Files
llvm/clang/lib/CodeGen/ABIInfoImpl.cpp

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

479 lines
17 KiB
C++
Raw Normal View History

//===- ABIInfoImpl.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
//
//===----------------------------------------------------------------------===//
#include "ABIInfoImpl.h"
using namespace clang;
using namespace clang::CodeGen;
// Pin the vtable to this file.
DefaultABIInfo::~DefaultABIInfo() = default;
ABIArgInfo DefaultABIInfo::classifyArgumentType(QualType Ty) const {
Ty = useFirstFieldIfTransparentUnion(Ty);
if (isAggregateTypeForABI(Ty)) {
// Records with non-trivial destructors/copy-constructors should not be
// passed by value.
if (CGCXXABI::RecordArgABI RAA = getRecordArgABI(Ty, getCXXABI()))
return getNaturalAlignIndirect(Ty, getDataLayout().getAllocaAddrSpace(),
RAA == CGCXXABI::RAA_DirectInMemory);
return getNaturalAlignIndirect(Ty, getDataLayout().getAllocaAddrSpace());
}
// Treat an enum type as its underlying type.
if (const auto *ED = Ty->getAsEnumDecl())
Ty = ED->getIntegerType();
ASTContext &Context = getContext();
if (const auto *EIT = Ty->getAs<BitIntType>())
if (EIT->getNumBits() >
Context.getTypeSize(Context.getTargetInfo().hasInt128Type()
? Context.Int128Ty
: Context.LongLongTy))
return getNaturalAlignIndirect(Ty, getDataLayout().getAllocaAddrSpace());
return (isPromotableIntegerTypeForABI(Ty)
? ABIArgInfo::getExtend(Ty, CGT.ConvertType(Ty))
: ABIArgInfo::getDirect());
}
ABIArgInfo DefaultABIInfo::classifyReturnType(QualType RetTy) const {
if (RetTy->isVoidType())
return ABIArgInfo::getIgnore();
if (isAggregateTypeForABI(RetTy))
return getNaturalAlignIndirect(RetTy, getDataLayout().getAllocaAddrSpace());
// Treat an enum type as its underlying type.
if (const auto *ED = RetTy->getAsEnumDecl())
RetTy = ED->getIntegerType();
if (const auto *EIT = RetTy->getAs<BitIntType>())
if (EIT->getNumBits() >
getContext().getTypeSize(getContext().getTargetInfo().hasInt128Type()
? getContext().Int128Ty
: getContext().LongLongTy))
return getNaturalAlignIndirect(RetTy,
getDataLayout().getAllocaAddrSpace());
return (isPromotableIntegerTypeForABI(RetTy) ? ABIArgInfo::getExtend(RetTy)
: ABIArgInfo::getDirect());
}
void DefaultABIInfo::computeInfo(CGFunctionInfo &FI) const {
if (!getCXXABI().classifyReturnType(FI))
FI.getReturnInfo() = classifyReturnType(FI.getReturnType());
for (auto &I : FI.arguments())
I.info = classifyArgumentType(I.type);
}
RValue DefaultABIInfo::EmitVAArg(CodeGenFunction &CGF, Address VAListAddr,
QualType Ty, AggValueSlot Slot) const {
return CGF.EmitLoadOfAnyValue(
CGF.MakeAddrLValue(
EmitVAArgInstr(CGF, VAListAddr, Ty, classifyArgumentType(Ty)), Ty),
Slot);
}
void CodeGen::AssignToArrayRange(CodeGen::CGBuilderTy &Builder,
llvm::Value *Array, llvm::Value *Value,
unsigned FirstIndex, unsigned LastIndex) {
// Alternatively, we could emit this as a loop in the source.
