[clang] improve class type sugar preservation in pointers to members #130537
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
mizvekov
requested review from
zygoloid,
cor3ntin,
HighCommander4,
erichkeane and
zyn0217
llvmbot
added
clang
|
@llvm/pr-subscribers-clangd @llvm/pr-subscribers-clang Author: Matheus Izvekov (mizvekov) ChangesThis changes the MemberPointerType representation to use a NestedNameSpecifier instead of a Type to represent the class. Since the qualifiers are always parsed as nested names, there was an impedance mismatch when converting these back and forth into types, and this led to issues in preserving sugar. The nested names are indeed a better match for these, as the differences which a QualType can represent cannot be expressed syntactically, and it also represents the use case more exactly, being either dependent or referring to a CXXRecord, unqualified. This patch also makes the MemberPointerType able to represent sugar for a {up/downcast}cast conversion of the base class, although for now the underlying type is canonical, as preserving the sugar up to that point requires further work. As usual, includes a few drive-by fixes in order to make use of the improvements, and removing some duplications, for example CheckBaseClassAccess is deduplicated from across SemaAccess and SemaCast. Patch is 183.99 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/130537.diff 81 Files Affected:
diff --git a/clang-tools-extra/clang-tidy/bugprone/ComparePointerToMemberVirtualFunctionCheck.cpp b/clang-tools-extra/clang-tidy/bugprone/ComparePointerToMemberVirtualFunctionCheck.cpp
index 9d1d92b989bf1..a8a9e6bdcdff8 100644
--- a/clang-tools-extra/clang-tidy/bugprone/ComparePointerToMemberVirtualFunctionCheck.cpp
+++ b/clang-tools-extra/clang-tidy/bugprone/ComparePointerToMemberVirtualFunctionCheck.cpp
@@ -70,10 +70,7 @@ void ComparePointerToMemberVirtualFunctionCheck::check(
// compare with variable which type is pointer to member function.
llvm::SmallVector<SourceLocation, 12U> SameSignatureVirtualMethods{};
const auto *MPT = cast<MemberPointerType>(DRE->getType().getCanonicalType());
- const Type *T = MPT->getClass();
- if (T == nullptr)
- return;
- const CXXRecordDecl *RD = T->getAsCXXRecordDecl();
+ const CXXRecordDecl *RD = MPT->getMostRecentCXXRecordDecl();
if (RD == nullptr)
return;
diff --git a/clang-tools-extra/clang-tidy/modernize/UseNullptrCheck.cpp b/clang-tools-extra/clang-tidy/modernize/UseNullptrCheck.cpp
index 108717e151b57..a6b00be75abf8 100644
--- a/clang-tools-extra/clang-tidy/modernize/UseNullptrCheck.cpp
+++ b/clang-tools-extra/clang-tidy/modernize/UseNullptrCheck.cpp
@@ -493,8 +493,7 @@ UseNullptrCheck::UseNullptrCheck(StringRef Name, ClangTidyContext *Context)
: ClangTidyCheck(Name, Context),
NullMacrosStr(Options.get("NullMacros", "NULL")),
IgnoredTypes(utils::options::parseStringList(Options.get(
- "IgnoredTypes",
- "std::_CmpUnspecifiedParam::;^std::__cmp_cat::__unspec"))) {
+ "IgnoredTypes", "_CmpUnspecifiedParam;^std::__cmp_cat::__unspec"))) {
StringRef(NullMacrosStr).split(NullMacros, ",");
}
diff --git a/clang-tools-extra/clang-tidy/utils/ExceptionAnalyzer.cpp b/clang-tools-extra/clang-tidy/utils/ExceptionAnalyzer.cpp
index 0fea7946a59f9..bc49fa856bafc 100644
--- a/clang-tools-extra/clang-tidy/utils/ExceptionAnalyzer.cpp
+++ b/clang-tools-extra/clang-tidy/utils/ExceptionAnalyzer.cpp
@@ -178,7 +178,9 @@ bool isFunctionPointerConvertible(QualType From, QualType To) {
// Note: converting Derived::* to Base::* is a different kind of conversion,
// called Pointer-to-member conversion.
