructor.h"
#include "flang/Lower/ConvertCall.h"
#include "flang/Lower/ConvertConstant.h"
#include "flang/Lower/ConvertProcedureDesignator.h"
#include "flang/Lower/ConvertType.h"
#include "flang/Lower/ConvertVariable.h"
#include "flang/Lower/StatementContext.h"
#include "flang/Lower/SymbolMap.h"
#include "flang/Optimizer/Builder/Complex.h"
#include "flang/Optimizer/Builder/IntrinsicCall.h"
#include "flang/Optimizer/Builder/MutableBox.h"
#include "flang/Optimizer/Builder/Runtime/Derived.h"
#include "flang/Optimizer/Builder/Runtime/Pointer.h"
#include "flang/Optimizer/Builder/Todo.h"
#include "flang/Optimizer/Dialect/FIRAttr.h"
#include "flang/Optimizer/HLFIR/HLFIROps.h"
#include "mlir/IR/IRMapping.h"
#include "llvm/ADT/STLExtras.h"
#include "llvm/ADT/TypeSwitch.h"
#include <optional>
namespace {
// This was modelled after isParenthesizedVariable()
template <typename T>
static bool isParenthesized(const Fortran::evaluate::Expr<T> &expr) {
using ExprVariant = decltype(Fortran::evaluate::Expr<T>::u);
using Parentheses = Fortran::evaluate::Parentheses<T>;
if constexpr (Fortran::common::HasMember<Parentheses, ExprVariant>) {
return std::get_if<Parentheses>(&expr.u) != nullptr;
} else {
return Fortran::common::visit(
[&](const auto &x) { return isParenthesized(x); }, expr.u);
}
}
/// Lower Designators to HLFIR.
class HlfirDesignatorBuilder {
private:
/// Internal entry point on the rightest part of a evaluate::Designator.
template <typename T>
hlfir::EntityWithAttributes
genLeafPartRef(const T &designatorNode,
bool vectorSubscriptDesignatorToValue) {
hlfir::EntityWithAttributes result = gen(designatorNode);
if (vectorSubscriptDesignatorToValue)
return turnVectorSubscriptedDesignatorIntoValue(result);
return result;
}
hlfir::EntityWithAttributes
genDesignatorExpr(const Fortran::lower::SomeExpr &designatorExpr,
bool vectorSubscriptDesignatorToValue = true);
public:
HlfirDesignatorBuilder(mlir::Location loc,
Fortran::lower::AbstractConverter &converter,
Fortran::lower::SymMap &symMap,
Fortran::lower::StatementContext &stmtCtx)
: converter{converter}, symMap{symMap}, stmtCtx{stmtCtx}, loc{loc} {}
/// Public entry points to lower a Designator<T> (given its .u member, to
/// avoid the template arguments which does not matter here).
/// This lowers a designator to an hlfir variable SSA value (that can be
/// assigned to), except for vector subscripted designators that are
/// lowered by default to hlfir.expr value since they cannot be
/// represented as HLFIR variable SSA values.
// Character designators variant contains substrings
using CharacterDesignators =
decltype(Fortran::evaluate::Designator<Fortran::evaluate::Type<
Fortran::evaluate::TypeCategory::Character, 1>>::u);
hlfir::EntityWithAttributes
gen(const CharacterDesignators &designatorVariant,
bool vectorSubscriptDesignatorToValue = true) {
return Fortran::common::visit(
[&](const auto &x) -> hlfir::EntityWithAttributes {
return genLeafPartRef(x, vectorSubscriptDesignatorToValue);
},
designatorVariant);
}
// Character designators variant contains complex parts
using RealDesignators =
decltype(Fortran::evaluate::Designator<Fortran::evaluate::Type<
Fortran::evaluate::TypeCategory::Real, 4>>::u);
hlfir::EntityWithAttributes
gen(const RealDesignators &designatorVariant,
bool vectorSubscriptDesignatorToValue = true) {
return Fortran::common::visit(
[&](const auto &x) -> hlfir::EntityWithAttributes {
return genLeafPartRef(x, vectorSubscriptDesignatorToValue);
},
designatorVariant);
}
// All other designators are similar
using OtherDesignators =
decltype(Fortran::evaluate::Designator<Fortran::evaluate::Type<
Fortran::evaluate::TypeCategory::Integer, 4>>::u);
hlfir::EntityWithAttributes
gen(const OtherDesignators &designatorVariant,
bool vectorSubscriptDesignatorToValue = true) {
return Fortran::common::visit(
[&](const auto &x) -> hlfir::EntityWithAttributes {
return genLeafPartRef(x, vectorSubscriptDesignatorToValue);
},
designatorVariant);
}
hlfir::EntityWithAttributes
genNamedEntity(const Fortran::evaluate::NamedEntity &namedEntity,
bool vectorSubscriptDesignatorToValue = true) {
if (namedEntity.IsSymbol())
return genLeafPartRef(
Fortran::evaluate::SymbolRef{namedEntity.GetLastSymbol()},
vectorSubscriptDesignatorToValue);
return genLeafPartRef(namedEntity.GetComponent(),
vectorSubscriptDesignatorToValue);
}
/// Public entry point to lower a vector subscripted designator to
/// an hlfir::ElementalAddrOp.
hlfir::ElementalAddrOp convertVectorSubscriptedExprToElementalAddr(
const Fortran::lower::SomeExpr &designatorExpr);
std::tuple<mlir::Type, fir::FortranVariableFlagsEnum>
genComponentDesignatorTypeAndAttributes(
const Fortran::semantics::Symbol &componentSym, mlir::Type fieldType,
bool isVolatile) {
if (mayHaveNonDefaultLowerBounds(componentSym)) {
mlir::Type boxType =
fir::BoxType::get(fieldType, isVolatile, componentSym.Corank());
return std::make_tuple(boxType,
fir::FortranVariableFlagsEnum::contiguous);
}
auto refType = fir::ReferenceType::get(fieldType, isVolatile);
return std::make_tuple(refType, fir::FortranVariableFlagsEnum{});
}
mlir::Value genComponentShape(const Fortran::semantics::Symbol &componentSym,
mlir::Type fieldType) {
// For pointers and allocatable components, the
// shape is deferred and should not be loaded now to preserve
// pointer/allocatable aspects.
if (componentSym.Rank() == 0 ||
Fortran::semantics::IsAllocatableOrObjectPointer(&componentSym) ||
Fortran::semantics::IsProcedurePointer(&componentSym))
return mlir::Value{};
fir::FirOpBuilder &builder = getBuilder();
mlir::Location loc = getLoc();
mlir::Type idxTy = builder.getIndexType();
llvm::SmallVector<mlir::Value> extents;
auto seqTy = mlir::cast<fir::SequenceType>(
hlfir::getFortranElementOrSequenceType(fieldType));
for (auto extent : seqTy.getShape()) {
if (extent == fir::SequenceType::getUnknownExtent()) {
// We have already generated invalid hlfir.declare
// without the type parameters and probably invalid storage
// for the variable (e.g. fir.alloca without type parameters).
// So this TODO here is a little bit late, but it matches
// the non-HLFIR path.
TODO(loc, "array component shape depending on length parameters");
}
extents.push_back(builder.createIntegerConstant(loc, idxTy, extent));
}
if (!mayHaveNonDefaultLowerBounds(componentSym))
return fir::ShapeOp::create(builder, loc, extents);
llvm::SmallVector<mlir::Value> lbounds;
if (const auto *objDetails =
componentSym.detailsIf<Fortran::semantics::ObjectEntityDetails>())
for (const Fortran::semantics::ShapeSpec &bounds : objDetails->shape())
if (auto lb = bounds.lbound().GetExplicit())
if (auto constant = Fortran::evaluate::ToInt64(*lb))
lbounds.push_back(
builder.createIntegerConstant(loc, idxTy, *constant));
assert(extents.size() == lbounds.size() &&
"extents and lower bounds must match");
return builder.genShape(loc, lbounds, extents);
}
fir::FortranVariableOpInterface
gen(const Fortran::evaluate::DataRef &dataRef) {
return Fortran::common::visit(
Fortran::common::visitors{[&](const auto &x) { return gen(x); }},
dataRef.u);
}
private:
/// Struct that is filled while visiting a part-ref (in the "visit" member
/// function) before the top level "gen" generates an hlfir.declare for the
/// part ref. It contains the lowered pieces of the part-ref that will
/// become the operands of an hlfir.declare.
struct PartInfo {
std::optional<hlfir::Entity> base;
std::string componentName{};
mlir::Value componentShape;
hlfir::DesignateOp::Subscripts subscripts;
std::optional<bool> complexPart;
mlir::Value resultShape;
llvm::SmallVector<mlir::Value> typeParams;
llvm::SmallVector<mlir::Value, 2> substring;
};
// Given the value type of a designator (T or fir.array<T>) and the front-end
// node for the designator, compute the memory type (fir.class, fir.ref, or
// fir.box)...
template <typename T>
mlir::Type computeDesignatorType(mlir::Type resultValueType,
PartInfo &partInfo,
const T &designatorNode) {
// Get base's shape if its a sequence type with no previously computed
// result shape
if (partInfo.base && mlir::isa<fir::SequenceType>(resultValueType) &&
!partInfo.resultShape)
partInfo.resultShape =
hlfir::genShape(getLoc(), getBuilder(), *partInfo.base);
// Enable volatility on the designatory type if it has the VOLATILE
// attribute or if the base is volatile.
bool isVolatile = false;
// Check if this should be a volatile reference
if constexpr (std::is_same_v<std::decay_t<T>,
Fortran::evaluate::SymbolRef>) {
if (designatorNode.get().GetUltimate().attrs().test(
Fortran::semantics::Attr::VOLATILE))
isVolatile = true;
} else if constexpr (std::is_same_v<std::decay_t<T>,
Fortran::evaluate::ArrayRef>) {
if (designatorNode.base().GetLastSymbol().attrs().test(
Fortran::semantics::Attr::VOLATILE))
isVolatile = true;
} else if constexpr (std::is_same_v<std::decay_t<T>,
Fortran::evaluate::Component>) {
if (designatorNode.GetLastSymbol().attrs().test(
Fortran::semantics::Attr::VOLATILE))
isVolatile = true;
}
// Check if the base type is volatile
if (partInfo.base.has_value()) {
mlir::Type baseType = partInfo.base.value().getType();
isVolatile = isVolatile || fir::isa_volatile_type(baseType);
}
// Dynamic type of polymorphic base must be kept if the designator is
// polymorphic.
if (isPolymorphic(designatorNode))
return fir::ClassType::get(resultValueType, isVolatile);
// Character scalar with dynamic length needs a fir.boxchar to hold the
// designator length.
auto charType = mlir::dyn_cast<fir::CharacterType>(resultValueType);
if (charType && charType.hasDynamicLen())
return fir::BoxCharType::get(charType.getContext(), charType.getFKind());
// Arrays with non default lower bounds or dynamic length or dynamic extent
// need a fir.box to hold the dynamic or lower bound information.
if (fir::hasDynamicSize(resultValueType) ||
mayHaveNonDefaultLowerBounds(partInfo))
return fir::BoxType::get(resultValueType, isVolatile);
// Non simply contiguous ref require a fir.box to carry the byte stride.
if (mlir::isa<fir::SequenceType>(resultValueType) &&
!Fortran::evaluate::IsSimplyContiguous(
designatorNode, getConverter().getFoldingContext(),
/*namedConstantSectionsAreAlwaysContiguous=*/false))
return fir::BoxType::get(resultValueType, isVolatile);
// Other designators can be handled as raw addresses.
return fir::ReferenceType::get(resultValueType, isVolatile);
}
template <typename T>
static bool isPolymorphic(const T &designatorNode) {
if constexpr (!std::is_same_v<T, Fortran::evaluate::Substring>) {
return Fortran::semantics::IsPolymorphic(designatorNode.GetLastSymbol());
}
return false;
}
template <typename T>
/// Generate an hlfir.designate for a part-ref given a filled PartInfo and the
/// FIR type for this part-ref.
fir::FortranVariableOpInterface genDesignate(mlir::Type resultValueType,
PartInfo &partInfo,
const T &designatorNode) {
mlir::Type designatorType =
computeDesignatorType(resultValueType, partInfo, designatorNode);
return genDesignate(designatorType, partInfo, /*attributes=*/{});
}
fir::FortranVariableOpInterface
genDesignate(mlir::Type designatorType, PartInfo &partInfo,
fir::FortranVariableFlagsAttr attributes) {
fir::FirOpBuilder &builder = getBuilder();
// Once a part with vector subscripts has been lowered, the following
// hlfir.designator (for the parts on the right of the designator) must
// be lowered inside the hlfir.elemental_addr because they depend on the
// hlfir.elemental_addr indices.
// All the subsequent Fortran indices however, should be lowered before
// the hlfir.elemental_addr because they should only be evaluated once,
// hence, the insertion point is restored outside of the
// hlfir.elemental_addr after generating the hlfir.designate. Example: in
// "X(VECTOR)%COMP(FOO(), BAR())", the calls to bar() and foo() must be
// generated outside of the hlfir.elemental, but the related hlfir.designate
// that depends on the scalar hlfir.designate of X(VECTOR) that was
// generated inside the hlfir.elemental_addr should be generated in the
// hlfir.elemental_addr.
if (auto elementalAddrOp = getVectorSubscriptElementAddrOp())
builder.setInsertionPointToEnd(&elementalAddrOp->getBody().front());
auto designate = hlfir::DesignateOp::create(
builder, getLoc(), designatorType, partInfo.base.value().getBase(),
partInfo.componentName, partInfo.componentShape, partInfo.subscripts,
partInfo.substring, partInfo.complexPart, partInfo.resultShape,
partInfo.typeParams, attributes);
if (auto elementalAddrOp = getVectorSubscriptElementAddrOp())
builder.setInsertionPoint(*elementalAddrOp);
return mlir::cast<fir::FortranVariableOpInterface>(
designate.getOperation());
}
fir::FortranVariableOpInterface
gen(const Fortran::evaluate::SymbolRef &symbolRef) {
if (std::optional<fir::FortranVariableOpInterface> varDef =
getSymMap().lookupVariableDefinition(symbolRef)) {
if (symbolRef.get().GetUltimate().test(
Fortran::semantics::Symbol::Flag::CrayPointee)) {
// The pointee is represented with a descriptor inheriting
// the shape and type parameters of the pointee.