for (unsigned I = FirstIndex; I <= LastIndex; ++I) {
llvm::Value *Cell =
Builder.CreateConstInBoundsGEP1_32(Builder.getInt8Ty(), Array, I);
Builder.CreateAlignedStore(Value, Cell, CharUnits::One());
}
}
bool CodeGen::isAggregateTypeForABI(QualType T) {
return !CodeGenFunction::hasScalarEvaluationKind(T) ||
T->isMemberFunctionPointerType();
}
llvm::Type *CodeGen::getVAListElementType(CodeGenFunction &CGF) {
return CGF.ConvertTypeForMem(
CGF.getContext().getBuiltinVaListType()->getPointeeType());
}
CGCXXABI::RecordArgABI CodeGen::getRecordArgABI(const RecordType *RT,
CGCXXABI &CXXABI) {
[clang] Improve nested name specifier AST representation (#147835) This is a major change on how we represent nested name qualifications in the AST. * The nested name specifier itself and how it's stored is changed. The prefixes for types are handled within the type hierarchy, which makes canonicalization for them super cheap, no memory allocation required. Also translating a type into nested name specifier form becomes a no-op. An identifier is stored as a DependentNameType. The nested name specifier gains a lightweight handle class, to be used instead of passing around pointers, which is similar to what is implemented for TemplateName. There is still one free bit available, and this handle can be used within a PointerUnion and PointerIntPair, which should keep bit-packing aficionados happy. * The ElaboratedType node is removed, all type nodes in which it could previously apply to can now store the elaborated keyword and name qualifier, tail allocating when present. * TagTypes can now point to the exact declaration found when producing these, as opposed to the previous situation of there only existing one TagType per entity. This increases the amount of type sugar retained, and can have several applications, for example in tracking module ownership, and other tools which care about source file origins, such as IWYU. These TagTypes are lazily allocated, in order to limit the increase in AST size. This patch offers a great performance benefit. It greatly improves compilation time for [stdexec](https://github.com/NVIDIA/stdexec). For one datapoint, for `test_on2.cpp` in that project, which is the slowest compiling test, this patch improves `-c` compilation time by about 7.2%, with the `-fsyntax-only` improvement being at ~12%. This has great results on compile-time-tracker as well: ![image](https://github.com/user-attachments/assets/700dce98-2cab-4aa8-97d1-b038c0bee831) This patch also further enables other optimziations in the future, and will reduce the performance impact of template specialization resugaring when that lands. It has some other miscelaneous drive-by fixes. About the review: Yes the patch is huge, sorry about that. Part of the reason is that I started by the nested name specifier part, before the ElaboratedType part, but that had a huge performance downside, as ElaboratedType is a big performance hog. I didn't have the steam to go back and change the patch after the fact. There is also a lot of internal API changes, and it made sense to remove ElaboratedType in one go, versus removing it from one type at a time, as that would present much more churn to the users. Also, the nested name specifier having a different API avoids missing changes related to how prefixes work now, which could make existing code compile but not work. How to review: The important changes are all in `clang/include/clang/AST` and `clang/lib/AST`, with also important changes in `clang/lib/Sema/TreeTransform.h`. The rest and bulk of the changes are mostly consequences of the changes in API. PS: TagType::getDecl is renamed to `getOriginalDecl` in this patch, just for easier to rebasing. I plan to rename it back after this lands. Fixes #136624 Fixes https://github.com/llvm/llvm-project/issues/43179 Fixes https://github.com/llvm/llvm-project/issues/68670 Fixes https://github.com/llvm/llvm-project/issues/92757
2025-08-09 05:06:53 -03:00
const RecordDecl *RD = RT->getOriginalDecl()->getDefinitionOrSelf();
if (const auto *CXXRD = dyn_cast<CXXRecordDecl>(RD))
return CXXABI.getRecordArgABI(CXXRD);
if (!RD->canPassInRegisters())
return CGCXXABI::RAA_Indirect;
return CGCXXABI::RAA_Default;
}
CGCXXABI::RecordArgABI CodeGen::getRecordArgABI(QualType T, CGCXXABI &CXXABI) {
const RecordType *RT = T->getAsCanonical<RecordType>();
if (!RT)
return CGCXXABI::RAA_Default;
return getRecordArgABI(RT, CXXABI);
}
bool CodeGen::classifyReturnType(const CGCXXABI &CXXABI, CGFunctionInfo &FI,
const ABIInfo &Info) {
QualType Ty = FI.getReturnType();
if (const auto *RD = Ty->getAsRecordDecl();
RD && !isa<CXXRecordDecl>(RD) && !RD->canPassInRegisters()) {
FI.getReturnInfo() = Info.getNaturalAlignIndirect(