- return FromMember->getClass() == ToMember->getClass() &&
+ return FromMember->getQualifier() == ToMember->getQualifier() &&
+ FromMember->getMostRecentCXXRecordDecl() ==
+ ToMember->getMostRecentCXXRecordDecl() &&
FromMember->getPointeeType() == ToMember->getPointeeType();
}
@@ -219,8 +221,8 @@ bool isQualificationConvertiblePointer(QualType From, QualType To,
if (P1->isMemberPointerType())
return P2->isMemberPointerType() &&
- P1->getAs<MemberPointerType>()->getClass() ==
- P2->getAs<MemberPointerType>()->getClass();
+ P1->getAs<MemberPointerType>()->getMostRecentCXXRecordDecl() ==
+ P2->getAs<MemberPointerType>()->getMostRecentCXXRecordDecl();
if (P1->isConstantArrayType())
return P2->isConstantArrayType() &&
diff --git a/clang-tools-extra/clangd/unittests/FindTargetTests.cpp b/clang-tools-extra/clangd/unittests/FindTargetTests.cpp
index fc54f89f4941e..602f61d9ecb41 100644
--- a/clang-tools-extra/clangd/unittests/FindTargetTests.cpp
+++ b/clang-tools-extra/clangd/unittests/FindTargetTests.cpp
@@ -1489,7 +1489,7 @@ TEST_F(FindExplicitReferencesTest, AllRefsInFoo) {
"4: targets = {a}\n"
"5: targets = {a::b}, qualifier = 'a::'\n"
"6: targets = {a::b::S}\n"
- "7: targets = {a::b::S::type}, qualifier = 'struct S::'\n"
+ "7: targets = {a::b::S::type}, qualifier = 'S::'\n"
"8: targets = {y}, decl\n"},
{R"cpp(
void foo() {
diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index 7859871b0493a..c2130acbe51bf 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -161,8 +161,8 @@ related warnings within the method body.
``__attribute__((model("large")))`` on non-TLS globals in x86-64 compilations.
This forces the global to be considered small or large in regards to the
x86-64 code model, regardless of the code model specified for the compilation.
-- Clang now emits a warning ``-Wreserved-init-priority`` instead of a hard error
- when ``__attribute__((init_priority(n)))`` is used with values of n in the
+- Clang now emits a warning ``-Wreserved-init-priority`` instead of a hard error
+ when ``__attribute__((init_priority(n)))`` is used with values of n in the
reserved range [0, 100]. The warning will be treated as an error by default.
- There is a new ``format_matches`` attribute to complement the existing
@@ -233,6 +233,7 @@ Improvements to Clang's diagnostics
under the subgroup ``-Wunsafe-buffer-usage-in-libc-call``.
- Diagnostics on chained comparisons (``a < b < c``) are now an error by default. This can be disabled with
``-Wno-error=parentheses``.
+- Better preservation of member pointer type sugar.
- The ``-Wshift-bool`` warning has been added to warn about shifting a boolean. (#GH28334)
- The :doc:`ThreadSafetyAnalysis` now supports ``-Wthread-safety-pointer``,
diff --git a/clang/include/clang/AST/ASTContext.h b/clang/include/clang/AST/ASTContext.h
index f9a12260a6590..af8c49e99a7ce 100644
--- a/clang/include/clang/AST/ASTContext.h
+++ b/clang/include/clang/AST/ASTContext.h
@@ -1558,10 +1558,9 @@ class ASTContext : public RefCountedBase<ASTContext> {
QualType getRValueReferenceType(QualType T) const;
/// Return the uniqued reference to the type for a member pointer to
- /// the specified type in the specified class.
- ///
- /// The class \p Cls is a \c Type because it could be a dependent name.
- QualType getMemberPointerType(QualType T, const Type *Cls) const;
+ /// the specified type in the specified nested name.
+ QualType getMemberPointerType(QualType T, NestedNameSpecifier *Qualifier,
+ const CXXRecordDecl *Cls) const;
/// Return a non-unique reference to the type for a variable array of
/// the specified element type.
diff --git a/clang/include/clang/AST/ASTNodeTraverser.h b/clang/include/clang/AST/ASTNodeTraverser.h
index 3bc0bdff2bdd1..f557555e96e59 100644
--- a/clang/include/clang/AST/ASTNodeTraverser.h
+++ b/clang/include/clang/AST/ASTNodeTraverser.h
@@ -393,7 +393,9 @@ class ASTNodeTraverser
Visit(T->getPointeeType());
}
void VisitMemberPointerType(const MemberPointerType *T) {
- Visit(T->getClass());
+ // FIXME: Provide a NestedNameSpecifier visitor.