// We have to update the base_addr to point to the current
// value of the Cray pointer variable.
fir::FirOpBuilder &builder = getBuilder();
fir::FortranVariableOpInterface ptrVar =
gen(Fortran::semantics::GetCrayPointer(symbolRef));
mlir::Value ptrAddr = ptrVar.getBase();
// Reinterpret the reference to a Cray pointer so that
// we have a pointer-compatible value after loading
// the Cray pointer value.
mlir::Type refPtrType = builder.getRefType(
fir::PointerType::get(fir::dyn_cast_ptrEleTy(ptrAddr.getType())));
mlir::Value cast = builder.createConvert(loc, refPtrType, ptrAddr);
mlir::Value ptrVal = fir::LoadOp::create(builder, loc, cast);
// Update the base_addr to the value of the Cray pointer.
// This is a hacky way to do the update, and it may harm
// performance around Cray pointer references.
// TODO: we should introduce an operation that updates
// just the base_addr of the given box. The CodeGen
// will just convert it into a single store.
fir::runtime::genPointerAssociateScalar(builder, loc, varDef->getBase(),
ptrVal);
}
return *varDef;
}
llvm::errs() << *symbolRef << "\n";
TODO(getLoc(), "lowering symbol to HLFIR");
}
fir::FortranVariableOpInterface
gen(const Fortran::semantics::Symbol &symbol) {
Fortran::evaluate::SymbolRef symref{symbol};
return gen(symref);
}
fir::FortranVariableOpInterface
gen(const Fortran::evaluate::Component &component) {
if (auto remapped = symMap.lookupComponentOverride(component))
return *remapped;
if (Fortran::semantics::IsAllocatableOrPointer(component.GetLastSymbol()))
return genWholeAllocatableOrPointerComponent(component);
PartInfo partInfo;
mlir::Type resultType = visit(component, partInfo);
return genDesignate(resultType, partInfo, component);
}
fir::FortranVariableOpInterface
gen(const Fortran::evaluate::ArrayRef &arrayRef) {
PartInfo partInfo;
mlir::Type resultType = visit(arrayRef, partInfo);
return genDesignate(resultType, partInfo, arrayRef);
}
fir::FortranVariableOpInterface
gen(const Fortran::evaluate::CoarrayRef &coarrayRef) {
PartInfo partInfo;
mlir::Type resultType = visit(coarrayRef, partInfo);
return genDesignate(resultType, partInfo, coarrayRef);
}
mlir::Type visit(const Fortran::evaluate::CoarrayRef &coarrayRef,
PartInfo &partInfo) {
// Coarray is a data entity with corank > 0 that must be scalar
// or array.
mlir::Type baseType = visit(coarrayRef.base().GetLastSymbol(), partInfo);
if (auto seqType = mlir::dyn_cast<fir::SequenceType>(baseType)) {
fir::FirOpBuilder &builder = getBuilder();
mlir::Location loc = getLoc();
mlir::Type idxTy = builder.getIndexType();
llvm::SmallVector<std::pair<mlir::Value, mlir::Value>> bounds;
auto getBaseBounds = [&](unsigned i) {
if (bounds.empty()) {
bounds = hlfir::genBounds(loc, builder, partInfo.base.value());
assert(!bounds.empty() &&
"failed to compute implicit array section bounds");
}
return bounds[i];
};
auto frontEndResultShape = Fortran::evaluate::GetShape(
converter.getFoldingContext(), coarrayRef);
auto tryGettingExtentFromFrontEnd = [&](unsigned dim)
-> std::pair<mlir::Value, fir::SequenceType::Extent> {
// Use constant extent if possible. The main advantage to do this now
// is to get the best FIR array types as possible while lowering.
if (frontEndResultShape)
if (auto maybeI64 =
Fortran::evaluate::ToInt64(frontEndResultShape->at(dim)))
return {builder.createIntegerConstant(loc, idxTy, *maybeI64),
*maybeI64};
return {mlir::Value{}, fir::SequenceType::getUnknownExtent()};
};
llvm::SmallVector<mlir::Value> resultExtents;
fir::SequenceType::Shape resultTypeShape;
bool sawVectorSubscripts = false;
if (auto *arrayRef{
std::get_if<Fortran::evaluate::ArrayRef>(&coarrayRef.base().u)}) {
for (auto subscript : llvm::enumerate(arrayRef->subscript())) {
if (const auto *triplet = std::get_if<Fortran::evaluate::Triplet>(
&subscript.value().u)) {
mlir::Value lb, ub;
if (const auto &lbExpr = triplet->lower())
lb = genSubscript(*lbExpr);
else
lb = getBaseBounds(subscript.index()).first;
if (const auto &ubExpr = triplet->upper())
ub = genSubscript(*ubExpr);
else
ub = getBaseBounds(subscript.index()).second;
lb = builder.createConvert(loc, idxTy, lb);
ub = builder.createConvert(loc, idxTy, ub);
mlir::Value stride = genSubscript(triplet->stride());
stride = builder.createConvert(loc, idxTy, stride);
auto [extentValue, shapeExtent] =
tryGettingExtentFromFrontEnd(resultExtents.size());
resultTypeShape.push_back(shapeExtent);
if (!extentValue)
extentValue =
builder.genExtentFromTriplet(loc, lb, ub, stride, idxTy);
resultExtents.push_back(extentValue);
partInfo.subscripts.emplace_back(
hlfir::DesignateOp::Triplet{lb, ub, stride});
} else {
const auto &expr =
std::get<Fortran::evaluate::IndirectSubscriptIntegerExpr>(
subscript.value().u)
.value();
hlfir::Entity subscript = genSubscript(expr);
partInfo.subscripts.push_back(subscript);
if (expr.Rank() > 0) {
sawVectorSubscripts = true;
auto [extentValue, shapeExtent] =
tryGettingExtentFromFrontEnd(resultExtents.size());
resultTypeShape.push_back(shapeExtent);
if (!extentValue)
extentValue =
hlfir::genExtent(loc, builder, subscript, /*dim=*/0);
resultExtents.push_back(extentValue);
}
}
}
}
assert(resultExtents.size() == resultTypeShape.size() &&
"inconsistent hlfir.designate shape");
// For vector subscripts, create an hlfir.elemental_addr and continue
// lowering the designator inside it as if it was addressing an element of
// the vector subscripts.
if (sawVectorSubscripts)
TODO(loc, "coarray: with vector subscripts.");
mlir::Type resultType = seqType.getEleTy();
if (!resultTypeShape.empty()) {
// Ranked array section. The result shape comes from the array section
// subscripts.
resultType = fir::SequenceType::get(resultTypeShape, resultType);
assert(!partInfo.resultShape &&
"Fortran designator can only have one ranked part");
partInfo.resultShape = builder.genShape(loc, resultExtents);
}
return resultType;
} else {
return baseType;
}
}
fir::FortranVariableOpInterface
gen(const Fortran::evaluate::ComplexPart &complexPart) {
PartInfo partInfo;
fir::factory::Complex cmplxHelper(getBuilder(), getLoc());
bool complexBit =
complexPart.part() == Fortran::evaluate::ComplexPart::Part::IM;
partInfo.complexPart = {complexBit};
mlir::Type resultType = visit(complexPart.complex(), partInfo);
// Determine complex part type
mlir::Type base = hlfir::getFortranElementType(resultType);
mlir::Type cmplxValueType = cmplxHelper.getComplexPartType(base);
mlir::Type designatorType = changeElementType(resultType, cmplxValueType);
return genDesignate(designatorType, partInfo, complexPart);
}
fir::FortranVariableOpInterface
gen(const Fortran::evaluate::Substring &substring) {
PartInfo partInfo;
mlir::Type baseStringType = Fortran::common::visit(
[&](const auto &x) { return visit(x, partInfo); }, substring.parent());
assert(partInfo.typeParams.size() == 1 && "expect base string length");
// Compute the substring lower and upper bound.
partInfo.substring.push_back(genSubscript(substring.lower()));
if (Fortran::evaluate::MaybeExtentExpr upperBound = substring.upper())
partInfo.substring.push_back(genSubscript(*upperBound));
else
partInfo.substring.push_back(partInfo.typeParams[0]);
fir::FirOpBuilder &builder = getBuilder();
mlir::Location loc = getLoc();
mlir::Type idxTy = builder.getIndexType();
partInfo.substring[0] =
builder.createConvert(loc, idxTy, partInfo.substring[0]);
partInfo.substring[1] =
builder.createConvert(loc, idxTy, partInfo.substring[1]);
// Try using constant length if available. mlir::arith folding would
// most likely be able to fold "max(ub-lb+1,0)" too, but getting
// the constant length in the FIR types would be harder.
std::optional<int64_t> cstLen =
Fortran::evaluate::ToInt64(Fortran::evaluate::Fold(
getConverter().getFoldingContext(), substring.LEN()));
if (cstLen) {
partInfo.typeParams[0] =
builder.createIntegerConstant(loc, idxTy, *cstLen);
} else {
// Compute "len = max(ub-lb+1,0)" (Fortran 2018 9.4.1).
mlir::Value one = builder.createIntegerConstant(loc, idxTy, 1);
auto boundsDiff = mlir::arith::SubIOp::create(
builder, loc, partInfo.substring[1], partInfo.substring[0]);
auto rawLen = mlir::arith::AddIOp::create(builder, loc, boundsDiff, one);
partInfo.typeParams[0] =
fir::factory::genMaxWithZero(builder, loc, rawLen);
}
auto kind = mlir::cast<fir::CharacterType>(
hlfir::getFortranElementType(baseStringType))
.getFKind();
auto newCharTy = fir::CharacterType::get(
baseStringType.getContext(), kind,
cstLen ? *cstLen : fir::CharacterType::unknownLen());
mlir::Type resultType = changeElementType(baseStringType, newCharTy);
return genDesignate(resultType, partInfo, substring);
}
static mlir::Type changeElementType(mlir::Type type, mlir::Type newEleTy) {
return llvm::TypeSwitch<mlir::Type, mlir::Type>(type)
.Case<fir::SequenceType>([&](fir::SequenceType seqTy) -> mlir::Type {
return fir::SequenceType::get(seqTy.getShape(), newEleTy);
})
.Case<fir::ReferenceType, fir::BoxType, fir::ClassType>(
[&](auto t) -> mlir::Type {
using FIRT = decltype(t);
return FIRT::get(changeElementType(t.getEleTy(), newEleTy),
t.isVolatile());
})
.Case<fir::PointerType, fir::HeapType>([&](auto t) -> mlir::Type {
using FIRT = decltype(t);
return FIRT::get(changeElementType(t.getEleTy(), newEleTy));
})
.Default([newEleTy](mlir::Type t) -> mlir::Type { return newEleTy; });
}
fir::FortranVariableOpInterface genWholeAllocatableOrPointerComponent(
const Fortran::evaluate::Component &component) {
if (auto remapped = symMap.lookupComponentOverride(component))
return *remapped;
// Generate whole allocatable or pointer component reference. The
// hlfir.designate result will be a pointer/allocatable.
PartInfo partInfo;
mlir::Type componentType = visitComponentImpl(component, partInfo).second;
const auto isVolatile =
fir::isa_volatile_type(partInfo.base.value().getBase().getType());
mlir::Type designatorType =
fir::ReferenceType::get(componentType, isVolatile);
fir::FortranVariableFlagsAttr attributes =
Fortran::lower::translateSymbolAttributes(getBuilder().getContext(),
component.GetLastSymbol());
return genDesignate(designatorType, partInfo, attributes);
}
mlir::Type visit(const Fortran::evaluate::DataRef &dataRef,
PartInfo &partInfo) {
return Fortran::common::visit(
[&](const auto &x) { return visit(x, partInfo); }, dataRef.u);
}
mlir::Type
visit(const Fortran::evaluate::StaticDataObject::Pointer &staticObject,
PartInfo &partInfo) {
fir::FirOpBuilder &builder = getBuilder();
mlir::Location loc = getLoc();
std::optional<std::string> string = staticObject->AsString();
// TODO: see if StaticDataObject can be replaced by something based on
// Constant<T> to avoid dealing with endianness here for KIND>1.