Ty, Info.getDataLayout().getAllocaAddrSpace());
return true;
[clang] Improve nested name specifier AST representation (#147835) This is a major change on how we represent nested name qualifications in the AST. * The nested name specifier itself and how it's stored is changed. The prefixes for types are handled within the type hierarchy, which makes canonicalization for them super cheap, no memory allocation required. Also translating a type into nested name specifier form becomes a no-op. An identifier is stored as a DependentNameType. The nested name specifier gains a lightweight handle class, to be used instead of passing around pointers, which is similar to what is implemented for TemplateName. There is still one free bit available, and this handle can be used within a PointerUnion and PointerIntPair, which should keep bit-packing aficionados happy. * The ElaboratedType node is removed, all type nodes in which it could previously apply to can now store the elaborated keyword and name qualifier, tail allocating when present. * TagTypes can now point to the exact declaration found when producing these, as opposed to the previous situation of there only existing one TagType per entity. This increases the amount of type sugar retained, and can have several applications, for example in tracking module ownership, and other tools which care about source file origins, such as IWYU. These TagTypes are lazily allocated, in order to limit the increase in AST size. This patch offers a great performance benefit. It greatly improves compilation time for [stdexec](https://github.com/NVIDIA/stdexec). For one datapoint, for `test_on2.cpp` in that project, which is the slowest compiling test, this patch improves `-c` compilation time by about 7.2%, with the `-fsyntax-only` improvement being at ~12%. This has great results on compile-time-tracker as well: ![image](https://github.com/user-attachments/assets/700dce98-2cab-4aa8-97d1-b038c0bee831) This patch also further enables other optimziations in the future, and will reduce the performance impact of template specialization resugaring when that lands. It has some other miscelaneous drive-by fixes. About the review: Yes the patch is huge, sorry about that. Part of the reason is that I started by the nested name specifier part, before the ElaboratedType part, but that had a huge performance downside, as ElaboratedType is a big performance hog. I didn't have the steam to go back and change the patch after the fact. There is also a lot of internal API changes, and it made sense to remove ElaboratedType in one go, versus removing it from one type at a time, as that would present much more churn to the users. Also, the nested name specifier having a different API avoids missing changes related to how prefixes work now, which could make existing code compile but not work. How to review: The important changes are all in `clang/include/clang/AST` and `clang/lib/AST`, with also important changes in `clang/lib/Sema/TreeTransform.h`. The rest and bulk of the changes are mostly consequences of the changes in API. PS: TagType::getDecl is renamed to `getOriginalDecl` in this patch, just for easier to rebasing. I plan to rename it back after this lands. Fixes #136624 Fixes https://github.com/llvm/llvm-project/issues/43179 Fixes https://github.com/llvm/llvm-project/issues/68670 Fixes https://github.com/llvm/llvm-project/issues/92757
2025-08-09 05:06:53 -03:00
}
return CXXABI.classifyReturnType(FI);
}
QualType CodeGen::useFirstFieldIfTransparentUnion(QualType Ty) {
if (const RecordType *UT = Ty->getAsUnionType()) {
[clang] Improve nested name specifier AST representation (#147835) This is a major change on how we represent nested name qualifications in the AST. * The nested name specifier itself and how it's stored is changed. The prefixes for types are handled within the type hierarchy, which makes canonicalization for them super cheap, no memory allocation required. Also translating a type into nested name specifier form becomes a no-op. An identifier is stored as a DependentNameType. The nested name specifier gains a lightweight handle class, to be used instead of passing around pointers, which is similar to what is implemented for TemplateName. There is still one free bit available, and this handle can be used within a PointerUnion and PointerIntPair, which should keep bit-packing aficionados happy. * The ElaboratedType node is removed, all type nodes in which it could previously apply to can now store the elaborated keyword and name qualifier, tail allocating when present. * TagTypes can now point to the exact declaration found when producing these, as opposed to the previous situation of there only existing one TagType per entity. This