+ Visit(T->getQualifier()->getAsType());
+ Visit(T->getMostRecentCXXRecordDecl());
Visit(T->getPointeeType());
}
void VisitArrayType(const ArrayType *T) { Visit(T->getElementType()); }
@@ -485,7 +487,8 @@ class ASTNodeTraverser
}
}
void VisitMemberPointerTypeLoc(MemberPointerTypeLoc TL) {
- Visit(TL.getClassTInfo()->getTypeLoc());
+ // FIXME: Provide NestedNamespecifierLoc visitor.
+ Visit(TL.getQualifierLoc().getTypeLoc());
}
void VisitVariableArrayTypeLoc(VariableArrayTypeLoc TL) {
Visit(TL.getSizeExpr());
diff --git a/clang/include/clang/AST/CanonicalType.h b/clang/include/clang/AST/CanonicalType.h
index 6699284d215bd..35db68971e029 100644
--- a/clang/include/clang/AST/CanonicalType.h
+++ b/clang/include/clang/AST/CanonicalType.h
@@ -453,7 +453,9 @@ template<>
struct CanProxyAdaptor<MemberPointerType>
: public CanProxyBase<MemberPointerType> {
LLVM_CLANG_CANPROXY_TYPE_ACCESSOR(getPointeeType)
- LLVM_CLANG_CANPROXY_SIMPLE_ACCESSOR(const Type *, getClass)
+ LLVM_CLANG_CANPROXY_SIMPLE_ACCESSOR(NestedNameSpecifier *, getQualifier)
+ LLVM_CLANG_CANPROXY_SIMPLE_ACCESSOR(const CXXRecordDecl *,
+ getMostRecentCXXRecordDecl)
};
// CanProxyAdaptors for arrays are intentionally unimplemented because
diff --git a/clang/include/clang/AST/RecursiveASTVisitor.h b/clang/include/clang/AST/RecursiveASTVisitor.h
index 5ca3e435f033b..67025eb7a5a38 100644
--- a/clang/include/clang/AST/RecursiveASTVisitor.h
+++ b/clang/include/clang/AST/RecursiveASTVisitor.h
@@ -1004,7 +1004,8 @@ DEF_TRAVERSE_TYPE(RValueReferenceType,
{ TRY_TO(TraverseType(T->getPointeeType())); })
DEF_TRAVERSE_TYPE(MemberPointerType, {
- TRY_TO(TraverseType(QualType(T->getClass(), 0)));
+ TRY_TO(TraverseNestedNameSpecifier(T->getQualifier()));
+ TRY_TO(TraverseDecl(T->getMostRecentCXXRecordDecl()));
TRY_TO(TraverseType(T->getPointeeType()));
})
@@ -1269,10 +1270,10 @@ DEF_TRAVERSE_TYPELOC(RValueReferenceType,
// We traverse this in the type case as well, but how is it not reached through
// the pointee type?
DEF_TRAVERSE_TYPELOC(MemberPointerType, {
- if (auto *TSI = TL.getClassTInfo())
- TRY_TO(TraverseTypeLoc(TSI->getTypeLoc()));
+ if (NestedNameSpecifierLoc QL = TL.getQualifierLoc())
+ TRY_TO(TraverseNestedNameSpecifierLoc(QL));
else
- TRY_TO(TraverseType(QualType(TL.getTypePtr()->getClass(), 0)));
+ TRY_TO(TraverseNestedNameSpecifier(TL.getTypePtr()->getQualifier()));
TRY_TO(TraverseTypeLoc(TL.getPointeeLoc()));
})
diff --git a/clang/include/clang/AST/Type.h b/clang/include/clang/AST/Type.h
index ef59bd1621fb8..84f40025b07fc 100644
--- a/clang/include/clang/AST/Type.h
+++ b/clang/include/clang/AST/Type.h
@@ -3524,14 +3524,16 @@ class MemberPointerType : public Type, public llvm::FoldingSetNode {
QualType PointeeType;
/// The class of which the pointee is a member. Must ultimately be a
- /// RecordType, but could be a typedef or a template parameter too.
- const Type *Class;
+ /// CXXRecordType, but could be a typedef or a template parameter too.