// This will also avoid making string copies here.
if (!string)
TODO(loc, "StaticDataObject::Pointer substring with kind > 1");
fir::ExtendedValue exv =
fir::factory::createStringLiteral(builder, getLoc(), *string);
auto flags = fir::FortranVariableFlagsAttr::get(
builder.getContext(), fir::FortranVariableFlagsEnum::parameter);
partInfo.base = hlfir::genDeclare(loc, builder, exv, ".stringlit", flags);
partInfo.typeParams.push_back(fir::getLen(exv));
return partInfo.base->getElementOrSequenceType();
}
mlir::Type visit(const Fortran::evaluate::SymbolRef &symbolRef,
PartInfo &partInfo) {
// A symbol is only visited if there is a following array, substring, or
// complex reference. If the entity is a pointer or allocatable, this
// reference designates the target, so the pointer, allocatable must be
// dereferenced here.
partInfo.base =
hlfir::derefPointersAndAllocatables(loc, getBuilder(), gen(symbolRef));
hlfir::genLengthParameters(loc, getBuilder(), *partInfo.base,
partInfo.typeParams);
return partInfo.base->getElementOrSequenceType();
}
mlir::Type visit(const Fortran::evaluate::ArrayRef &arrayRef,
PartInfo &partInfo) {
mlir::Type baseType;
if (const auto *component = arrayRef.base().UnwrapComponent()) {
// Pointers and allocatable components must be dereferenced since the
// array ref designates the target (this is done in "visit"). Other
// components need special care to deal with the array%array_comp(indices)
// case.
if (Fortran::semantics::IsAllocatableOrObjectPointer(
&component->GetLastSymbol()) ||
symMap.lookupComponentOverride(*component))
baseType = visit(*component, partInfo);
else
baseType = hlfir::getFortranElementOrSequenceType(
visitComponentImpl(*component, partInfo).second);
} else {
baseType = visit(arrayRef.base().GetLastSymbol(), partInfo);
}
fir::FirOpBuilder &builder = getBuilder();
mlir::Location loc = getLoc();
mlir::Type idxTy = builder.getIndexType();
llvm::SmallVector<std::pair<mlir::Value, mlir::Value>> bounds;
auto getBaseBounds = [&](unsigned i) {
if (bounds.empty()) {
if (partInfo.componentName.empty()) {
bounds = hlfir::genBounds(loc, builder, partInfo.base.value());
} else {
assert(
partInfo.componentShape &&
"implicit array section bounds must come from component shape");
bounds = hlfir::genBounds(loc, builder, partInfo.componentShape);
}
assert(!bounds.empty() &&
"failed to compute implicit array section bounds");
}
return bounds[i];
};
auto frontEndResultShape =
Fortran::evaluate::GetShape(converter.getFoldingContext(), arrayRef);
auto tryGettingExtentFromFrontEnd =
[&](unsigned dim) -> std::pair<mlir::Value, fir::SequenceType::Extent> {
// Use constant extent if possible. The main advantage to do this now
// is to get the best FIR array types as possible while lowering.
if (frontEndResultShape)
if (auto maybeI64 =
Fortran::evaluate::ToInt64(frontEndResultShape->at(dim)))
return {builder.createIntegerConstant(loc, idxTy, *maybeI64),
*maybeI64};
return {mlir::Value{}, fir::SequenceType::getUnknownExtent()};
};
llvm::SmallVector<mlir::Value> resultExtents;
fir::SequenceType::Shape resultTypeShape;
bool sawVectorSubscripts = false;
for (auto subscript : llvm::enumerate(arrayRef.subscript())) {
if (const auto *triplet =
std::get_if<Fortran::evaluate::Triplet>(&subscript.value().u)) {
mlir::Value lb, ub;
if (const auto &lbExpr = triplet->lower())
lb = genSubscript(*lbExpr);
else
lb = getBaseBounds(subscript.index()).first;
if (const auto &ubExpr = triplet->upper())
ub = genSubscript(*ubExpr);
else
ub = getBaseBounds(subscript.index()).second;
lb = builder.createConvert(loc, idxTy, lb);
ub = builder.createConvert(loc, idxTy, ub);
mlir::Value stride = genSubscript(triplet->stride());
stride = builder.createConvert(loc, idxTy, stride);
auto [extentValue, shapeExtent] =
tryGettingExtentFromFrontEnd(resultExtents.size());
resultTypeShape.push_back(shapeExtent);
if (!extentValue)
extentValue =
builder.genExtentFromTriplet(loc, lb, ub, stride, idxTy);
resultExtents.push_back(extentValue);
partInfo.subscripts.emplace_back(
hlfir::DesignateOp::Triplet{lb, ub, stride});
} else {
const auto &expr =
std::get<Fortran::evaluate::IndirectSubscriptIntegerExpr>(
subscript.value().u)
.value();
hlfir::Entity subscript = genSubscript(expr);
partInfo.subscripts.push_back(subscript);
if (expr.Rank() > 0) {
sawVectorSubscripts = true;
auto [extentValue, shapeExtent] =
tryGettingExtentFromFrontEnd(resultExtents.size());
resultTypeShape.push_back(shapeExtent);
if (!extentValue)
extentValue = hlfir::genExtent(loc, builder, subscript, /*dim=*/0);
resultExtents.push_back(extentValue);
}
}
}
assert(resultExtents.size() == resultTypeShape.size() &&
"inconsistent hlfir.designate shape");
// For vector subscripts, create an hlfir.elemental_addr and continue
// lowering the designator inside it as if it was addressing an element of
// the vector subscripts.
if (sawVectorSubscripts)
return createVectorSubscriptElementAddrOp(partInfo, baseType,
resultExtents);
mlir::Type resultType =
mlir::cast<fir::SequenceType>(baseType).getElementType();
if (!resultTypeShape.empty()) {
// Ranked array section. The result shape comes from the array section
// subscripts.
resultType = fir::SequenceType::get(resultTypeShape, resultType);
assert(!partInfo.resultShape &&
"Fortran designator can only have one ranked part");
partInfo.resultShape = builder.genShape(loc, resultExtents);
} else if (!partInfo.componentName.empty() &&
partInfo.base.value().isArray()) {
// This is an array%array_comp(indices) reference. Keep the
// shape of the base array and not the array_comp.
auto compBaseTy = partInfo.base->getElementOrSequenceType();
resultType = changeElementType(compBaseTy, resultType);
assert(!partInfo.resultShape && "should not have been computed already");
partInfo.resultShape = hlfir::genShape(loc, builder, *partInfo.base);
}
return resultType;
}
static bool
mayHaveNonDefaultLowerBounds(const Fortran::semantics::Symbol &componentSym) {
if (const auto *objDetails =
componentSym.detailsIf<Fortran::semantics::ObjectEntityDetails>())
for (const Fortran::semantics::ShapeSpec &bounds : objDetails->shape())
if (auto lb = bounds.lbound().GetExplicit())
if (auto constant = Fortran::evaluate::ToInt64(*lb))
if (!constant || *constant != 1)
return true;
return false;
}
static bool mayHaveNonDefaultLowerBounds(const PartInfo &partInfo) {
return partInfo.resultShape &&
mlir::isa<fir::ShiftType, fir::ShapeShiftType>(
partInfo.resultShape.getType());
}
mlir::Type visit(const Fortran::evaluate::Component &component,
PartInfo &partInfo) {
if (Fortran::semantics::IsAllocatableOrPointer(component.GetLastSymbol())) {
// In a visit, the following reference will address the target. Insert
// the dereference here.
partInfo.base = genWholeAllocatableOrPointerComponent(component);
partInfo.base = hlfir::derefPointersAndAllocatables(loc, getBuilder(),
*partInfo.base);
hlfir::genLengthParameters(loc, getBuilder(), *partInfo.base,
partInfo.typeParams);
return partInfo.base->getElementOrSequenceType();
}
if (auto remapped = symMap.lookupComponentOverride(component)) {
// Do not generate field for the designate if the component
// is overridden, the override value is already addressing
// the component.
partInfo.base = *remapped;
hlfir::genLengthParameters(loc, getBuilder(), *partInfo.base,
partInfo.typeParams);
return partInfo.base->getElementOrSequenceType();
}
// This function must be called from contexts where the component is not the
// base of an ArrayRef. In these cases, the component cannot be an array
// if the base is an array. The code below determines the shape of the
// component reference if any.
auto [baseType, componentType] = visitComponentImpl(component, partInfo);
mlir::Type componentBaseType =
hlfir::getFortranElementOrSequenceType(componentType);
if (partInfo.base.value().isArray()) {
// For array%scalar_comp, the result shape is
// the one of the base. Compute it here. Note that the lower bounds of the
// base are not the ones of the resulting reference (that are default
// ones).
partInfo.resultShape = hlfir::genShape(loc, getBuilder(), *partInfo.base);
assert(!partInfo.componentShape &&
"Fortran designators can only have one ranked part");
return changeElementType(baseType, componentBaseType);
}
if (partInfo.complexPart && partInfo.componentShape) {
// Treat ...array_comp%im/re as ...array_comp(:,:,...)%im/re
// so that the codegen has the full slice triples for the component
// readily available.
fir::FirOpBuilder &builder = getBuilder();
mlir::Type idxTy = builder.getIndexType();
mlir::Value one = builder.createIntegerConstant(loc, idxTy, 1);
llvm::SmallVector<mlir::Value> resultExtents;
// Collect <lb, ub> pairs from the component shape.
auto bounds = hlfir::genBounds(loc, builder, partInfo.componentShape);
for (auto &boundPair : bounds) {
// The default subscripts are <lb, ub, 1>:
partInfo.subscripts.emplace_back(hlfir::DesignateOp::Triplet{
boundPair.first, boundPair.second, one});
auto extentValue = builder.genExtentFromTriplet(
loc, boundPair.first, boundPair.second, one, idxTy);
resultExtents.push_back(extentValue);
}
// The result shape is: <max((ub - lb + 1) / 1, 0), ...>.
partInfo.resultShape = builder.genShape(loc, resultExtents);
return componentBaseType;
}
// scalar%array_comp or scalar%scalar. In any case the shape of this
// part-ref is coming from the component.
partInfo.resultShape = partInfo.componentShape;
partInfo.componentShape = {};
return componentBaseType;
}
// Returns the <BaseType, ComponentType> pair, computes partInfo.base,
// partInfo.componentShape and partInfo.typeParams, but does not set the
// partInfo.resultShape yet. The result shape will be computed after
// processing a following ArrayRef, if any, and in "visit" otherwise.
std::pair<mlir::Type, mlir::Type>
visitComponentImpl(const Fortran::evaluate::Component &component,
PartInfo &partInfo) {
fir::FirOpBuilder &builder = getBuilder();
// Break the Designator visit here: if the base is an array-ref, a
// coarray-ref, or another component, this creates another hlfir.designate
// for it. hlfir.designate is not meant to represent more than one
// part-ref.
partInfo.base = gen(component.base());
// If the base is an allocatable/pointer, dereference it here since the
// component ref designates its target.
partInfo.base =
hlfir::derefPointersAndAllocatables(loc, builder, *partInfo.base);
assert(partInfo.typeParams.empty() && "should not have been computed yet");
hlfir::genLengthParameters(getLoc(), getBuilder(), *partInfo.base,
partInfo.typeParams);
mlir::Type baseType = partInfo.base->getElementOrSequenceType();
// Lower the information about the component (type, length parameters and
// shape).
const Fortran::semantics::Symbol &componentSym = component.GetLastSymbol();
partInfo.componentName = converter.getRecordTypeFieldName(componentSym);
auto recordType =
mlir::cast<fir::RecordType>(hlfir::getFortranElementType(baseType));
if (recordType.isDependentType())
TODO(getLoc(), "Designate derived type with length parameters in HLFIR");
mlir::Type fieldType = recordType.getType(partInfo.componentName);
assert(fieldType && "component name is not known");
mlir::Type fieldBaseType =
hlfir::getFortranElementOrSequenceType(fieldType);
partInfo.componentShape = genComponentShape(componentSym, fieldBaseType);
mlir::Type fieldEleType = hlfir::getFortranElementType(fieldBaseType);
if (fir::isRecordWithTypeParameters(fieldEleType))
TODO(loc,
"lower a component that is a parameterized derived type to HLFIR");
if (auto charTy = mlir::dyn_cast<fir::CharacterType>(fieldEleType)) {
mlir::Location loc = getLoc();
mlir::Type idxTy = builder.getIndexType();
if (charTy.hasConstantLen())
partInfo.typeParams.push_back(
builder.createIntegerConstant(loc, idxTy, charTy.getLen()));
else if (!Fortran::semantics::IsAllocatableOrObjectPointer(&componentSym))
TODO(loc, "compute character length of automatic character component "
"in a PDT");
// Otherwise, the length of the component is deferred and will only
// be read when the component is dereferenced.
}
return {baseType, fieldType};
}
// Compute: "lb + (i-1)*step".
mlir::Value computeTripletPosition(mlir::Location loc,
fir::FirOpBuilder &builder,
hlfir::DesignateOp::Triplet &triplet,
mlir::Value oneBasedIndex) {
mlir::Type idxTy = builder.getIndexType();
mlir::Value lb = builder.createConvert(loc, idxTy, std::get<0>(triplet));
mlir::Value step = builder.createConvert(loc, idxTy, std::get<2>(triplet));
mlir::Value one = builder.createIntegerConstant(loc, idxTy, 1);
oneBasedIndex = builder.createConvert(loc, idxTy, oneBasedIndex);
mlir::Value zeroBased =
mlir::arith::SubIOp::create(builder, loc, oneBasedIndex, one);
mlir::Value offset =
mlir::arith::MulIOp::create(builder, loc, zeroBased, step);
return mlir::arith::AddIOp::create(builder, loc, lb, offset);
}
/// Create an hlfir.element_addr operation to deal with vector subscripted
/// entities. This transforms the current vector subscripted array-ref into a
/// a scalar array-ref that is addressing the vector subscripted part given
/// the one based indices of the hlfir.element_addr.
/// The rest of the designator lowering will continue lowering any further
/// parts inside the hlfir.elemental as a scalar reference.