increases the amount of type sugar retained, and can have several applications, for example in tracking module ownership, and other tools which care about source file origins, such as IWYU. These TagTypes are lazily allocated, in order to limit the increase in AST size. This patch offers a great performance benefit. It greatly improves compilation time for [stdexec](https://github.com/NVIDIA/stdexec). For one datapoint, for `test_on2.cpp` in that project, which is the slowest compiling test, this patch improves `-c` compilation time by about 7.2%, with the `-fsyntax-only` improvement being at ~12%. This has great results on compile-time-tracker as well: ![image](https://github.com/user-attachments/assets/700dce98-2cab-4aa8-97d1-b038c0bee831) This patch also further enables other optimziations in the future, and will reduce the performance impact of template specialization resugaring when that lands. It has some other miscelaneous drive-by fixes. About the review: Yes the patch is huge, sorry about that. Part of the reason is that I started by the nested name specifier part, before the ElaboratedType part, but that had a huge performance downside, as ElaboratedType is a big performance hog. I didn't have the steam to go back and change the patch after the fact. There is also a lot of internal API changes, and it made sense to remove ElaboratedType in one go, versus removing it from one type at a time, as that would present much more churn to the users. Also, the nested name specifier having a different API avoids missing changes related to how prefixes work now, which could make existing code compile but not work. How to review: The important changes are all in `clang/include/clang/AST` and `clang/lib/AST`, with also important changes in `clang/lib/Sema/TreeTransform.h`. The rest and bulk of the changes are mostly consequences of the changes in API. PS: TagType::getDecl is renamed to `getOriginalDecl` in this patch, just for easier to rebasing. I plan to rename it back after this lands. Fixes #136624 Fixes https://github.com/llvm/llvm-project/issues/43179 Fixes https://github.com/llvm/llvm-project/issues/68670 Fixes https://github.com/llvm/llvm-project/issues/92757
2025-08-09 05:06:53 -03:00
const RecordDecl *UD = UT->getOriginalDecl()->getDefinitionOrSelf();
if (UD->hasAttr<TransparentUnionAttr>()) {
assert(!UD->field_empty() && "sema created an empty transparent union");
return UD->field_begin()->getType();
}
}
return Ty;
}
llvm::Value *CodeGen::emitRoundPointerUpToAlignment(CodeGenFunction &CGF,
llvm::Value *Ptr,
CharUnits Align) {
// OverflowArgArea = (OverflowArgArea + Align - 1) & -Align;
llvm::Value *RoundUp = CGF.Builder.CreateConstInBoundsGEP1_32(
CGF.Builder.getInt8Ty(), Ptr, Align.getQuantity() - 1);
return CGF.Builder.CreateIntrinsic(
llvm::Intrinsic::ptrmask, {Ptr->getType(), CGF.IntPtrTy},
{RoundUp, llvm::ConstantInt::get(CGF.IntPtrTy, -Align.getQuantity())},
nullptr, Ptr->getName() + ".aligned");
}
Address
CodeGen::emitVoidPtrDirectVAArg(CodeGenFunction &CGF, Address VAListAddr,
llvm::Type *DirectTy, CharUnits DirectSize,
CharUnits DirectAlign, CharUnits SlotSize,
bool AllowHigherAlign, bool ForceRightAdjust) {
// Cast the element type to i8* if necessary. Some platforms define
// va_list as a struct containing an i8* instead of just an i8*.
if (VAListAddr.getElementType() != CGF.Int8PtrTy)
VAListAddr = VAListAddr.withElementType(CGF.Int8PtrTy);
llvm::Value *Ptr = CGF.Builder.CreateLoad(VAListAddr, "argp.cur");
// If the CC aligns values higher than the slot size, do so if needed.
Address Addr = Address::invalid();
if (AllowHigherAlign && DirectAlign > SlotSize) {
Addr = Address(emitRoundPointerUpToAlignment(CGF, Ptr, DirectAlign),
CGF.Int8Ty, DirectAlign);
} else {
Addr = Address(Ptr, CGF.Int8Ty, SlotSize);
}
// Advance the pointer past the argument, then store that back.
CharUnits FullDirectSize = DirectSize.alignTo(SlotSize);
Address NextPtr =
CGF.Builder.CreateConstInBoundsByteGEP(Addr, FullDirectSize, "argp.next");
CGF.Builder.CreateStore(NextPtr.emitRawPointer(CGF), VAListAddr);
// If the argument is smaller than a slot, and this is a big-endian
// target, the argument will be right-adjusted in its slot.
if (DirectSize < SlotSize && CGF.CGM.getDataLayout().isBigEndian() &&
(!DirectTy->isStructTy() || ForceRightAdjust)) {
Addr = CGF.Builder.CreateConstInBoundsByteGEP(Addr, SlotSize - DirectSize);
}
return Addr.withElementType(DirectTy);
}
RValue CodeGen::emitVoidPtrVAArg(CodeGenFunction &CGF, Address VAListAddr,
QualType ValueTy, bool IsIndirect,
TypeInfoChars ValueInfo,
CharUnits SlotSizeAndAlign,
bool AllowHigherAlign, AggValueSlot Slot,
bool ForceRightAdjust) {
// The size and alignment of the value that was passed directly.