+ NestedNameSpecifier *Qualifier;
- MemberPointerType(QualType Pointee, const Type *Cls, QualType CanonicalPtr)
+ MemberPointerType(QualType Pointee, NestedNameSpecifier *Qualifier,
+ QualType CanonicalPtr)
: Type(MemberPointer, CanonicalPtr,
- (Cls->getDependence() & ~TypeDependence::VariablyModified) |
+ (toTypeDependence(Qualifier->getDependence()) &
+ ~TypeDependence::VariablyModified) |
Pointee->getDependence()),
- PointeeType(Pointee), Class(Cls) {}
+ PointeeType(Pointee), Qualifier(Qualifier) {}
public:
QualType getPointeeType() const { return PointeeType; }
@@ -3548,21 +3550,21 @@ class MemberPointerType : public Type, public llvm::FoldingSetNode {
return !PointeeType->isFunctionProtoType();
}
- const Type *getClass() const { return Class; }
+ NestedNameSpecifier *getQualifier() const { return Qualifier; }
CXXRecordDecl *getMostRecentCXXRecordDecl() const;
- bool isSugared() const { return false; }
- QualType desugar() const { return QualType(this, 0); }
+ bool isSugared() const;
+ QualType desugar() const {
+ return isSugared() ? getCanonicalTypeInternal() : QualType(this, 0);
+ }
void Profile(llvm::FoldingSetNodeID &ID) {
- Profile(ID, getPointeeType(), getClass());
+ Profile(ID, getPointeeType(), getQualifier(), getMostRecentCXXRecordDecl());
}
static void Profile(llvm::FoldingSetNodeID &ID, QualType Pointee,
- const Type *Class) {
- ID.AddPointer(Pointee.getAsOpaquePtr());
- ID.AddPointer(Class);
- }
+ const NestedNameSpecifier *Qualifier,
+ const CXXRecordDecl *Cls);
static bool classof(const Type *T) {
return T->getTypeClass() == MemberPointer;
diff --git a/clang/include/clang/AST/TypeLoc.h b/clang/include/clang/AST/TypeLoc.h
index a55a38335ef6a..0fca6e522e03b 100644
--- a/clang/include/clang/AST/TypeLoc.h
+++ b/clang/include/clang/AST/TypeLoc.h
@@ -1355,7 +1355,7 @@ class BlockPointerTypeLoc : public PointerLikeTypeLoc<BlockPointerTypeLoc,
};
struct MemberPointerLocInfo : public PointerLikeLocInfo {
- TypeSourceInfo *ClassTInfo;
+ void *QualifierData;
};
/// Wrapper for source info for member pointers.
@@ -1371,28 +1371,32 @@ class MemberPointerTypeLoc : public PointerLikeTypeLoc<MemberPointerTypeLoc,
setSigilLoc(Loc);
}
- const Type *getClass() const {
- return getTypePtr()->getClass();
- }
-
- TypeSourceInfo *getClassTInfo() const {
- return getLocalData()->ClassTInfo;
+ NestedNameSpecifierLoc getQualifierLoc() const {
+ return NestedNameSpecifierLoc(getTypePtr()->getQualifier(),
+ getLocalData()->QualifierData);
}
- void setClassTInfo(TypeSourceInfo* TI) {
- getLocalData()->ClassTInfo = TI;
+ void setQualifierLoc(NestedNameSpecifierLoc QualifierLoc) {
+ assert(QualifierLoc.getNestedNameSpecifier() ==
+ getTypePtr()->getQualifier() &&
+ "Inconsistent nested-name-specifier pointer");
+ getLocalData()->QualifierData = QualifierLoc.getOpaqueData();
}
void initializeLocal(ASTContext &Context, SourceLocation Loc) {
setSigilLoc(Loc);
- setClassTInfo(nullptr);
+ if (auto *Qualifier = getTypePtr()->getQualifier()) {
+ NestedNameSpecifierLocBuilder Builder;
+ Builder.MakeTrivial(Context, Qualifier, Loc);
+ setQualifierLoc(Builder.getWithLocInContext(Context));
+ } else
+ getLocalData()->QualifierData = nullptr;
}
SourceRange getLocalSourceRange() const {
- if (TypeSourceInfo *TI = getClassTInfo())
- return SourceRange(TI->getTypeLoc().getBeginLoc(), getStarLoc());
- else
- return SourceRange(getStarLoc());
+ if (NestedNameSpecifierLoc QL = getQualifierLoc())
+ return SourceRange(QL.getBeginLoc(), getStarLoc());
+ return SourceRange(getStarLoc());
}
};
diff --git a/clang/include/clang/AST/TypeProperties.td b/clang/include/clang/AST/TypeProperties.td
index 6f1a76bd18fb5..27f71bf5cc62f 100644
--- a/clang/include/clang/AST/TypeProperties.td
+++ b/clang/include/clang/AST/TypeProperties.td
@@ -100,12 +100,15 @@ let Class = MemberPointerType in {