/// At the end of the designator lowering, the hlfir.elemental_addr will
/// be turned into an hlfir.elemental value, unless the caller of this
/// utility requested to get the hlfir.elemental_addr instead of lowering
/// the designator to an mlir::Value.
mlir::Type createVectorSubscriptElementAddrOp(
PartInfo &partInfo, mlir::Type baseType,
llvm::ArrayRef<mlir::Value> resultExtents) {
fir::FirOpBuilder &builder = getBuilder();
mlir::Value shape = builder.genShape(loc, resultExtents);
// The type parameters to be added on the hlfir.elemental_addr are the ones
// of the whole designator (not the ones of the vector subscripted part).
// These are not yet known and will be added when finalizing the designator
// lowering.
// The resulting designator may be polymorphic, in which case the resulting
// type is the base of the vector subscripted part because
// allocatable/pointer components cannot be referenced after a vector
// subscripted part. Set the mold to the current base. It will be erased if
// the resulting designator is not polymorphic.
assert(partInfo.base.has_value() &&
"vector subscripted part must have a base");
mlir::Value mold = *partInfo.base;
auto elementalAddrOp = hlfir::ElementalAddrOp::create(
builder, loc, shape, mold, mlir::ValueRange{},
/*isUnordered=*/true);
setVectorSubscriptElementAddrOp(elementalAddrOp);
builder.setInsertionPointToEnd(&elementalAddrOp.getBody().front());
mlir::Region::BlockArgListType indices = elementalAddrOp.getIndices();
auto indicesIterator = indices.begin();
auto getNextOneBasedIndex = [&]() -> mlir::Value {
assert(indicesIterator != indices.end() && "ill formed ElementalAddrOp");
return *(indicesIterator++);
};
// Transform the designator into a scalar designator computing the vector
// subscripted entity element address given one based indices (for the shape
// of the vector subscripted designator).
for (hlfir::DesignateOp::Subscript &subscript : partInfo.subscripts) {
if (auto *triplet =
std::get_if<hlfir::DesignateOp::Triplet>(&subscript)) {
// subscript = (lb + (i-1)*step)
mlir::Value scalarSubscript = computeTripletPosition(
loc, builder, *triplet, getNextOneBasedIndex());
subscript = scalarSubscript;
} else {
hlfir::Entity valueSubscript{std::get<mlir::Value>(subscript)};
if (valueSubscript.isScalar())
continue;
// subscript = vector(i + (vector_lb-1))
hlfir::Entity scalarSubscript = hlfir::getElementAt(
loc, builder, valueSubscript, {getNextOneBasedIndex()});
scalarSubscript =
hlfir::loadTrivialScalar(loc, builder, scalarSubscript);
subscript = scalarSubscript;
}
}
builder.setInsertionPoint(elementalAddrOp);
return mlir::cast<fir::SequenceType>(baseType).getElementType();
}
/// Yield the designator for the final part-ref inside the
/// hlfir.elemental_addr.
void finalizeElementAddrOp(hlfir::ElementalAddrOp elementalAddrOp,
hlfir::EntityWithAttributes elementAddr) {
fir::FirOpBuilder &builder = getBuilder();
builder.setInsertionPointToEnd(&elementalAddrOp.getBody().front());
if (!elementAddr.isPolymorphic())
elementalAddrOp.getMoldMutable().clear();
hlfir::YieldOp::create(builder, loc, elementAddr);
builder.setInsertionPointAfter(elementalAddrOp);
}
/// If the lowered designator has vector subscripts turn it into an
/// ElementalOp, otherwise, return the lowered designator. This should
/// only be called if the user did not request to get the
/// hlfir.elemental_addr. In Fortran, vector subscripted designators are only
/// writable on the left-hand side of an assignment and in input IO
/// statements. Otherwise, they are not variables (cannot be modified, their
/// value is taken at the place they appear).
hlfir::EntityWithAttributes turnVectorSubscriptedDesignatorIntoValue(
hlfir::EntityWithAttributes loweredDesignator) {
std::optional<hlfir::ElementalAddrOp> elementalAddrOp =
getVectorSubscriptElementAddrOp();
if (!elementalAddrOp)
return loweredDesignator;
finalizeElementAddrOp(*elementalAddrOp, loweredDesignator);
// This vector subscript designator is only being read, transform the
// hlfir.elemental_addr into an hlfir.elemental. The content of the
// hlfir.elemental_addr is cloned, and the resulting address is loaded to
// get the new element value.
fir::FirOpBuilder &builder = getBuilder();
mlir::Location loc = getLoc();
mlir::Value elemental =
hlfir::cloneToElementalOp(loc, builder, *elementalAddrOp);
(*elementalAddrOp)->erase();
setVectorSubscriptElementAddrOp(std::nullopt);
fir::FirOpBuilder *bldr = &builder;
getStmtCtx().attachCleanup(
[=]() { hlfir::DestroyOp::create(*bldr, loc, elemental); });
return hlfir::EntityWithAttributes{elemental};
}
/// Lower a subscript expression. If it is a scalar subscript that is a
/// variable, it is loaded into an integer value. If it is an array (for
/// vector subscripts) it is dereferenced if this is an allocatable or
/// pointer.
template <typename T>
hlfir::Entity genSubscript(const Fortran::evaluate::Expr<T> &expr);
const std::optional<hlfir::ElementalAddrOp> &
getVectorSubscriptElementAddrOp() const {
return vectorSubscriptElementAddrOp;
}
void setVectorSubscriptElementAddrOp(
std::optional<hlfir::ElementalAddrOp> elementalAddrOp) {
vectorSubscriptElementAddrOp = elementalAddrOp;
}
mlir::Location getLoc() const { return loc; }
Fortran::lower::AbstractConverter &getConverter() { return converter; }
fir::FirOpBuilder &getBuilder() { return converter.getFirOpBuilder(); }
Fortran::lower::SymMap &getSymMap() { return symMap; }
Fortran::lower::StatementContext &getStmtCtx() { return stmtCtx; }
Fortran::lower::AbstractConverter &converter;
Fortran::lower::SymMap &symMap;
Fortran::lower::StatementContext &stmtCtx;
// If there is a vector subscript, an elementalAddrOp is created
// to compute the address of the designator elements.
std::optional<hlfir::ElementalAddrOp> vectorSubscriptElementAddrOp{};
mlir::Location loc;
};
hlfir::EntityWithAttributes HlfirDesignatorBuilder::genDesignatorExpr(
const Fortran::lower::SomeExpr &designatorExpr,
bool vectorSubscriptDesignatorToValue) {
// Expr<SomeType> plumbing to unwrap Designator<T> and call
// gen(Designator<T>.u).
return Fortran::common::visit(
[&](const auto &x) -> hlfir::EntityWithAttributes {
using T = std::decay_t<decltype(x)>;
if constexpr (Fortran::common::HasMember<
T, Fortran::lower::CategoryExpression>) {
if constexpr (T::Result::category ==
Fortran::common::TypeCategory::Derived) {
return gen(std::get<Fortran::evaluate::Designator<
Fortran::evaluate::SomeDerived>>(x.u)
.u,
vectorSubscriptDesignatorToValue);
} else {
return Fortran::common::visit(
[&](const auto &preciseKind) {
using TK =
typename std::decay_t<decltype(preciseKind)>::Result;
return gen(
std::get<Fortran::evaluate::Designator<TK>>(preciseKind.u)
.u,
vectorSubscriptDesignatorToValue);
},
x.u);
}
} else {
fir::emitFatalError(loc, "unexpected typeless Designator");
}
},
designatorExpr.u);
}
hlfir::ElementalAddrOp
HlfirDesignatorBuilder::convertVectorSubscriptedExprToElementalAddr(
const Fortran::lower::SomeExpr &designatorExpr) {
hlfir::EntityWithAttributes elementAddrEntity = genDesignatorExpr(
designatorExpr, /*vectorSubscriptDesignatorToValue=*/false);
assert(getVectorSubscriptElementAddrOp().has_value() &&
"expected vector subscripts");
hlfir::ElementalAddrOp elementalAddrOp = *getVectorSubscriptElementAddrOp();
// Now that the type parameters have been computed, add then to the
// hlfir.elemental_addr.
fir::FirOpBuilder &builder = getBuilder();
llvm::SmallVector<mlir::Value, 1> lengths;
hlfir::genLengthParameters(loc, builder, elementAddrEntity, lengths);
if (!lengths.empty())
elementalAddrOp.getTypeparamsMutable().assign(lengths);
if (!elementAddrEntity.isPolymorphic())
elementalAddrOp.getMoldMutable().clear();
// Create the hlfir.yield terminator inside the hlfir.elemental_body.
builder.setInsertionPointToEnd(&elementalAddrOp.getBody().front());
hlfir::YieldOp::create(builder, loc, elementAddrEntity);
builder.setInsertionPointAfter(elementalAddrOp);
// Reset the HlfirDesignatorBuilder state, in case it is used on a new
// designator.
setVectorSubscriptElementAddrOp(std::nullopt);
return elementalAddrOp;
}
//===--------------------------------------------------------------------===//
// Binary Operation implementation
//===--------------------------------------------------------------------===//
template <typename T>
struct BinaryOp {};
#undef GENBIN
#define GENBIN(GenBinEvOp, GenBinTyCat, GenBinFirOp) \
template <int KIND> \
struct BinaryOp<Fortran::evaluate::GenBinEvOp<Fortran::evaluate::Type< \
Fortran::common::TypeCategory::GenBinTyCat, KIND>>> { \
using Op = Fortran::evaluate::GenBinEvOp<Fortran::evaluate::Type< \
Fortran::common::TypeCategory::GenBinTyCat, KIND>>; \
static hlfir::EntityWithAttributes gen(mlir::Location loc, \
fir::FirOpBuilder &builder, \
const Op &, hlfir::Entity lhs, \
hlfir::Entity rhs) { \
if constexpr (Fortran::common::TypeCategory::GenBinTyCat == \
Fortran::common::TypeCategory::Unsigned) { \
return hlfir::EntityWithAttributes{ \
builder.createUnsigned<GenBinFirOp>(loc, lhs.getType(), lhs, \
rhs)}; \
} else { \
return hlfir::EntityWithAttributes{ \
GenBinFirOp::create(builder, loc, lhs, rhs)}; \
} \
} \
};
GENBIN(Add, Integer, mlir::arith::AddIOp)
GENBIN(Add, Unsigned, mlir::arith::AddIOp)
GENBIN(Add, Real, mlir::arith::AddFOp)
GENBIN(Add, Complex, fir::AddcOp)
GENBIN(Subtract, Integer, mlir::arith::SubIOp)
GENBIN(Subtract, Unsigned, mlir::arith::SubIOp)
GENBIN(Subtract, Real, mlir::arith::SubFOp)
GENBIN(Subtract, Complex, fir::SubcOp)
GENBIN(Multiply, Integer, mlir::arith::MulIOp)
GENBIN(Multiply, Unsigned, mlir::arith::MulIOp)
GENBIN(Multiply, Real, mlir::arith::MulFOp)
GENBIN(Multiply, Complex, fir::MulcOp)
GENBIN(Divide, Integer, mlir::arith::DivSIOp)
GENBIN(Divide, Unsigned, mlir::arith::DivUIOp)
GENBIN(Divide, Real, mlir::arith::DivFOp)
template <int KIND>
struct BinaryOp<Fortran::evaluate::Divide<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Complex, KIND>>> {
using Op = Fortran::evaluate::Divide<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Complex, KIND>>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder, const Op &,
hlfir::Entity lhs, hlfir::Entity rhs) {
mlir::Type ty = Fortran::lower::getFIRType(
builder.getContext(), Fortran::common::TypeCategory::Complex, KIND,
/*params=*/{});
// TODO: Ideally, complex number division operations should always be
// lowered to MLIR. However, converting them to the runtime via MLIR causes
// ABI issues.
if (builder.getComplexDivisionToRuntimeFlag()) {
return hlfir::EntityWithAttributes{
fir::genDivC(builder, loc, ty, lhs, rhs)};
} else {
return hlfir::EntityWithAttributes{
mlir::complex::DivOp::create(builder, loc, lhs, rhs)};
}
}
};
template <Fortran::common::TypeCategory TC, int KIND>
struct BinaryOp<Fortran::evaluate::Power<Fortran::evaluate::Type<TC, KIND>>> {
using Op = Fortran::evaluate::Power<Fortran::evaluate::Type<TC, KIND>>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder, const Op &,
hlfir::Entity lhs, hlfir::Entity rhs) {
mlir::Type ty = Fortran::lower::getFIRType(builder.getContext(), TC, KIND,
/*params=*/{});
return hlfir::EntityWithAttributes{fir::genPow(builder, loc, ty, lhs, rhs)};
}
};
template <Fortran::common::TypeCategory TC, int KIND>
struct BinaryOp<
Fortran::evaluate::RealToIntPower<Fortran::evaluate::Type<TC, KIND>>> {
using Op =
Fortran::evaluate::RealToIntPower<Fortran::evaluate::Type<TC, KIND>>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder, const Op &,
hlfir::Entity lhs, hlfir::Entity rhs) {
mlir::Type ty = Fortran::lower::getFIRType(builder.getContext(), TC, KIND,
/*params=*/{});
return hlfir::EntityWithAttributes{fir::genPow(builder, loc, ty, lhs, rhs)};
}
};
template <Fortran::common::TypeCategory TC, int KIND>
struct BinaryOp<
Fortran::evaluate::Extremum<Fortran::evaluate::Type<TC, KIND>>> {
using Op = Fortran::evaluate::Extremum<Fortran::evaluate::Type<TC, KIND>>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder,
const Op &op, hlfir::Entity lhs,
hlfir::Entity rhs) {
llvm::SmallVector<mlir::Value, 2> args{lhs, rhs};
fir::ExtendedValue res = op.ordering == Fortran::evaluate::Ordering::Greater
? fir::genMax(builder, loc, args)
: fir::genMin(builder, loc, args);
return hlfir::EntityWithAttributes{fir::getBase(res)};
}
};
// evaluate::Extremum is only created by the front-end when building compiler
// generated expressions (like when folding LEN() or shape/bounds inquiries).