CharUnits DirectSize, DirectAlign;
if (IsIndirect) {
DirectSize = CGF.getPointerSize();
DirectAlign = CGF.getPointerAlign();
} else {
DirectSize = ValueInfo.Width;
DirectAlign = ValueInfo.Align;
}
// Cast the address we've calculated to the right type.
llvm::Type *DirectTy = CGF.ConvertTypeForMem(ValueTy), *ElementTy = DirectTy;
if (IsIndirect) {
unsigned AllocaAS = CGF.CGM.getDataLayout().getAllocaAddrSpace();
DirectTy = llvm::PointerType::get(CGF.getLLVMContext(), AllocaAS);
}
Address Addr = emitVoidPtrDirectVAArg(CGF, VAListAddr, DirectTy, DirectSize,
DirectAlign, SlotSizeAndAlign,
AllowHigherAlign, ForceRightAdjust);
if (IsIndirect) {
Addr = Address(CGF.Builder.CreateLoad(Addr), ElementTy, ValueInfo.Align);
}
return CGF.EmitLoadOfAnyValue(CGF.MakeAddrLValue(Addr, ValueTy), Slot);
}
Address CodeGen::emitMergePHI(CodeGenFunction &CGF, Address Addr1,
llvm::BasicBlock *Block1, Address Addr2,
llvm::BasicBlock *Block2,
const llvm::Twine &Name) {
assert(Addr1.getType() == Addr2.getType());
llvm::PHINode *PHI = CGF.Builder.CreatePHI(Addr1.getType(), 2, Name);
PHI->addIncoming(Addr1.emitRawPointer(CGF), Block1);
PHI->addIncoming(Addr2.emitRawPointer(CGF), Block2);
CharUnits Align = std::min(Addr1.getAlignment(), Addr2.getAlignment());
return Address(PHI, Addr1.getElementType(), Align);
}
bool CodeGen::isEmptyField(ASTContext &Context, const FieldDecl *FD,
bool AllowArrays, bool AsIfNoUniqueAddr) {
if (FD->isUnnamedBitField())
return true;
QualType FT = FD->getType();
// Constant arrays of empty records count as empty, strip them off.
// Constant arrays of zero length always count as empty.
bool WasArray = false;
if (AllowArrays)
while (const ConstantArrayType *AT = Context.getAsConstantArrayType(FT)) {
if (AT->isZeroSize())
return true;
FT = AT->getElementType();
// The [[no_unique_address]] special case below does not apply to
// arrays of C++ empty records, so we need to remember this fact.
WasArray = true;
}
const RecordType *RT = FT->getAsCanonical<RecordType>();
if (!RT)
return false;
// C++ record fields are never empty, at least in the Itanium ABI.
//
// FIXME: We should use a predicate for whether this behavior is true in the
// current ABI.
//
// The exception to the above rule are fields marked with the
// [[no_unique_address]] attribute (since C++20). Those do count as empty
// according to the Itanium ABI. The exception applies only to records,
// not arrays of records, so we must also check whether we stripped off an
// array type above.
[clang] Improve nested name specifier AST representation (#147835) This is a major change on how we represent nested name qualifications in the AST. * The nested name specifier itself and how it's stored is changed. The prefixes for types are handled within the type hierarchy, which makes canonicalization for them super cheap, no memory allocation required. Also translating a type into nested name specifier form becomes a no-op. An identifier is stored as a DependentNameType. The nested name specifier gains a lightweight handle class, to be used instead of passing around pointers, which is similar to what is implemented for TemplateName. There is still one free bit available, and this handle can be used within a PointerUnion and PointerIntPair, which should keep bit-packing aficionados happy. * The ElaboratedType node is removed, all type nodes in which it could previously apply to can now store the elaborated keyword and name qualifier, tail allocating when present. * TagTypes can now point to the exact declaration found when producing these, as opposed to the previous situation of there only existing one TagType per entity. This increases the amount of type sugar retained, and can have several applications, for example in tracking module ownership, and other tools which care about source file origins, such as IWYU. These TagTypes are lazily allocated, in order to limit the increase in AST size. This patch offers a great performance benefit. It greatly improves compilation time for [stdexec](https://github.com/NVIDIA/stdexec). For one datapoint, for `test_on2.cpp` in that project, which is the slowest compiling test, this patch improves `-c` compilation time by about 7.2%, with the `-fsyntax-only` improvement being at ~12%. This has great results on compile-time-tracker as well: ![image](https://github.com/user-attachments/assets/700dce98-2cab-4aa8-97d1-b038c0bee831) This patch also further enables other optimziations in the future, and will reduce the performance impact of template specialization resugaring when that lands. It has some other miscelaneous drive-by fixes. About the review: Yes the patch is huge, sorry about that. Part of the reason is that I started by the nested name specifier part, before the ElaboratedType part, but that had a huge performance downside, as ElaboratedType is a big performance hog. I didn't have the steam to go back and change the patch after the fact. There is also a lot of internal API changes, and it made sense to remove ElaboratedType in one go, versus removing it from one type at a time, as that would present much more churn to the users. Also, the nested name specifier having a different API avoids missing changes related to how prefixes work now, which could make existing code compile but not work. How to review: The important changes are all in `clang/include/clang/AST` and `clang/lib/AST`, with also important changes in `clang/lib/Sema/TreeTransform.h`. The rest and bulk of the changes are mostly consequences of the changes in API. PS: TagType::getDecl is renamed to `getOriginalDecl` in this patch, just for easier to rebasing. I plan to rename it back after this lands. Fixes #136624 Fixes https://github.com/llvm/llvm-project/issues/43179 Fixes https://github.com/llvm/llvm-project/issues/68670 Fixes https://github.com/llvm/llvm-project/issues/92757
2025-08-09 05:06:53 -03:00
if (isa<CXXRecordDecl>(RT->getOriginalDecl()) &&
(WasArray || (!AsIfNoUniqueAddr && !FD->hasAttr<NoUniqueAddressAttr>())))
return false;
return isEmptyRecord(Context, FT, AllowArrays, AsIfNoUniqueAddr);
}
bool CodeGen::isEmptyRecord(ASTContext &Context, QualType T, bool AllowArrays,
bool AsIfNoUniqueAddr) {
const auto *RD = T->getAsRecordDecl();
if (!RD)
return false;
if (RD->hasFlexibleArrayMember())
return false;
// If this is a C++ record, check the bases first.
if (const CXXRecordDecl *CXXRD = dyn_cast<CXXRecordDecl>(RD))
for (const auto &I : CXXRD->bases())
if (!isEmptyRecord(Context, I.getType(), true, AsIfNoUniqueAddr))
return false;
for (const auto *I : RD->fields())
if (!isEmptyField(Context, I, AllowArrays, AsIfNoUniqueAddr))
return false;
return true;
}
[clang][CGRecordLayout] Remove dependency on isZeroSize (#96422) This is a follow-up from the conversation starting at https://github.com/llvm/llvm-project/pull/93809#issuecomment-2173729801 The root problem that motivated the change are external AST sources that compute `ASTRecordLayout`s themselves instead of letting Clang compute them from the AST. One such example is LLDB using DWARF to get the definitive offsets and sizes of C++ structures. Such layouts should be considered correct (modulo buggy DWARF), but various assertions and lowering logic around the `CGRecordLayoutBuilder` relies on the AST having `[[no_unique_address]]` attached to them. This is a layout-altering attribute which is not encoded in DWARF. This causes us LLDB to trip over the various LLVM<->Clang layout consistency checks. There has been precedent for avoiding such layout-altering attributes from affecting lowering with externally-provided layouts (e.g., packed structs). This patch proposes to replace the `isZeroSize` checks in `CGRecordLayoutBuilder` (which roughly means "empty field with [[no_unique_address]]") with checks for `CodeGen::isEmptyField`/`CodeGen::isEmptyRecord`. **Details** The main strategy here was to change the `isZeroSize` check in `CGRecordLowering::accumulateFields` and `CGRecordLowering::accumulateBases` to use the `isEmptyXXX` APIs instead, preventing empty fields from being added to the `Members` and `Bases` structures. The rest of the changes fall out from here, to prevent lookups into these structures (for field numbers or base indices) from failing. Added `isEmptyRecordForLayout` and `isEmptyFieldForLayout` (open to better naming suggestions). The main difference to the existing `isEmptyRecord`/`isEmptyField` APIs, is that the `isEmptyXXXForLayout` counterparts don't have special treatment for `unnamed bitfields`/arrays and also treat fields of empty types as if they had `[[no_unique_address]]` (i.e., just like the `AsIfNoUniqueAddr` in `isEmptyField` does).