def : Property<"pointeeType", QualType> {
let Read = [{ node->getPointeeType() }];
}
- def : Property<"baseType", QualType> {
- let Read = [{ QualType(node->getClass(), 0) }];
+ def : Property<"Qualifier", NestedNameSpecifier> {
+ let Read = [{ node->getQualifier() }];
+ }
+ def : Property<"Cls", DeclRef> {
+ let Read = [{ node->getMostRecentCXXRecordDecl() }];
}
def : Creator<[{
- return ctx.getMemberPointerType(pointeeType, baseType.getTypePtr());
+ return ctx.getMemberPointerType(pointeeType, Qualifier, cast_or_null<CXXRecordDecl>(Cls));
}]>;
}
diff --git a/clang/include/clang/Basic/DiagnosticSemaKinds.td b/clang/include/clang/Basic/DiagnosticSemaKinds.td
index 88d3fb35eea5c..80d3ca6b16bbb 100644
--- a/clang/include/clang/Basic/DiagnosticSemaKinds.td
+++ b/clang/include/clang/Basic/DiagnosticSemaKinds.td
@@ -7038,10 +7038,8 @@ def err_illegal_decl_mempointer_to_reference : Error<
"'%0' declared as a member pointer to a reference of type %1">;
def err_illegal_decl_mempointer_to_void : Error<
"'%0' declared as a member pointer to void">;
-def err_illegal_decl_mempointer_in_nonclass : Error<
- "'%0' does not point into a class">;
-def err_mempointer_in_nonclass_type : Error<
- "member pointer refers into non-class type %0">;
+def err_illegal_decl_mempointer_in_nonclass
+ : Error<"'%0' does not point into a class">;
def err_reference_to_void : Error<"cannot form a reference to 'void'">;
def err_nonfunction_block_type : Error<
"block pointer to non-function type is invalid">;
diff --git a/clang/include/clang/Sema/Sema.h b/clang/include/clang/Sema/Sema.h
index fdef57e84ee3d..35a9eb48204de 100644
--- a/clang/include/clang/Sema/Sema.h
+++ b/clang/include/clang/Sema/Sema.h
@@ -1348,6 +1348,12 @@ class Sema final : public SemaBase {
unsigned DiagID, bool ForceCheck = false,
bool ForceUnprivileged = false);
+ AccessResult CheckBaseClassAccess(
+ SourceLocation AccessLoc, CXXRecordDecl *Base, CXXRecordDecl *Derived,
+ const CXXBasePath &Path, unsigned DiagID,
+ llvm::function_ref<void(PartialDiagnostic &PD)> SetupPDiag,
+ bool ForceCheck = false, bool ForceUnprivileged = false);
+
/// Checks access to all the declarations in the given result set.
void CheckLookupAccess(const LookupResult &R);
@@ -5747,10 +5753,11 @@ class Sema final : public SemaBase {
/// Determine whether the type \p Derived is a C++ class that is
/// derived from the type \p Base.
+ bool IsDerivedFrom(SourceLocation Loc, CXXRecordDecl *Derived,
+ CXXRecordDecl *Base, CXXBasePaths &Paths);
+ bool IsDerivedFrom(SourceLocation Loc, CXXRecordDecl *Derived,
+ CXXRecordDecl *Base);
bool IsDerivedFrom(SourceLocation Loc, QualType Derived, QualType Base);
-
- /// Determine whether the type \p Derived is a C++ class that is
- /// derived from the type \p Base.
bool IsDerivedFrom(SourceLocation Loc, QualType Derived, QualType Base,
CXXBasePaths &Paths);
@@ -10033,15 +10040,24 @@ class Sema final : public SemaBase {
bool InOverloadResolution,
QualType &ConvertedType);
+ enum class CheckMemberPointerConversionResult {
+ Success,
+ DifferentPointee,
+ NotDerived,
+ Ambiguous,
+ Virtual,
+ Inaccessible
+ };
+ enum class CheckMemberPointerConversionDirection : bool { Downcast, Upcast };
/// CheckMemberPointerConversion - Check the member pointer conversion from
/// the expression From to the type ToType. This routine checks for ambiguous
/// or virtual or inaccessible base-to-derived member pointer conversions for
- /// which IsMemberPointerConversion has already returned true. It returns true
- /// and produces a diagnostic if there was an error, or returns false
- /// otherwise.
- bool CheckMemberPointerConversion(Expr *From, QualType ToType, CastKind &Kind,
- CXXCastPath &BasePath,
- bool IgnoreBaseAccess);
+ /// which IsMemberPointerConversion has already returned true. It produces a
+ // diagnostic if there was an error.