// MIN and MAX are represented as evaluate::ProcedureRef and are not going
// through here. So far the frontend does not generate character Extremum so
// there is no way to test it.
template <int KIND>
struct BinaryOp<Fortran::evaluate::Extremum<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Character, KIND>>> {
using Op = Fortran::evaluate::Extremum<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Character, KIND>>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &, const Op &,
hlfir::Entity, hlfir::Entity) {
fir::emitFatalError(loc, "Fortran::evaluate::Extremum are unexpected");
}
static void genResultTypeParams(mlir::Location loc, fir::FirOpBuilder &,
hlfir::Entity, hlfir::Entity,
llvm::SmallVectorImpl<mlir::Value> &) {
fir::emitFatalError(loc, "Fortran::evaluate::Extremum are unexpected");
}
};
/// Convert parser's INTEGER relational operators to MLIR.
static mlir::arith::CmpIPredicate
translateSignedRelational(Fortran::common::RelationalOperator rop) {
switch (rop) {
case Fortran::common::RelationalOperator::LT:
return mlir::arith::CmpIPredicate::slt;
case Fortran::common::RelationalOperator::LE:
return mlir::arith::CmpIPredicate::sle;
case Fortran::common::RelationalOperator::EQ:
return mlir::arith::CmpIPredicate::eq;
case Fortran::common::RelationalOperator::NE:
return mlir::arith::CmpIPredicate::ne;
case Fortran::common::RelationalOperator::GT:
return mlir::arith::CmpIPredicate::sgt;
case Fortran::common::RelationalOperator::GE:
return mlir::arith::CmpIPredicate::sge;
}
llvm_unreachable("unhandled INTEGER relational operator");
}
template <int KIND>
struct BinaryOp<Fortran::evaluate::Relational<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Integer, KIND>>> {
using Op = Fortran::evaluate::Relational<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Integer, KIND>>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder,
const Op &op, hlfir::Entity lhs,
hlfir::Entity rhs) {
auto cmp = mlir::arith::CmpIOp::create(
builder, loc, Fortran::lower::translateSignedRelational(op.opr), lhs,
rhs);
return hlfir::EntityWithAttributes{cmp};
}
};
template <int KIND>
struct BinaryOp<Fortran::evaluate::Relational<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Unsigned, KIND>>> {
using Op = Fortran::evaluate::Relational<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Unsigned, KIND>>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder,
const Op &op, hlfir::Entity lhs,
hlfir::Entity rhs) {
int bits = Fortran::evaluate::Type<Fortran::common::TypeCategory::Integer,
KIND>::Scalar::bits;
auto signlessType = mlir::IntegerType::get(
builder.getContext(), bits,
mlir::IntegerType::SignednessSemantics::Signless);
mlir::Value lhsSL = builder.createConvert(loc, signlessType, lhs);
mlir::Value rhsSL = builder.createConvert(loc, signlessType, rhs);
auto cmp = mlir::arith::CmpIOp::create(
builder, loc, Fortran::lower::translateUnsignedRelational(op.opr),
lhsSL, rhsSL);
return hlfir::EntityWithAttributes{cmp};
}
};
template <int KIND>
struct BinaryOp<Fortran::evaluate::Relational<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Real, KIND>>> {
using Op = Fortran::evaluate::Relational<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Real, KIND>>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder,
const Op &op, hlfir::Entity lhs,
hlfir::Entity rhs) {
auto cmp = mlir::arith::CmpFOp::create(
builder, loc, Fortran::lower::translateFloatRelational(op.opr), lhs,
rhs);
return hlfir::EntityWithAttributes{cmp};
}
};
template <int KIND>
struct BinaryOp<Fortran::evaluate::Relational<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Complex, KIND>>> {
using Op = Fortran::evaluate::Relational<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Complex, KIND>>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder,
const Op &op, hlfir::Entity lhs,
hlfir::Entity rhs) {
auto cmp = fir::CmpcOp::create(
builder, loc, Fortran::lower::translateFloatRelational(op.opr), lhs,
rhs);
return hlfir::EntityWithAttributes{cmp};
}
};
template <int KIND>
struct BinaryOp<Fortran::evaluate::Relational<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Character, KIND>>> {
using Op = Fortran::evaluate::Relational<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Character, KIND>>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder,
const Op &op, hlfir::Entity lhs,
hlfir::Entity rhs) {
auto cmp = hlfir::CmpCharOp::create(
builder, loc, translateSignedRelational(op.opr), lhs, rhs);
return hlfir::EntityWithAttributes{cmp};
}
};
template <int KIND>
struct BinaryOp<Fortran::evaluate::LogicalOperation<KIND>> {
using Op = Fortran::evaluate::LogicalOperation<KIND>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder,
const Op &op, hlfir::Entity lhs,
hlfir::Entity rhs) {
bool lhsIsLogical = mlir::isa<fir::LogicalType>(lhs.getType());
bool rhsIsLogical = mlir::isa<fir::LogicalType>(rhs.getType());
if (lhsIsLogical || rhsIsLogical) {
// Use fir logical ops when at least one operand is a Fortran LOGICAL.
// Ensure both operands have the same type.
mlir::Type resTy = lhsIsLogical ? lhs.getType() : rhs.getType();
mlir::Value lhsVal = builder.createConvert(loc, resTy, lhs);
mlir::Value rhsVal = builder.createConvert(loc, resTy, rhs);
switch (op.logicalOperator) {
case Fortran::evaluate::LogicalOperator::And:
return hlfir::EntityWithAttributes{
fir::LogicalAndOp::create(builder, loc, resTy, lhsVal, rhsVal)};
case Fortran::evaluate::LogicalOperator::Or:
return hlfir::EntityWithAttributes{
fir::LogicalOrOp::create(builder, loc, resTy, lhsVal, rhsVal)};
case Fortran::evaluate::LogicalOperator::Eqv:
return hlfir::EntityWithAttributes{
fir::EqvOp::create(builder, loc, resTy, lhsVal, rhsVal)};
case Fortran::evaluate::LogicalOperator::Neqv:
return hlfir::EntityWithAttributes{
fir::NeqvOp::create(builder, loc, resTy, lhsVal, rhsVal)};
case Fortran::evaluate::LogicalOperator::Not:
llvm_unreachable(".NOT. is not a binary operator");
}
llvm_unreachable("unhandled logical operation");
}
// Both operands are i1 (from arithmetic comparisons): use arith ops.
mlir::Type i1Type = builder.getI1Type();
mlir::Value i1Lhs = builder.createConvert(loc, i1Type, lhs);
mlir::Value i1Rhs = builder.createConvert(loc, i1Type, rhs);
switch (op.logicalOperator) {
case Fortran::evaluate::LogicalOperator::And:
return hlfir::EntityWithAttributes{
mlir::arith::AndIOp::create(builder, loc, i1Lhs, i1Rhs)};
case Fortran::evaluate::LogicalOperator::Or:
return hlfir::EntityWithAttributes{
mlir::arith::OrIOp::create(builder, loc, i1Lhs, i1Rhs)};
case Fortran::evaluate::LogicalOperator::Eqv:
return hlfir::EntityWithAttributes{mlir::arith::CmpIOp::create(
builder, loc, mlir::arith::CmpIPredicate::eq, i1Lhs, i1Rhs)};
case Fortran::evaluate::LogicalOperator::Neqv:
return hlfir::EntityWithAttributes{mlir::arith::CmpIOp::create(
builder, loc, mlir::arith::CmpIPredicate::ne, i1Lhs, i1Rhs)};
case Fortran::evaluate::LogicalOperator::Not:
llvm_unreachable(".NOT. is not a binary operator");
}
llvm_unreachable("unhandled logical operation");
}
};
template <int KIND>
struct BinaryOp<Fortran::evaluate::ComplexConstructor<KIND>> {
using Op = Fortran::evaluate::ComplexConstructor<KIND>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder, const Op &,
hlfir::Entity lhs, hlfir::Entity rhs) {
mlir::Value res =
fir::factory::Complex{builder, loc}.createComplex(lhs, rhs);
return hlfir::EntityWithAttributes{res};
}
};
template <int KIND>
struct BinaryOp<Fortran::evaluate::SetLength<KIND>> {
using Op = Fortran::evaluate::SetLength<KIND>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder, const Op &,
hlfir::Entity string,
hlfir::Entity length) {
// The input length may be a user input and needs to be sanitized as per
// Fortran 2018 7.4.4.2 point 5.
mlir::Value safeLength = fir::factory::genMaxWithZero(builder, loc, length);
return hlfir::EntityWithAttributes{
hlfir::SetLengthOp::create(builder, loc, string, safeLength)};
}
static void
genResultTypeParams(mlir::Location, fir::FirOpBuilder &, hlfir::Entity,
hlfir::Entity rhs,
llvm::SmallVectorImpl<mlir::Value> &resultTypeParams) {
resultTypeParams.push_back(rhs);
}
};
template <int KIND>
struct BinaryOp<Fortran::evaluate::Concat<KIND>> {
using Op = Fortran::evaluate::Concat<KIND>;
hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder, const Op &,
hlfir::Entity lhs, hlfir::Entity rhs) {
assert(len && "genResultTypeParams must have been called");
auto concat =
hlfir::ConcatOp::create(builder, loc, mlir::ValueRange{lhs, rhs}, len);
return hlfir::EntityWithAttributes{concat.getResult()};
}
void
genResultTypeParams(mlir::Location loc, fir::FirOpBuilder &builder,
hlfir::Entity lhs, hlfir::Entity rhs,
llvm::SmallVectorImpl<mlir::Value> &resultTypeParams) {
llvm::SmallVector<mlir::Value> lengths;
hlfir::genLengthParameters(loc, builder, lhs, lengths);
hlfir::genLengthParameters(loc, builder, rhs, lengths);
assert(lengths.size() == 2 && "lacks rhs or lhs length");
mlir::Type idxType = builder.getIndexType();
mlir::Value lhsLen = builder.createConvert(loc, idxType, lengths[0]);
mlir::Value rhsLen = builder.createConvert(loc, idxType, lengths[1]);
len = mlir::arith::AddIOp::create(builder, loc, lhsLen, rhsLen);
resultTypeParams.push_back(len);
}
private:
mlir::Value len{};
};
//===--------------------------------------------------------------------===//
// Unary Operation implementation
//===--------------------------------------------------------------------===//
template <typename T>
struct UnaryOp {};
template <int KIND>
struct UnaryOp<Fortran::evaluate::Not<KIND>> {
using Op = Fortran::evaluate::Not<KIND>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder, const Op &,
hlfir::Entity lhs) {
mlir::Value one = builder.createBool(loc, true);
mlir::Value val = builder.createConvert(loc, builder.getI1Type(), lhs);
return hlfir::EntityWithAttributes{
mlir::arith::XOrIOp::create(builder, loc, val, one)};
}
};
template <int KIND>
struct UnaryOp<Fortran::evaluate::Negate<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Integer, KIND>>> {
using Op = Fortran::evaluate::Negate<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Integer, KIND>>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder, const Op &,
hlfir::Entity lhs) {
// Like LLVM, integer negation is the binary op "0 - value"
mlir::Type type = Fortran::lower::getFIRType(
builder.getContext(), Fortran::common::TypeCategory::Integer, KIND,
/*params=*/{});
mlir::Value zero = builder.createIntegerConstant(loc, type, 0);
return hlfir::EntityWithAttributes{
mlir::arith::SubIOp::create(builder, loc, zero, lhs)};
}
};
template <int KIND>
struct UnaryOp<Fortran::evaluate::Negate<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Unsigned, KIND>>> {
using Op = Fortran::evaluate::Negate<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Unsigned, KIND>>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder, const Op &,
hlfir::Entity lhs) {
int bits = Fortran::evaluate::Type<Fortran::common::TypeCategory::Integer,
KIND>::Scalar::bits;
mlir::Type signlessType = mlir::IntegerType::get(
builder.getContext(), bits,
mlir::IntegerType::SignednessSemantics::Signless);
mlir::Value zero = builder.createIntegerConstant(loc, signlessType, 0);
mlir::Value signless = builder.createConvert(loc, signlessType, lhs);
mlir::Value negated =
mlir::arith::SubIOp::create(builder, loc, zero, signless);
return hlfir::EntityWithAttributes(
builder.createConvert(loc, lhs.getType(), negated));
}
};
template <int KIND>
struct UnaryOp<Fortran::evaluate::Negate<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Real, KIND>>> {
using Op = Fortran::evaluate::Negate<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Real, KIND>>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder, const Op &,
hlfir::Entity lhs) {
return hlfir::EntityWithAttributes{
mlir::arith::NegFOp::create(builder, loc, lhs)};
}
};
template <int KIND>
struct UnaryOp<Fortran::evaluate::Negate<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Complex, KIND>>> {
using Op = Fortran::evaluate::Negate<
Fortran::evaluate::Type<Fortran::common::TypeCategory::Complex, KIND>>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder, const Op &,
hlfir::Entity lhs) {
return hlfir::EntityWithAttributes{fir::NegcOp::create(builder, loc, lhs)};
}
};
template <int KIND>
struct UnaryOp<Fortran::evaluate::ComplexComponent<KIND>> {
using Op = Fortran::evaluate::ComplexComponent<KIND>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder,
const Op &op, hlfir::Entity lhs) {
mlir::Value res = fir::factory::Complex{builder, loc}.extractComplexPart(
lhs, op.isImaginaryPart);
return hlfir::EntityWithAttributes{res};
}
};
template <typename T>
struct UnaryOp<Fortran::evaluate::Parentheses<T>> {
using Op = Fortran::evaluate::Parentheses<T>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder,
const Op &op, hlfir::Entity lhs) {
if (lhs.isVariable())
return hlfir::EntityWithAttributes{
hlfir::AsExprOp::create(builder, loc, lhs)};
return hlfir::EntityWithAttributes{
hlfir::NoReassocOp::create(builder, loc, lhs.getType(), lhs)};
}
static void
genResultTypeParams(mlir::Location loc, fir::FirOpBuilder &builder,
hlfir::Entity lhs,
llvm::SmallVectorImpl<mlir::Value> &resultTypeParams) {
hlfir::genLengthParameters(loc, builder, lhs, resultTypeParams);
}
};
template <Fortran::common::TypeCategory TC1, int KIND,
Fortran::common::TypeCategory TC2>
struct UnaryOp<
Fortran::evaluate::Convert<Fortran::evaluate::Type<TC1, KIND>, TC2>> {
using Op =
Fortran::evaluate::Convert<Fortran::evaluate::Type<TC1, KIND>, TC2>;
static hlfir::EntityWithAttributes gen(mlir::Location loc,
fir::FirOpBuilder &builder, const Op &,
hlfir::Entity lhs) {
if constexpr (TC1 == Fortran::common::TypeCategory::Character &&
TC2 == TC1) {
return hlfir::convertCharacterKind(loc, builder, lhs, KIND);
}
mlir::Type type = Fortran::lower::getFIRType(builder.getContext(), TC1,
KIND, /*params=*/{});
mlir::Value res = builder.convertWithSemantics(loc, type, lhs);
return hlfir::EntityWithAttributes{res};
}
static void
genResultTypeParams(mlir::Location loc, fir::FirOpBuilder &builder,
hlfir::Entity lhs,
llvm::SmallVectorImpl<mlir::Value> &resultTypeParams) {
hlfir::genLengthParameters(loc, builder, lhs, resultTypeParams);
}
};
static bool hasDeferredCharacterLength(const Fortran::semantics::Symbol &sym) {
const Fortran::semantics::DeclTypeSpec *type = sym.GetType();
return type &&
type->category() ==
Fortran::semantics::DeclTypeSpec::Category::Character &&
type->characterTypeSpec().length().isDeferred();
}
/// Lower Expr to HLFIR.