2024-07-16 04:59:51 +01:00
bool CodeGen::isEmptyFieldForLayout(const ASTContext &Context,
const FieldDecl *FD) {
if (FD->isZeroLengthBitField())
[clang][CGRecordLayout] Remove dependency on isZeroSize (#96422) This is a follow-up from the conversation starting at https://github.com/llvm/llvm-project/pull/93809#issuecomment-2173729801 The root problem that motivated the change are external AST sources that compute `ASTRecordLayout`s themselves instead of letting Clang compute them from the AST. One such example is LLDB using DWARF to get the definitive offsets and sizes of C++ structures. Such layouts should be considered correct (modulo buggy DWARF), but various assertions and lowering logic around the `CGRecordLayoutBuilder` relies on the AST having `[[no_unique_address]]` attached to them. This is a layout-altering attribute which is not encoded in DWARF. This causes us LLDB to trip over the various LLVM<->Clang layout consistency checks. There has been precedent for avoiding such layout-altering attributes from affecting lowering with externally-provided layouts (e.g., packed structs). This patch proposes to replace the `isZeroSize` checks in `CGRecordLayoutBuilder` (which roughly means "empty field with [[no_unique_address]]") with checks for `CodeGen::isEmptyField`/`CodeGen::isEmptyRecord`. **Details** The main strategy here was to change the `isZeroSize` check in `CGRecordLowering::accumulateFields` and `CGRecordLowering::accumulateBases` to use the `isEmptyXXX` APIs instead, preventing empty fields from being added to the `Members` and `Bases` structures. The rest of the changes fall out from here, to prevent lookups into these structures (for field numbers or base indices) from failing. Added `isEmptyRecordForLayout` and `isEmptyFieldForLayout` (open to better naming suggestions). The main difference to the existing `isEmptyRecord`/`isEmptyField` APIs, is that the `isEmptyXXXForLayout` counterparts don't have special treatment for `unnamed bitfields`/arrays and also treat fields of empty types as if they had `[[no_unique_address]]` (i.e., just like the `AsIfNoUniqueAddr` in `isEmptyField` does).
2024-07-16 04:59:51 +01:00
return true;
if (FD->isUnnamedBitField())
return false;
return isEmptyRecordForLayout(Context, FD->getType());
}
bool CodeGen::isEmptyRecordForLayout(const ASTContext &Context, QualType T) {
const auto *RD = T->getAsRecordDecl();
if (!RD)
[clang][CGRecordLayout] Remove dependency on isZeroSize (#96422) This is a follow-up from the conversation starting at https://github.com/llvm/llvm-project/pull/93809#issuecomment-2173729801 The root problem that motivated the change are external AST sources that compute `ASTRecordLayout`s themselves instead of letting Clang compute them from the AST. One such example is LLDB using DWARF to get the definitive offsets and sizes of C++ structures. Such layouts should be considered correct (modulo buggy DWARF), but various assertions and lowering logic around the `CGRecordLayoutBuilder` relies on the AST having `[[no_unique_address]]` attached to them. This is a layout-altering attribute which is not encoded in DWARF. This causes us LLDB to trip over the various LLVM<->Clang layout consistency checks. There has been precedent for avoiding such layout-altering attributes from affecting lowering with externally-provided layouts (e.g., packed structs). This patch proposes to replace the `isZeroSize` checks in `CGRecordLayoutBuilder` (which roughly means "empty field with [[no_unique_address]]") with checks for `CodeGen::isEmptyField`/`CodeGen::isEmptyRecord`. **Details** The main strategy here was to change the `isZeroSize` check in `CGRecordLowering::accumulateFields` and `CGRecordLowering::accumulateBases` to use the `isEmptyXXX` APIs instead, preventing empty fields from being added to the `Members` and `Bases` structures. The rest of the changes fall out from here, to prevent lookups into these structures (for field numbers or base indices) from failing. Added `isEmptyRecordForLayout` and `isEmptyFieldForLayout` (open to better naming suggestions). The main difference to the existing `isEmptyRecord`/`isEmptyField` APIs, is that the `isEmptyXXXForLayout` counterparts don't have special treatment for `unnamed bitfields`/arrays and also treat fields of empty types as if they had `[[no_unique_address]]` (i.e., just like the `AsIfNoUniqueAddr` in `isEmptyField` does).
2024-07-16 04:59:51 +01:00
return false;
// If this is a C++ record, check the bases first.
if (const CXXRecordDecl *CXXRD = dyn_cast<CXXRecordDecl>(RD)) {
if (CXXRD->isDynamicClass())
return false;
for (const auto &I : CXXRD->bases())
if (!isEmptyRecordForLayout(Context, I.getType()))
return false;
}
for (const auto *I : RD->fields())
if (!isEmptyFieldForLayout(Context, I))
return false;
return true;
}
const Type *CodeGen::isSingleElementStruct(QualType T, ASTContext &Context) {
const auto *RD = T->getAsRecordDecl();
if (!RD)
return nullptr;
if (RD->hasFlexibleArrayMember())
return nullptr;
const Type *Found = nullptr;
// If this is a C++ record, check the bases first.
if (const CXXRecordDecl *CXXRD = dyn_cast<CXXRecordDecl>(RD)) {
for (const auto &I : CXXRD->bases()) {
// Ignore empty records.
if (isEmptyRecord(Context, I.getType(), true))
continue;
// If we already found an element then this isn't a single-element struct.
if (Found)
return nullptr;
// If this is non-empty and not a single element struct, the composite
// cannot be a single element struct.