+ CheckMemberPointerConversionResult CheckMemberPointerConversion(
+ QualType FromType, const MemberPointerType *ToPtrType, CastKind &Kind,
+ CXXCastPath &BasePath, SourceLocation CheckLoc, SourceRange OpRange,
+ bool IgnoreBaseAccess, CheckMemberPointerConversionDirection Direction);
/// IsQualificationConversion - Determines whether the conversion from
/// an rvalue of type FromType to ToType is a qualification conversion
@@ -14869,8 +14885,9 @@ class Sema final : public SemaBase {
///
/// \returns a member pointer type, if successful, or a NULL type if there was
/// an error.
- QualType BuildMemberPointerType(QualType T, QualType Class,
- SourceLocation Loc, DeclarationName Entity);
+ QualType BuildMemberPointerType(QualType T, NestedNameSpecifier *Qualifier,
+ CXXRecordDecl *Cls, SourceLocation Loc,
+ DeclarationName Entity);
/// Build a block pointer type.
///
diff --git a/clang/lib/AST/ASTContext.cpp b/clang/lib/AST/ASTContext.cpp
index fce1c34897da7..a98e749e025fc 100644
--- a/clang/lib/AST/ASTContext.cpp
+++ b/clang/lib/AST/ASTContext.cpp
@@ -3321,7 +3321,8 @@ static void encodeTypeForFunctionPointerAuth(const ASTContext &Ctx,
case Type::MemberPointer: {
OS << "M";
const auto *MPT = T->castAs<MemberPointerType>();
- encodeTypeForFunctionPointerAuth(Ctx, OS, QualType(MPT->getClass(), 0));
+ encodeTypeForFunctionPointerAuth(
+ Ctx, OS, QualType(MPT->getQualifier()->getAsType(), 0));
encodeTypeForFunctionPointerAuth(Ctx, OS, MPT->getPointeeType());
return;
}
@@ -3510,7 +3511,8 @@ uint16_t ASTContext::getPointerAuthTypeDiscriminator(QualType T) {
if (PointeeType->castAs<FunctionProtoType>()->getExceptionSpecType() !=
EST_None) {
QualType FT = getFunctionTypeWithExceptionSpec(PointeeType, EST_None);
- T = getMemberPointerType(FT, MPT->getClass());
+ T = getMemberPointerType(FT, MPT->getQualifier(),
+ MPT->getMostRecentCXXRecordDecl());
}
}
std::unique_ptr<MangleContext> MC(createMangleContext());
@@ -4024,32 +4026...
[truncated]
|
Endilll
reviewed
Mar 10, 2025
mizvekov
force-pushed
the
users/mizvekov/clang-nns-print-improv
branch
2 times, most recently
from
87d712f to
7b6eae7
Compare
erichkeane
reviewed
Mar 10, 2025
There was a problem hiding this comment.
This patch ended up being pretty massive... ANYTHING we can do to reduce the size of this would be very appreciated.
I DID NOT do more than a cursory look over the tests, so if someone else could do some work to make sure there isn't anything awkward/weird happening, I'd appreciate it.
Otherwise, I have some minor concerns/issues, but the code itself seems fine.
It might be possible to split some stuff, but it would be hard to test it individually and avoid regressions in between the patches. |
mizvekov
force-pushed
the
users/mizvekov/clang-member-pointer-qualifier
branch
from
fb38f66 to
94e604f
Compare
|
A dependent PR had been merged meanwhile, and this was not rebased. While this patch is still big, you might have been seeing way more stuff than was actually in the PR. I just rebased, but have not addressed any comments yet. |
|
Could you run this through the compile time tracker? |
|
I had already done so, it's practically no change: https://llvm-compile-time-tracker.com/compare.php?from=cce9f360b50ca5a95d5cb550e27a8f6757423460&to=20841e444c0f8d48c6c9d34a77fe48164f7ae728&stat=instructions:u |
mizvekov
force-pushed
the
users/mizvekov/clang-member-pointer-qualifier
branch
from
94e604f to
c1c419d
Compare
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/10/builds/1752 Here is the relevant piece of the build log for the reference |
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/18/builds/13228 Here is the relevant piece of the build log for the reference |
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/73/builds/14958 Here is the relevant piece of the build log for the reference |
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/181/builds/15922 Here is the relevant piece of the build log for the reference |
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/59/builds/14624 Here is the relevant piece of the build log for the reference |
|
Oof, looks like lldb needs updating too. |
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/163/builds/15578 Here is the relevant piece of the build log for the reference |
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/60/builds/22602 Here is the relevant piece of the build log for the reference |
llvm-sync bot
pushed a commit
to arm/arm-toolchain
that referenced
this pull request
Mar 20, 2025
…ointers to members" (#132215) Reverts llvm/llvm-project#130537 This missed updating lldb, which we didn't notice due to lack of pre-commit CI.