class HlfirBuilder {
public:
HlfirBuilder(mlir::Location loc, Fortran::lower::AbstractConverter &converter,
Fortran::lower::SymMap &symMap,
Fortran::lower::StatementContext &stmtCtx)
: converter{converter}, symMap{symMap}, stmtCtx{stmtCtx}, loc{loc} {}
template <typename T>
hlfir::EntityWithAttributes gen(const Fortran::evaluate::Expr<T> &expr) {
if (const Fortran::lower::ExprToValueMap *map =
getConverter().getExprOverrides()) {
if constexpr (std::is_same_v<T, Fortran::evaluate::SomeType>) {
if (auto match = map->find(&expr); match != map->end())
return hlfir::EntityWithAttributes{match->second};
} else {
Fortran::lower::SomeExpr someExpr = toEvExpr(expr);
if (auto match = map->find(&someExpr); match != map->end())
return hlfir::EntityWithAttributes{match->second};
}
}
return Fortran::common::visit([&](const auto &x) { return gen(x); },
expr.u);
}
private:
hlfir::EntityWithAttributes
gen(const Fortran::evaluate::BOZLiteralConstant &expr) {
// BOZ literals reach lowering only from BGE/BGT/BLE/BLT intrinsics when at
// least one operand is not constant (otherwise folds). For all other
// intrinsics, semantics converts BOZ to the expected type before lowering.
Fortran::evaluate::Constant<Fortran::evaluate::LargestInt> intConstant{
expr};
return gen(intConstant);
}
hlfir::EntityWithAttributes gen(const Fortran::evaluate::NullPointer &expr) {
auto nullop = hlfir::NullOp::create(getBuilder(), getLoc());
return mlir::cast<fir::FortranVariableOpInterface>(nullop.getOperation());
}
hlfir::EntityWithAttributes
gen(const Fortran::evaluate::ProcedureDesignator &proc) {
return Fortran::lower::convertProcedureDesignatorToHLFIR(
getLoc(), getConverter(), proc, getSymMap(), getStmtCtx());
}
hlfir::EntityWithAttributes gen(const Fortran::evaluate::ProcedureRef &expr) {
Fortran::evaluate::ProcedureDesignator proc{expr.proc()};
auto procTy{Fortran::lower::translateSignature(proc, getConverter())};
auto result = Fortran::lower::convertCallToHLFIR(getLoc(), getConverter(),
expr, procTy.getResult(0),
getSymMap(), getStmtCtx());
assert(result.has_value());
return *result;
}
template <typename T>
hlfir::EntityWithAttributes
gen(const Fortran::evaluate::Designator<T> &designator) {
return HlfirDesignatorBuilder(getLoc(), getConverter(), getSymMap(),
getStmtCtx())
.gen(designator.u);
}
template <typename T>
hlfir::EntityWithAttributes
gen(const Fortran::evaluate::FunctionRef<T> &expr) {
mlir::Type resType =
Fortran::lower::TypeBuilder<T>::genType(getConverter(), expr);
auto result = Fortran::lower::convertCallToHLFIR(
getLoc(), getConverter(), expr, resType, getSymMap(), getStmtCtx());
assert(result.has_value());
return *result;
}
template <typename T>
hlfir::EntityWithAttributes gen(const Fortran::evaluate::Constant<T> &expr) {
mlir::Location loc = getLoc();
fir::FirOpBuilder &builder = getBuilder();
fir::ExtendedValue exv = Fortran::lower::convertConstant(
converter, loc, expr, /*outlineBigConstantInReadOnlyMemory=*/true);
if (const auto *scalarBox = exv.getUnboxed())
if (fir::isa_trivial(scalarBox->getType()))
return hlfir::EntityWithAttributes(*scalarBox);
if (auto addressOf = fir::getBase(exv).getDefiningOp<fir::AddrOfOp>()) {
auto flags = fir::FortranVariableFlagsAttr::get(
builder.getContext(), fir::FortranVariableFlagsEnum::parameter);
return hlfir::genDeclare(
loc, builder, exv,
addressOf.getSymbol().getRootReference().getValue(), flags);
}
fir::emitFatalError(loc, "Constant<T> was lowered to unexpected format");
}
template <typename T>
hlfir::EntityWithAttributes
gen(const Fortran::evaluate::ArrayConstructor<T> &arrayCtor) {
return Fortran::lower::ArrayConstructorBuilder<T>::gen(
getLoc(), getConverter(), arrayCtor, getSymMap(), getStmtCtx());
}
template <typename D, typename R, typename O>
hlfir::EntityWithAttributes
gen(const Fortran::evaluate::Operation<D, R, O> &op) {
auto &builder = getBuilder();
mlir::Location loc = getLoc();
const int rank = op.Rank();
UnaryOp<D> unaryOp;
auto left = hlfir::loadTrivialScalar(loc, builder, gen(op.left()));
llvm::SmallVector<mlir::Value, 1> typeParams;
if constexpr (R::category == Fortran::common::TypeCategory::Character) {
unaryOp.genResultTypeParams(loc, builder, left, typeParams);
}
if (rank == 0)
return unaryOp.gen(loc, builder, op.derived(), left);
// Elemental expression.
mlir::Type elementType;
if constexpr (R::category == Fortran::common::TypeCategory::Derived) {
if (op.derived().GetType().IsUnlimitedPolymorphic())
elementType = mlir::NoneType::get(builder.getContext());
else
elementType = Fortran::lower::translateDerivedTypeToFIRType(
getConverter(), op.derived().GetType().GetDerivedTypeSpec());
} else {
elementType =
Fortran::lower::getFIRType(builder.getContext(), R::category, R::kind,
/*params=*/{});
}
mlir::Value shape = hlfir::genShape(loc, builder, left);
auto genKernel = [&op, &left, &unaryOp](
mlir::Location l, fir::FirOpBuilder &b,
mlir::ValueRange oneBasedIndices) -> hlfir::Entity {
auto leftElement = hlfir::getElementAt(l, b, left, oneBasedIndices);
auto leftVal = hlfir::loadTrivialScalar(l, b, leftElement);
return unaryOp.gen(l, b, op.derived(), leftVal);
};
mlir::Value elemental = hlfir::genElementalOp(
loc, builder, elementType, shape, typeParams, genKernel,
/*isUnordered=*/true, left.isPolymorphic() ? left : mlir::Value{});
fir::FirOpBuilder *bldr = &builder;
getStmtCtx().attachCleanup(
[=]() { hlfir::DestroyOp::create(*bldr, loc, elemental); });
return hlfir::EntityWithAttributes{elemental};
}
template <typename D, typename R, typename LO, typename RO>
hlfir::EntityWithAttributes
gen(const Fortran::evaluate::Operation<D, R, LO, RO> &op) {
auto &builder = getBuilder();
mlir::Location loc = getLoc();
const int rank = op.Rank();
BinaryOp<D> binaryOp;
auto left = hlfir::loadTrivialScalar(loc, builder, gen(op.left()));
auto right = hlfir::loadTrivialScalar(loc, builder, gen(op.right()));
// "A op (...)" or "(...) op A" may need to have their reassoc flag
// turned off
const bool leftIsParens = isParenthesized(op.left());
const bool rightIsParens = isParenthesized(op.right());
const bool noReassoc =
getConverter().getLoweringOptions().getProtectParens() &&
(leftIsParens || rightIsParens);
llvm::SmallVector<mlir::Value, 1> typeParams;
if constexpr (R::category == Fortran::common::TypeCategory::Character) {
binaryOp.genResultTypeParams(loc, builder, left, right, typeParams);
}
if (rank == 0) {
auto fmfBackup = builder.getFastMathFlags();
if (noReassoc)
builder.setFastMathFlags(fmfBackup &
~mlir::arith::FastMathFlags::reassoc);
auto res = binaryOp.gen(loc, builder, op.derived(), left, right);
builder.setFastMathFlags(fmfBackup);
return res;
}
// Elemental expression.
mlir::Type elementType =
Fortran::lower::getFIRType(builder.getContext(), R::category, R::kind,
/*params=*/{});
// TODO: "merge" shape, get cst shape from front-end if possible.
// Prefer a compile-time constant shape to get a statically shaped result.
auto hasConstantShape = [](hlfir::Entity entity) -> bool {
if (!entity.isArray())
return false;
auto seqTy =
mlir::dyn_cast<fir::SequenceType>(entity.getElementOrSequenceType());
return seqTy && !fir::sequenceWithNonConstantShape(seqTy);
};
mlir::Value shape;
if (hasConstantShape(left))
shape = hlfir::genShape(loc, builder, left);
else if (hasConstantShape(right))
shape = hlfir::genShape(loc, builder, right);
else if (left.isArray())
shape = hlfir::genShape(loc, builder, left);
else {
assert(right.isArray() && "must have at least one array operand");
shape = hlfir::genShape(loc, builder, right);
}
auto genKernel = [&op, &left, &right, &binaryOp, noReassoc](
mlir::Location l, fir::FirOpBuilder &b,
mlir::ValueRange oneBasedIndices) -> hlfir::Entity {
auto fmfBackup = b.getFastMathFlags();
if (noReassoc)
b.setFastMathFlags(fmfBackup & ~mlir::arith::FastMathFlags::reassoc);
auto leftElement = hlfir::getElementAt(l, b, left, oneBasedIndices);
auto rightElement = hlfir::getElementAt(l, b, right, oneBasedIndices);
auto leftVal = hlfir::loadTrivialScalar(l, b, leftElement);
auto rightVal = hlfir::loadTrivialScalar(l, b, rightElement);
auto result = binaryOp.gen(l, b, op.derived(), leftVal, rightVal);
b.setFastMathFlags(fmfBackup);
return result;
};
auto iofBackup = builder.getIntegerOverflowFlags();
// nsw is never added to operations on vector subscripts
// even if -fno-wrapv is enabled.
builder.setIntegerOverflowFlags(mlir::arith::IntegerOverflowFlags::none);
mlir::Value elemental = hlfir::genElementalOp(loc, builder, elementType,
shape, typeParams, genKernel,
/*isUnordered=*/true);
builder.setIntegerOverflowFlags(iofBackup);
fir::FirOpBuilder *bldr = &builder;
getStmtCtx().attachCleanup(
[=]() { hlfir::DestroyOp::create(*bldr, loc, elemental); });
return hlfir::EntityWithAttributes{elemental};
}
hlfir::EntityWithAttributes
gen(const Fortran::evaluate::Relational<Fortran::evaluate::SomeType> &op) {
return Fortran::common::visit([&](const auto &x) { return gen(x); }, op.u);
}
hlfir::EntityWithAttributes gen(const Fortran::evaluate::TypeParamInquiry &) {
TODO(getLoc(), "lowering type parameter inquiry to HLFIR");
}
hlfir::EntityWithAttributes
gen(const Fortran::evaluate::DescriptorInquiry &desc) {
mlir::Location loc = getLoc();
auto &builder = getBuilder();
hlfir::EntityWithAttributes entity =
HlfirDesignatorBuilder(getLoc(), getConverter(), getSymMap(),
getStmtCtx())
.genNamedEntity(desc.base());
using ResTy = Fortran::evaluate::DescriptorInquiry::Result;
mlir::Type resultType =
getConverter().genType(ResTy::category, ResTy::kind);
auto castResult = [&](mlir::Value v) {
return hlfir::EntityWithAttributes{
builder.createConvert(loc, resultType, v)};
};
switch (desc.field()) {
case Fortran::evaluate::DescriptorInquiry::Field::Len:
return castResult(hlfir::genCharLength(loc, builder, entity));
case Fortran::evaluate::DescriptorInquiry::Field::LowerBound:
return castResult(
hlfir::genLBound(loc, builder, entity, desc.dimension()));
case Fortran::evaluate::DescriptorInquiry::Field::Extent:
return castResult(
hlfir::genExtent(loc, builder, entity, desc.dimension()));
case Fortran::evaluate::DescriptorInquiry::Field::Rank:
return castResult(hlfir::genRank(loc, builder, entity, resultType));
case Fortran::evaluate::DescriptorInquiry::Field::Stride:
// So far the front end does not generate this inquiry.