Found = isSingleElementStruct(I.getType(), Context);
if (!Found)
return nullptr;
}
}
// Check for single element.
for (const auto *FD : RD->fields()) {
QualType FT = FD->getType();
// Ignore empty fields.
if (isEmptyField(Context, FD, true))
continue;
// If we already found an element then this isn't a single-element
// struct.
if (Found)
return nullptr;
// Treat single element arrays as the element.
while (const ConstantArrayType *AT = Context.getAsConstantArrayType(FT)) {
if (AT->getZExtSize() != 1)
break;
FT = AT->getElementType();
}
if (!isAggregateTypeForABI(FT)) {
Found = FT.getTypePtr();
} else {
Found = isSingleElementStruct(FT, Context);
if (!Found)
return nullptr;
}
}
// We don't consider a struct a single-element struct if it has
// padding beyond the element type.
if (Found && Context.getTypeSize(Found) != Context.getTypeSize(T))
return nullptr;
return Found;
}
Address CodeGen::EmitVAArgInstr(CodeGenFunction &CGF, Address VAListAddr,
QualType Ty, const ABIArgInfo &AI) {
// This default implementation defers to the llvm backend's va_arg
// instruction. It can handle only passing arguments directly
// (typically only handled in the backend for primitive types), or
// aggregates passed indirectly by pointer (NOTE: if the "byval"
// flag has ABI impact in the callee, this implementation cannot
// work.)
// Only a few cases are covered here at the moment -- those needed
// by the default abi.
llvm::Value *Val;
if (AI.isIndirect()) {
assert(!AI.getPaddingType() &&
"Unexpected PaddingType seen in arginfo in generic VAArg emitter!");
assert(
!AI.getIndirectRealign() &&
"Unexpected IndirectRealign seen in arginfo in generic VAArg emitter!");
auto TyInfo = CGF.getContext().getTypeInfoInChars(Ty);
CharUnits TyAlignForABI = TyInfo.Align;
llvm::Type *ElementTy = CGF.ConvertTypeForMem(Ty);
llvm::Type *BaseTy = llvm::PointerType::getUnqual(CGF.getLLVMContext());
llvm::Value *Addr =
CGF.Builder.CreateVAArg(VAListAddr.emitRawPointer(CGF), BaseTy);
return Address(Addr, ElementTy, TyAlignForABI);
} else {
assert((AI.isDirect() || AI.isExtend()) &&
"Unexpected ArgInfo Kind in generic VAArg emitter!");
assert(!AI.getInReg() &&
"Unexpected InReg seen in arginfo in generic VAArg emitter!");
assert(!AI.getPaddingType() &&
"Unexpected PaddingType seen in arginfo in generic VAArg emitter!");
assert(!AI.getDirectOffset() &&
"Unexpected DirectOffset seen in arginfo in generic VAArg emitter!");
assert(!AI.getCoerceToType() &&
"Unexpected CoerceToType seen in arginfo in generic VAArg emitter!");
Address Temp = CGF.CreateMemTemp(Ty, "varet");
Val = CGF.Builder.CreateVAArg(VAListAddr.emitRawPointer(CGF),
CGF.ConvertTypeForMem(Ty));
CGF.Builder.CreateStore(Val, Temp);
return Temp;
}
}
bool CodeGen::isSIMDVectorType(ASTContext &Context, QualType Ty) {
return Ty->getAs<VectorType>() && Context.getTypeSize(Ty) == 128;
}
bool CodeGen::isRecordWithSIMDVectorType(ASTContext &Context, QualType Ty) {
const auto *RD = Ty->getAsRecordDecl();
if (!RD)
return false;
// If this is a C++ record, check the bases first.
if (const CXXRecordDecl *CXXRD = dyn_cast<CXXRecordDecl>(RD))
for (const auto &I : CXXRD->bases())
if (!isRecordWithSIMDVectorType(Context, I.getType()))
return false;
for (const auto *i : RD->fields()) {
QualType FT = i->getType();
if (isSIMDVectorType(Context, FT))
return true;
if (isRecordWithSIMDVectorType(Context, FT))
return true;
}
return false;
}