mizvekov
added a commit
that referenced
this pull request
Mar 20, 2025
Original PR: #130537 Reland after updating lldb too. This changes the MemberPointerType representation to use a NestedNameSpecifier instead of a Type to represent the base class. Since the qualifiers are always parsed as nested names, there was an impedance mismatch when converting these back and forth into types, and this led to issues in preserving sugar. The nested names are indeed a better match for these, as the differences which a QualType can represent cannot be expressed syntatically, and they represent the use case more exactly, being either dependent or referring to a CXXRecord, unqualified. This patch also makes the MemberPointerType able to represent sugar for a {up/downcast}cast conversion of the base class, although for now the underlying type is canonical, as preserving the sugar up to that point requires further work. As usual, includes a few drive-by fixes in order to make use of the improvements.
mizvekov
added a commit
that referenced
this pull request
Mar 20, 2025
…er (#132234) Original PR: #130537 Reland after updating lldb too. This changes the MemberPointerType representation to use a NestedNameSpecifier instead of a Type to represent the base class. Since the qualifiers are always parsed as nested names, there was an impedance mismatch when converting these back and forth into types, and this led to issues in preserving sugar. The nested names are indeed a better match for these, as the differences which a QualType can represent cannot be expressed syntatically, and they represent the use case more exactly, being either dependent or referring to a CXXRecord, unqualified. This patch also makes the MemberPointerType able to represent sugar for a {up/downcast}cast conversion of the base class, although for now the underlying type is canonical, as preserving the sugar up to that point requires further work. As usual, includes a few drive-by fixes in order to make use of the improvements.
mizvekov
added a commit
that referenced
this pull request
Mar 21, 2025
Original PR: #130537 Originally reverted due to revert of dependent commit. Relanding with no changes. This changes the MemberPointerType representation to use a NestedNameSpecifier instead of a Type to represent the base class. Since the qualifiers are always parsed as nested names, there was an impedance mismatch when converting these back and forth into types, and this led to issues in preserving sugar. The nested names are indeed a better match for these, as the differences which a QualType can represent cannot be expressed syntatically, and they represent the use case more exactly, being either dependent or referring to a CXXRecord, unqualified. This patch also makes the MemberPointerType able to represent sugar for a {up/downcast}cast conversion of the base class, although for now the underlying type is canonical, as preserving the sugar up to that point requires further work. As usual, includes a few drive-by fixes in order to make use of the improvements.
mizvekov
added a commit
that referenced
this pull request
Mar 21, 2025
…er (#132401) Original PR: #130537 Originally reverted due to revert of dependent commit. Relanding with no changes. This changes the MemberPointerType representation to use a NestedNameSpecifier instead of a Type to represent the base class. Since the qualifiers are always parsed as nested names, there was an impedance mismatch when converting these back and forth into types, and this led to issues in preserving sugar. The nested names are indeed a better match for these, as the differences which a QualType can represent cannot be expressed syntatically, and they represent the use case more exactly, being either dependent or referring to a CXXRecord, unqualified. This patch also makes the MemberPointerType able to represent sugar for a {up/downcast}cast conversion of the base class, although for now the underlying type is canonical, as preserving the sugar up to that point requires further work. As usual, includes a few drive-by fixes in order to make use of the improvements.
mizvekov
added a commit
that referenced
this pull request
Mar 26, 2025
This fixes a regression when interpreting a nested name specifier for deduction purposes in member pointers. This introduces a helper for fully translating a nested name specifier into a type. Other existing potential users will be deduplicated by using this helper in subsequent patches. This regression was introduced here: #130537 and was reported here: #132401 (comment) No release notes, since the regression was never released.
mizvekov
added a commit
that referenced
this pull request
Mar 26, 2025
This fixes a regression when interpreting a nested name specifier for deduction purposes in member pointers. This introduces a helper for fully translating a nested name specifier into a type. Other existing potential users will be deduplicated by using this helper in subsequent patches. This regression was introduced here: #130537 and was reported here: #132401 (comment) No release notes, since the regression was never released.