TODO(loc, "stride inquiry");
}
llvm_unreachable("unknown descriptor inquiry");
}
hlfir::EntityWithAttributes
gen(const Fortran::evaluate::RankOneBoundElement &x) {
TODO(getLoc(), "rank-1 bound element lowering");
}
/// Generate a conditional expression as an hlfir.conditional op whose
/// regions yield the then/else values. Materialization into memory is
/// deferred to the bufferization pass.
template <typename T>
hlfir::Entity
genConditionalOp(const Fortran::evaluate::ConditionalExpr<T> &condExpr,
mlir::Type elementType, bool isPolymorphic) {
const mlir::Location loc{getLoc()};
fir::FirOpBuilder &builder{getBuilder()};
// Lower the condition to i1.
getStmtCtx().pushScope();
const hlfir::EntityWithAttributes condEntity{gen(condExpr.condition())};
mlir::Value condition{hlfir::loadTrivialScalar(loc, builder, condEntity)};
condition = builder.createConvert(loc, builder.getI1Type(), condition);
getStmtCtx().finalizeAndPop();
// Build the hlfir.expr result type.
const hlfir::ExprType::Shape shape(condExpr.Rank(),
hlfir::ExprType::getUnknownExtent());
const mlir::Type exprType{hlfir::ExprType::get(builder.getContext(), shape,
elementType, isPolymorphic)};
auto condOp =
hlfir::ConditionalOp::create(builder, loc, exprType, condition);
// Cleanups are placed in the yield's cleanup region so that
// bufferization can replay them after the assign into the temp.
auto genBranch = [&](mlir::Region &region, const auto &value) {
builder.setInsertionPointToStart(&region.front());
getStmtCtx().pushScope();
hlfir::Entity entity{hlfir::derefPointersAndAllocatables(
loc, builder, hlfir::Entity{gen(value)})};
auto yieldOp = hlfir::YieldOp::create(builder, loc, entity);
Fortran::lower::genCleanUpInRegionIfAny(
loc, builder, yieldOp.getCleanup(), getStmtCtx());
if (yieldOp.getCleanup().empty())
getStmtCtx().pop();
};
genBranch(condOp.getThenRegion(), condExpr.thenValue());
genBranch(condOp.getElseRegion(), condExpr.elseValue());
builder.setInsertionPointAfter(condOp);
fir::FirOpBuilder *const bldr{&builder};
mlir::Value result{condOp.getResult()};
getStmtCtx().attachCleanup(
[=]() { hlfir::DestroyOp::create(*bldr, loc, result); });
return hlfir::Entity{result};
}
/// Generate scalar conditional for trivial scalar types using fir.if SSA
/// results; avoids temporary and assignment.
template <typename T>
hlfir::Entity genTrivialScalarConditional(
const Fortran::evaluate::ConditionalExpr<T> &condExpr,
mlir::Type elementType) {
assert(fir::isa_trivial(elementType) &&
"genTrivialScalarConditional only handles trivial scalar types");
const mlir::Location loc{getLoc()};
fir::FirOpBuilder &builder{getBuilder()};
getStmtCtx().pushScope();
const hlfir::EntityWithAttributes condEntity{gen(condExpr.condition())};
mlir::Value condition{hlfir::loadTrivialScalar(loc, builder, condEntity)};
condition = builder.createConvert(loc, builder.getI1Type(), condition);
auto results =
builder
.genIfOp(loc, {elementType}, condition,
/*withElseRegion=*/true)
.genThen([&]() {
getStmtCtx().pushScope();
hlfir::Entity entity{gen(condExpr.thenValue())};
entity = hlfir::loadTrivialScalar(loc, builder, entity);
getStmtCtx().finalizeAndPop();
mlir::Value result =
builder.createConvert(loc, elementType, entity);
fir::ResultOp::create(builder, loc, result);
})
.genElse([&]() {
getStmtCtx().pushScope();
hlfir::Entity entity{gen(condExpr.elseValue())};
entity = hlfir::loadTrivialScalar(loc, builder, entity);
getStmtCtx().finalizeAndPop();
mlir::Value result =
builder.createConvert(loc, elementType, entity);
fir::ResultOp::create(builder, loc, result);
})
.getResults();
getStmtCtx().finalizeAndPop();
return hlfir::Entity{results[0]};
}
/// Conditional expression (Fortran 2023)
template <typename T>
hlfir::EntityWithAttributes
gen(const Fortran::evaluate::ConditionalExpr<T> &condExpr) {
mlir::Type resultType{Fortran::lower::translateSomeExprToFIRType(
converter, toEvExpr(condExpr))};
if (fir::isRecordWithTypeParameters(
hlfir::getFortranElementType(resultType)))
TODO(getLoc(), "conditional expression with length-parameterized "
"derived type");
// Trivial scalar types (INTEGER, REAL, COMPLEX, LOGICAL, UNSIGNED)
// use fir.if SSA results directly — no temporary needed.
const mlir::Type elementType{hlfir::getFortranElementType(resultType)};
if (condExpr.Rank() == 0 && !fir::isPolymorphicType(resultType) &&
fir::isa_trivial(elementType))
return hlfir::EntityWithAttributes{
genTrivialScalarConditional(condExpr, elementType)};
// All other cases: arrays, CHARACTER, polymorphic, non-trivial derived.
// Emit hlfir.conditional to delay materialization to bufferization.
return hlfir::EntityWithAttributes{genConditionalOp(
condExpr, elementType, fir::isPolymorphicType(resultType))};
}
hlfir::EntityWithAttributes
gen(const Fortran::evaluate::ImpliedDoIndex &var) {
mlir::Value value = symMap.lookupImpliedDo(toStringRef(var.name));
if (!value)
fir::emitFatalError(getLoc(), "ac-do-variable has no binding");
// The index value generated by the implied-do has Index type,
// while computations based on it inside the loop body are using
// the original data type. So we need to cast it appropriately.
mlir::Type varTy = getConverter().genType(toEvExpr(var));
value = getBuilder().createConvert(getLoc(), varTy, value);
return hlfir::EntityWithAttributes{value};
}
static bool
isDerivedTypeWithLenParameters(const Fortran::semantics::Symbol &sym) {
if (const Fortran::semantics::DeclTypeSpec *declTy = sym.GetType())
if (const Fortran::semantics::DerivedTypeSpec *derived =
declTy->AsDerived())
return Fortran::semantics::CountLenParameters(*derived) > 0;
return false;
}
// Construct an entity holding the value specified by the
// StructureConstructor. The initialization of the temporary entity
// is done component by component with the help of HLFIR operations
// DesignateOp and AssignOp.
hlfir::EntityWithAttributes
gen(const Fortran::evaluate::StructureConstructor &ctor) {
mlir::Location loc = getLoc();
fir::FirOpBuilder &builder = getBuilder();
mlir::Type ty = translateSomeExprToFIRType(converter, toEvExpr(ctor));
auto recTy = mlir::cast<fir::RecordType>(ty);
if (recTy.isDependentType())
TODO(loc, "structure constructor for derived type with length parameters "
"in HLFIR");
// Allocate scalar temporary that will be initialized
// with the values specified by the constructor.
mlir::Value storagePtr = builder.createTemporary(loc, recTy);
auto varOp = hlfir::EntityWithAttributes{
hlfir::DeclareOp::create(builder, loc, storagePtr, "ctor.temp")};
// Initialize any components that need initialization.
mlir::Value box = builder.createBox(loc, fir::ExtendedValue{varOp});
fir::runtime::genDerivedTypeInitialize(builder, loc, box);
// StructureConstructor values may relate to name of components in parent
// types. These components cannot be addressed directly, the parent
// components must be addressed first. The loop below creates all the
// required chains of hlfir.designate to address the parent components so
// that the StructureConstructor can later be lowered by addressing these
// parent components if needed. Note: the front-end orders the components in
// structure constructors.
using ValueAndParent = std::tuple<const Fortran::lower::SomeExpr &,
const Fortran::semantics::Symbol &,
hlfir::EntityWithAttributes>;
llvm::SmallVector<ValueAndParent> valuesAndParents;
for (const auto &value : llvm::reverse(ctor.values())) {
const Fortran::semantics::Symbol &compSym = *value.first;
hlfir::EntityWithAttributes currentParent = varOp;
for (Fortran::lower::ComponentReverseIterator compIterator(
ctor.result().derivedTypeSpec());
!compIterator.lookup(compSym.name());) {
// Private parent components have mangled names. Get the name from the
// parent symbol.
const Fortran::semantics::Symbol *parentCompSym =
compIterator.getParentComponent();
assert(parentCompSym && "failed to get parent component symbol");
std::string parentName =
converter.getRecordTypeFieldName(*parentCompSym);
// Advance the iterator, but don't use its return value.
compIterator.advanceToParentType();
auto baseRecTy = mlir::cast<fir::RecordType>(
hlfir::getFortranElementType(currentParent.getType()));
auto parentCompType = baseRecTy.getType(parentName);
assert(parentCompType && "failed to retrieve parent component type");
mlir::Type designatorType = builder.getRefType(parentCompType);
mlir::Value newParent = hlfir::DesignateOp::create(
builder, loc, designatorType, currentParent, parentName,
/*compShape=*/mlir::Value{}, hlfir::DesignateOp::Subscripts{},
/*substring=*/mlir::ValueRange{},
/*complexPart=*/std::nullopt,
/*shape=*/mlir::Value{}, /*typeParams=*/mlir::ValueRange{},
fir::FortranVariableFlagsAttr{});
currentParent = hlfir::EntityWithAttributes{newParent};
}
valuesAndParents.emplace_back(
ValueAndParent{value.second.value(), compSym, currentParent});
}
HlfirDesignatorBuilder designatorBuilder(loc, converter, symMap, stmtCtx);
for (const auto &iter : llvm::reverse(valuesAndParents)) {
auto &sym = std::get<const Fortran::semantics::Symbol &>(iter);
auto &expr = std::get<const Fortran::lower::SomeExpr &>(iter);
auto &baseOp = std::get<hlfir::EntityWithAttributes>(iter);
std::string name = converter.getRecordTypeFieldName(sym);
// Generate DesignateOp for the component.
// The designator's result type is just a reference to the component type,
// because the whole component is being designated.
auto baseRecTy = mlir::cast<fir::RecordType>(
hlfir::getFortranElementType(baseOp.getType()));
auto compType = baseRecTy.getType(name);
assert(compType && "failed to retrieve component type");
mlir::Value compShape =
designatorBuilder.genComponentShape(sym, compType);
const bool isDesignatorVolatile =
fir::isa_volatile_type(baseOp.getType());
auto [designatorType, extraAttributeFlags] =
designatorBuilder.genComponentDesignatorTypeAndAttributes(
sym, compType, isDesignatorVolatile);
mlir::Type fieldElemType = hlfir::getFortranElementType(compType);
llvm::SmallVector<mlir::Value, 1> typeParams;
if (auto charType = mlir::dyn_cast<fir::CharacterType>(fieldElemType)) {
if (charType.hasConstantLen()) {
mlir::Type idxType = builder.getIndexType();
typeParams.push_back(
builder.createIntegerConstant(loc, idxType, charType.getLen()));
} else if (!hasDeferredCharacterLength(sym)) {
// If the length is not deferred, this is a parametrized derived type
// where the character length depends on the derived type length
// parameters. Otherwise, this is a pointer/allocatable component and
// the length will be set during the assignment.
TODO(loc, "automatic character component in structure constructor");
}
}
// Convert component symbol attributes to variable attributes.
fir::FortranVariableFlagsAttr attrs =
Fortran::lower::translateSymbolAttributes(builder.getContext(), sym,
extraAttributeFlags);
// Get the component designator.
auto lhs = hlfir::DesignateOp::create(
builder, loc, designatorType, baseOp, name, compShape,
hlfir::DesignateOp::Subscripts{},
/*substring=*/mlir::ValueRange{},
/*complexPart=*/std::nullopt,
/*shape=*/compShape, typeParams, attrs);
if (attrs && bitEnumContainsAny(attrs.getFlags(),
fir::FortranVariableFlagsEnum::pointer)) {
if (Fortran::semantics::IsProcedure(sym)) {
// Procedure pointer components.
if (Fortran::evaluate::UnwrapExpr<Fortran::evaluate::NullPointer>(
expr)) {
auto boxTy{
Fortran::lower::getUntypedBoxProcType(builder.getContext())};
hlfir::Entity rhs(
fir::factory::createNullBoxProc(builder, loc, boxTy));
builder.createStoreWithConvert(loc, rhs, lhs);
continue;
}
hlfir::Entity rhs(getBase(Fortran::lower::convertExprToAddress(
loc, converter, expr, symMap, stmtCtx)));
builder.createStoreWithConvert(loc, rhs, lhs);
continue;
}
// Pointer component construction is just a copy of the box contents.
fir::ExtendedValue lhsExv =
hlfir::translateToExtendedValue(loc, builder, lhs);
auto *toBox = lhsExv.getBoxOf<fir::MutableBoxValue>();
if (!toBox)
fir::emitFatalError(loc, "pointer component designator could not be "
"lowered to mutable box");
Fortran::lower::associateMutableBox(converter, loc, *toBox, expr,
/*lbounds=*/{}, stmtCtx);
continue;
}
// Use generic assignment for all the other cases.
bool allowRealloc =
attrs &&
bitEnumContainsAny(attrs.getFlags(),
fir::FortranVariableFlagsEnum::allocatable);
// If the component is allocatable, then we have to check
// whether the RHS value is allocatable or not.
// If it is not allocatable, then AssignOp can be used directly.
// If it is allocatable, then using AssignOp for unallocated RHS
// will cause illegal dereference. When an unallocated allocatable
// value is used to construct an allocatable component, the component
// must just stay unallocated (see Fortran 2018 7.5.10 point 7).
// If the component is allocatable and RHS is NULL() expression, then
// we can just skip it: the LHS must remain unallocated with its
// defined rank.
if (allowRealloc &&
Fortran::evaluate::UnwrapExpr<Fortran::evaluate::NullPointer>(expr))
continue;
bool keepLhsLength = false;
if (allowRealloc)
if (const Fortran::semantics::DeclTypeSpec *declType = sym.GetType())
keepLhsLength =
declType->category() ==
Fortran::semantics::DeclTypeSpec::Category::Character &&
!declType->characterTypeSpec().length().isDeferred();
// Handle special case when the initializer expression is
// '{%SET_LENGTH(x,const_kind)}'. In structure constructor,
// SET_LENGTH is used for initializers of non-allocatable character
// components so that the front-end can better
// fold and work with these structure constructors.
// Here, they are just noise since the assignment semantics will deal
// with any length mismatch, and creating an extra temp with the lhs
// length is useless.
// TODO: should this be moved into an hlfir.assign + hlfir.set_length
// pattern rewrite?
hlfir::Entity rhs = gen(expr);
if (auto set_length = rhs.getDefiningOp<hlfir::SetLengthOp>())
rhs = hlfir::Entity{set_length.getString()};
// lambda to generate `lhs = rhs` and deal with potential rhs implicit
// cast
auto genAssign = [&] {
rhs = hlfir::loadTrivialScalar(loc, builder, rhs);
auto rhsCastAndCleanup =
hlfir::genTypeAndKindConvert(loc, builder, rhs, lhs.getType(),
/*preserveLowerBounds=*/allowRealloc);
hlfir::AssignOp::create(builder, loc, rhsCastAndCleanup.first, lhs,
allowRealloc,
allowRealloc ? keepLhsLength : false,
/*temporary_lhs=*/true);
if (rhsCastAndCleanup.second)
(*rhsCastAndCleanup.second)();
};
if (!allowRealloc || !rhs.isMutableBox()) {
genAssign();
continue;
}
auto [rhsExv, cleanup] =
hlfir::translateToExtendedValue(loc, builder, rhs);
assert(!cleanup && "unexpected cleanup");
auto *fromBox = rhsExv.getBoxOf<fir::MutableBoxValue>();
if (!fromBox)
fir::emitFatalError(loc, "allocatable entity could not be lowered "
"to mutable box");
mlir::Value isAlloc =
fir::factory::genIsAllocatedOrAssociatedTest(builder, loc, *fromBox);
builder.genIfThen(loc, isAlloc).genThen(genAssign).end();
}
if (fir::isRecordWithAllocatableMember(recTy)) {
// Deallocate allocatable components without calling final subroutines.
// The Fortran 2018 section 9.7.3.2 about deallocation is not ruling
// about the fate of allocatable components of structure constructors,
// and there is no behavior consensus in other compilers.
fir::FirOpBuilder *bldr = &builder;
getStmtCtx().attachCleanup([=]() {
fir::runtime::genDerivedTypeDestroyWithoutFinalization(*bldr, loc, box);
});
}
return varOp;
}
mlir::Location getLoc() const { return loc; }
Fortran::lower::AbstractConverter &getConverter() { return converter; }
fir::FirOpBuilder &getBuilder() { return converter.getFirOpBuilder(); }
Fortran::lower::SymMap &getSymMap() { return symMap; }
Fortran::lower::StatementContext &getStmtCtx() { return stmtCtx; }
Fortran::lower::AbstractConverter &converter;
Fortran::lower::SymMap &symMap;
Fortran::lower::StatementContext &stmtCtx;
mlir::Location loc;
};
template <typename T>
hlfir::Entity
HlfirDesignatorBuilder::genSubscript(const Fortran::evaluate::Expr<T> &expr) {
fir::FirOpBuilder &builder = getBuilder();
mlir::arith::IntegerOverflowFlags iofBackup{};
if (!getConverter().getLoweringOptions().getIntegerWrapAround()) {
iofBackup = builder.getIntegerOverflowFlags();
builder.setIntegerOverflowFlags(mlir::arith::IntegerOverflowFlags::nsw);
}
auto loweredExpr =
HlfirBuilder(getLoc(), getConverter(), getSymMap(), getStmtCtx())
.gen(expr);
if (!getConverter().getLoweringOptions().getIntegerWrapAround())
builder.setIntegerOverflowFlags(iofBackup);
// Skip constant conversions that litters designators and makes generated
// IR harder to read: directly use index constants for constant subscripts.
mlir::Type idxTy = builder.getIndexType();
if (!loweredExpr.isArray() && loweredExpr.getType() != idxTy)
if (std::optional<llvm::APInt> cstIndex =
fir::getIntIfConstant(loweredExpr))
if (std::optional<std::int64_t> cstIndex64 = cstIndex->trySExtValue())
return hlfir::EntityWithAttributes{
builder.createIntegerConstant(getLoc(), idxTy, *cstIndex64)};
return hlfir::loadTrivialScalar(loc, builder, loweredExpr);
}
} // namespace
hlfir::EntityWithAttributes Fortran::lower::convertExprToHLFIR(
mlir::Location loc, Fortran::lower::AbstractConverter &converter,
const Fortran::lower::SomeExpr &expr, Fortran::lower::SymMap &symMap,
Fortran::lower::StatementContext &stmtCtx) {
return HlfirBuilder(loc, converter, symMap, stmtCtx).gen(expr);
}
fir::ExtendedValue Fortran::lower::convertToBox(
mlir::Location loc, Fortran::lower::AbstractConverter &converter,
hlfir::Entity entity, Fortran::lower::StatementContext &stmtCtx,
mlir::Type fortranType, unsigned corank) {
fir::FirOpBuilder &builder = converter.getFirOpBuilder();
auto [exv, cleanup] =
hlfir::convertToBox(loc, builder, entity, fortranType, corank);
if (cleanup)
stmtCtx.attachCleanup(*cleanup);
return exv;
}
fir::ExtendedValue Fortran::lower::convertExprToBox(
mlir::Location loc, Fortran::lower::AbstractConverter &converter,
const Fortran::lower::SomeExpr &expr, Fortran::lower::SymMap &symMap,
Fortran::lower::StatementContext &stmtCtx) {
hlfir::EntityWithAttributes loweredExpr =
HlfirBuilder(loc, converter, symMap, stmtCtx).gen(expr);
return convertToBox(loc, converter, loweredExpr, stmtCtx,
converter.genType(expr));
}
fir::ExtendedValue Fortran::lower::convertToAddress(
mlir::Location loc, Fortran::lower::AbstractConverter &converter,
hlfir::Entity entity, Fortran::lower::StatementContext &stmtCtx,
mlir::Type fortranType) {
fir::FirOpBuilder &builder = converter.getFirOpBuilder();
auto [exv, cleanup] =
hlfir::convertToAddress(loc, builder, entity, fortranType);
if (cleanup)
stmtCtx.attachCleanup(*cleanup);
return exv;
}
fir::ExtendedValue Fortran::lower::convertExprToAddress(
mlir::Location loc, Fortran::lower::AbstractConverter &converter,
const Fortran::lower::SomeExpr &expr, Fortran::lower::SymMap &symMap,
Fortran::lower::StatementContext &stmtCtx) {
hlfir::EntityWithAttributes loweredExpr =
HlfirBuilder(loc, converter, symMap, stmtCtx).gen(expr);
return convertToAddress(loc, converter, loweredExpr, stmtCtx,
converter.genType(expr));
}
fir::ExtendedValue Fortran::lower::convertToValue(
mlir::Location loc, Fortran::lower::AbstractConverter &converter,
hlfir::Entity entity, Fortran::lower::StatementContext &stmtCtx) {
auto &builder = converter.getFirOpBuilder();
auto [exv, cleanup] = hlfir::convertToValue(loc, builder, entity);
if (cleanup)
stmtCtx.attachCleanup(*cleanup);
return exv;
}
fir::ExtendedValue Fortran::lower::convertExprToValue(
mlir::Location loc, Fortran::lower::AbstractConverter &converter,
const Fortran::lower::SomeExpr &expr, Fortran::lower::SymMap &symMap,
Fortran::lower::StatementContext &stmtCtx) {
hlfir::EntityWithAttributes loweredExpr =
HlfirBuilder(loc, converter, symMap, stmtCtx).gen(expr);
return convertToValue(loc, converter, loweredExpr, stmtCtx);
}
fir::ExtendedValue Fortran::lower::convertDataRefToValue(
mlir::Location loc, Fortran::lower::AbstractConverter &converter,
const Fortran::evaluate::DataRef &dataRef, Fortran::lower::SymMap &symMap,
Fortran::lower::StatementContext &stmtCtx) {
fir::FortranVariableOpInterface loweredExpr =
HlfirDesignatorBuilder(loc, converter, symMap, stmtCtx).gen(dataRef);
return convertToValue(loc, converter, loweredExpr, stmtCtx);
}
fir::MutableBoxValue Fortran::lower::convertExprToMutableBox(
mlir::Location loc, Fortran::lower::AbstractConverter &converter,
const Fortran::lower::SomeExpr &expr, Fortran::lower::SymMap &symMap) {
// Pointers and Allocatable cannot be temporary expressions. Temporaries may
// be created while lowering it (e.g. if any indices expression of a
// designator create temporaries), but they can be destroyed before using the
// lowered pointer or allocatable;
Fortran::lower::StatementContext localStmtCtx;
hlfir::EntityWithAttributes loweredExpr =
HlfirBuilder(loc, converter, symMap, localStmtCtx).gen(expr);
fir::ExtendedValue exv = Fortran::lower::translateToExtendedValue(
loc, converter.getFirOpBuilder(), loweredExpr, localStmtCtx);
auto *mutableBox = exv.getBoxOf<fir::MutableBoxValue>();
assert(mutableBox && "expression could not be lowered to mutable box");
return *mutableBox;
}
hlfir::ElementalAddrOp
Fortran::lower::convertVectorSubscriptedExprToElementalAddr(
mlir::Location loc, Fortran::lower::AbstractConverter &converter,
const Fortran::lower::SomeExpr &designatorExpr,
Fortran::lower::SymMap &symMap, Fortran::lower::StatementContext &stmtCtx) {
return HlfirDesignatorBuilder(loc, converter, symMap, stmtCtx)
.convertVectorSubscriptedExprToElementalAddr(designatorExpr);
}
hlfir::Entity Fortran::lower::genVectorSubscriptedDesignatorFirstElementAddress(
mlir::Location loc, Fortran::lower::AbstractConverter &converter,
const Fortran::lower::SomeExpr &expr, Fortran::lower::SymMap &symMap,
Fortran::lower::StatementContext &stmtCtx) {
fir::FirOpBuilder &builder = converter.getFirOpBuilder();
// Get a hlfir.elemental_addr op describing the address of the value
// indexed from the original array.
// Note: the hlfir.elemental_addr op verifier requires it to be inside
// of a hlfir.region_assign op. This operation is never seen by the
// verifier because it is immediately inlined.
hlfir::ElementalAddrOp addrOp = convertVectorSubscriptedExprToElementalAddr(
loc, converter, expr, symMap, stmtCtx);
if (!addrOp.getCleanup().empty())
TODO(converter.getCurrentLocation(),
"Vector subscript requring a cleanup region");
// hlfir.elemental_addr doesn't have a normal lowering because it
// can't return a value. Instead we need to inline it here using
// values for the first element. Similar to hlfir::inlineElementalOp.
mlir::Value one = builder.createIntegerConstant(
converter.getCurrentLocation(), builder.getIndexType(), 1);
mlir::SmallVector<mlir::Value> oneBasedIndices;
oneBasedIndices.resize(addrOp.getIndices().size(), one);
mlir::IRMapping mapper;
mapper.map(addrOp.getIndices(), oneBasedIndices);
assert(addrOp.getElementalRegion().hasOneBlock());
mlir::Operation *newOp;
for (mlir::Operation &op : addrOp.getElementalRegion().back().getOperations())
newOp = builder.clone(op, mapper);
auto yield = mlir::cast<hlfir::YieldOp>(newOp);
addrOp->erase();
if (!yield.getCleanup().empty())
TODO(converter.getCurrentLocation(),
"Vector subscript requring element cleanup");
hlfir::Entity result{yield.getEntity()};
yield->erase();
return result;
}