mizvekov
added a commit
that referenced
this pull request
Mar 26, 2025
This fixes a regression when interpreting a nested name specifier for deduction purposes in member pointers. This introduces a helper for fully translating a nested name specifier into a type. Other existing potential users will be deduplicated by using this helper in subsequent patches. This regression was introduced here: #130537 and was reported here: #132401 (comment) No release notes, since the regression was never released.
mizvekov
added a commit
that referenced
this pull request
Mar 26, 2025
This fixes a regression when interpreting a nested name specifier for deduction purposes in member pointers. This introduces a helper for fully translating a nested name specifier into a type. This regression was introduced here: #130537 and was reported here: #132401 (comment) No release notes, since the regression was never released.
mizvekov
added a commit
that referenced
this pull request
Mar 26, 2025
…#133113) This fixes a regression when interpreting a nested name specifier for deduction purposes in member pointers. This introduces a helper for fully translating a nested name specifier into a type. This regression was introduced here: #130537 and was reported here: #132401 (comment) No release notes, since this regression was never released.
llvm-sync bot
pushed a commit
to arm/arm-toolchain
that referenced
this pull request
Mar 26, 2025
…med classes (#133113) This fixes a regression when interpreting a nested name specifier for deduction purposes in member pointers. This introduces a helper for fully translating a nested name specifier into a type. This regression was introduced here: llvm/llvm-project#130537 and was reported here: llvm/llvm-project#132401 (comment) No release notes, since this regression was never released.
mizvekov
added a commit
that referenced
this pull request
Mar 27, 2025
Fixes a regression introduced in #130537 and reported here #133144 This fixes a crash in ASTStructuralEquivalence where the non-null precondition for IsStructurallyEquivalent would be violated, when comparing member pointers with a dependent class. This also drive-by fixes the ast node traverser for member pointers so it doesn't traverse into the qualifier in case it's not a type, or the class declaration in case there is none. This avoids printing of `<<<NULL>>>` on the text node dumper, which is redundant. No release notes since the regression was never released. Fixes #133144
mizvekov
added a commit
that referenced
this pull request
Mar 28, 2025
…133343) Fixes a regression introduced in #130537 and reported here #133144 This fixes a crash in ASTStructuralEquivalence where the non-null precondition for IsStructurallyEquivalent would be violated, when comparing member pointers with a dependent class. This also drive-by fixes the ast node traverser for member pointers so it doesn't traverse into the qualifier in case it's not a type, or the class declaration in case it would be equivalent to what the qualifier refers. This avoids printing of `<<<NULL>>>` on the text node dumper, which is redundant. No release notes since the regression was never released. Fixes #133144
llvm-sync bot
pushed a commit
to arm/arm-toolchain
that referenced
this pull request
Mar 28, 2025
…er pointer (#133343) Fixes a regression introduced in llvm/llvm-project#130537 and reported here llvm/llvm-project#133144 This fixes a crash in ASTStructuralEquivalence where the non-null precondition for IsStructurallyEquivalent would be violated, when comparing member pointers with a dependent class. This also drive-by fixes the ast node traverser for member pointers so it doesn't traverse into the qualifier in case it's not a type, or the class declaration in case it would be equivalent to what the qualifier refers. This avoids printing of `<<<NULL>>>` on the text node dumper, which is redundant. No release notes since the regression was never released. Fixes llvm/llvm-project#133144
mizvekov
added a commit
that referenced
this pull request
Mar 30, 2025
This implements a missing case for an adjusted member-pointer in getCommonSugaredType, when that was originally implemented here: #130537 This missing case could otherwise cause a crash, so this is a regression fix. This should fix the crash reported here: #132401 (comment) No release notes, since this regression and the underlying feature were never released.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This changes the MemberPointerType representation to use a NestedNameSpecifier instead of a Type to represent the class.
Since the qualifiers are always parsed as nested names, there was an impedance mismatch when converting these back and forth into types, and this led to issues in preserving sugar.
The nested names are indeed a better match for these, as the differences which a QualType can represent cannot be expressed syntactically, and it also represents the use case more exactly, being either dependent or referring to a CXXRecord, unqualified.
This patch also makes the MemberPointerType able to represent sugar for a {up/downcast}cast conversion of the base class, although for now the underlying type is canonical, as preserving the sugar up to that point requires further work.
As usual, includes a few drive-by fixes in order to make use of the improvements, and removing some duplications, for example CheckBaseClassAccess is deduplicated from across SemaAccess and SemaCast.