[clang] use getCommonSugar in an assortment of places

[mizvekov]

Summary

For this patch, a simple search was performed for patterns where there are
two types (usually an LHS and an RHS) which are structurally the same, and there
is some result type which is resolved as either one of them (typically LHS for
consistency).

We change those cases to resolve as the common sugared type between those two,
utilizing the new infrastructure created for this purpose.

Signed-off-by: Matheus Izvekov <[e-mail]>

[mizvekov]

.

[mizvekov]

.

[mizvekov]

.

[mizvekov]

.

[mizvekov]

.

[mizvekov]

.

[mizvekov]

.

[mizvekov]

.

[mizvekov]

rebase

[mizvekov]

• Run libcxx CI

[shafik]

Changing how types are printed can effect LLDB expression parsing tests. I ran the LLDB test suite with this change:

ninja check-lldb

and it changes to the results for TestScalarURem.py which can be run using:

llvm-lit -sv lldb/test --filter TestScalarURem.py

The change looks expected.

[mizvekov]

In D111509#3168178, @shafik wrote:

llvm-lit -sv lldb/test --filter TestScalarURem.py

The change looks expected.

Thanks for letting me know. Unfortunately that test is not supported on my platform (Windows).
I have added a dummy file to lldb dir to try to trigger pre-commit CI there, and hopefully see what the error is.

[rsmith]

Looks good to me.

clang/lib/Sema/SemaExprCXX.cpp

6342–6353

Can we remove this if now? getCommonSugaredType should unify the exception specifications so I think this stops being a special case.

6406–6413

Likewise here, can we remove this if?

clang/test/SemaObjC/format-strings-objc.m

271 ↗

(On Diff #444066)

Not related to this patch, but this new diagnostic is in some ways worse than the old one: casting to long doesn't fix the problem, given that the format specifier here %d expects an int.

[mizvekov]

In D111509#3168178, @shafik wrote:

Changing how types are printed can effect LLDB expression parsing tests. I ran the LLDB test suite with this change:

ninja check-lldb

and it changes to the results for TestScalarURem.py which can be run using:

llvm-lit -sv lldb/test --filter TestScalarURem.py

The change looks expected.

Fixed now, thanks. (And also API/commands/expression/rdar44436068/Test128BitsInteger.py).

[tra]

It appears that this change may result in clang crashing (e.g. it broke CUDA compilation: https://lab.llvm.org/buildbot/#/builders/55/builds/35047/steps/3/logs/stdio).

Reproducer:

class {
  typedef __complex__ a;
  operator=(double b){c *= b} a c

$bin/clang++ -cc1 -triple x86_64-unknown-linux-gnu -fsyntax-only -std=c++14 -disable-llvm-passes -x c++ reproducer.cc
reproducer.cc:2:11: warning: plain '_Complex' requires a type specifier; assuming '_Complex double'
  typedef __complex__ a;
          ^
                      double
reproducer.cc:3:3: error: a type specifier is required for all declarations
  operator=(double b){c *= b} a c
  ^
reproducer.cc:3:34: error: expected ';' at end of declaration list
  operator=(double b){c *= b} a c
                                 ^
                                 ;
reproducer.cc:4:1: error: expected '}'
^
reproducer.cc:1:7: note: to match this '{'
class {
      ^
clang++: work/llvm/repo/llvm/include/llvm/Support/Casting.h:566: decltype(auto) llvm::cast(const From &) [To = clang::BuiltinType, From = clang::QualType]: Assertion `isa<To>(Val) && "cast<Ty>() argument of incompatible type!"' failed.
PLEASE submit a bug report to https://github.com/llvm/llvm-project/issues/ and include the crash backtrace, preprocessed source, and associated run script.
Stack dump:
0.      Program arguments: bin/clang++ -cc1 -triple x86_64-unknown-linux-gnu -fsyntax-only -std=c++14 -disable-llvm-passes -x c++ reproducer.cc
1.      reproducer.cc:3:29: current parser token '}'
2.      reproducer.cc:1:1: parsing struct/union/class body '(anonymous)'
3.      reproducer.cc:3:22: parsing function body '(anonymous class)::operator='
4.      reproducer.cc:3:22: in compound statement ('{}')
 #0 0x000055ad5d7fe49a llvm::sys::PrintStackTrace(llvm::raw_ostream&, int) work/llvm/repo/llvm/lib/Support/Unix/Signals.inc:569:11
 #1 0x000055ad5d7fe64b PrintStackTraceSignalHandler(void*) work/llvm/repo/llvm/lib/Support/Unix/Signals.inc:636:1
 #2 0x000055ad5d7fcc96 llvm::sys::RunSignalHandlers() work/llvm/repo/llvm/lib/Support/Signals.cpp:103:5
 #3 0x000055ad5d7fed75 SignalHandler(int) work/llvm/repo/llvm/lib/Support/Unix/Signals.inc:407:1
 #4 0x00007f4b4e083200 __restore_rt (/lib/x86_64-linux-gnu/libpthread.so.0+0x12200)
 #5 0x00007f4b4daf28c1 raise ./signal/../sysdeps/unix/sysv/linux/raise.c:50:1
 #6 0x00007f4b4dadc546 abort ./stdlib/abort.c:81:7
 #7 0x00007f4b4dadc42f get_sysdep_segment_value ./intl/loadmsgcat.c:509:8
 #8 0x00007f4b4dadc42f _nl_load_domain ./intl/loadmsgcat.c:970:34
 #9 0x00007f4b4daeb242 (/lib/x86_64-linux-gnu/libc.so.6+0x31242)
#10 0x000055ad5e2f39e8 decltype(auto) llvm::cast<clang::BuiltinType, clang::QualType>(clang::QualType const&) work/llvm/repo/llvm/include/llvm/Support/Casting.h:567:43
#11 0x000055ad5e2f3929 clang::BuiltinType const* clang::Type::castAs<clang::BuiltinType>() const work/llvm/build/debug/tools/clang/include/clang/AST/TypeNodes.inc:86:1
#12 0x000055ad62cfe279 clang::ASTContext::getFloatTypeSemantics(clang::QualType) const work/llvm/repo/clang/lib/AST/ASTContext.cpp:1707:14
#13 0x000055ad61e8ad61 unsupportedTypeConversion(clang::Sema const&, clang::QualType, clang::QualType) work/llvm/repo/clang/lib/Sema/SemaExpr.cpp:1256:29
#14 0x000055ad61eb3961 clang::Sema::CheckAssignmentConstraints(clang::QualType, clang::ActionResult<clang::Expr*, true>&, clang::CastKind&, bool) work/llvm/repo/clang/lib/Sema/SemaExpr.cpp:9619:7
#15 0x000055ad61eb30c5 clang::Sema::CheckAssignmentConstraints(clang::SourceLocation, clang::QualType, clang::QualType) work/llvm/repo/clang/lib/Sema/SemaExpr.cpp:9465:3
#16 0x000055ad61ec565f clang::Sema::CheckAssignmentOperands(clang::Expr*, clang::ActionResult<clang::Expr*, true>&, clang::SourceLocation, clang::QualType, clang::BinaryOperatorKind) work/llvm/repo/clang/lib/Sema/SemaExpr.cpp:13919:12
#17 0x000055ad61ea1903 clang::Sema::CreateBuiltinBinOp(clang::SourceLocation, clang::BinaryOperatorKind, clang::Expr*, clang::Expr*) work/llvm/repo/clang/lib/Sema/SemaExpr.cpp:14988:11
#18 0x000055ad61ecaf02 clang::Sema::BuildBinOp(clang::Scope*, clang::SourceLocation, clang::BinaryOperatorKind, clang::Expr*, clang::Expr*) work/llvm/repo/clang/lib/Sema/SemaExpr.cpp:15511:10
#19 0x000055ad61e8a180 clang::Sema::ActOnBinOp(clang::Scope*, clang::SourceLocation, clang::tok::TokenKind, clang::Expr*, clang::Expr*) work/llvm/repo/clang/lib/Sema/SemaExpr.cpp:15323:10

[mizvekov]

Thanks for the repro!

[ronlieb]

i am seeing similar issues in openmp buildbot failing with this patch

[mizvekov]

In D111509#3778277, @ronlieb wrote:

i am seeing similar issues in openmp buildbot failing with this patch

I got a small patch coming up soon.

[mizvekov]

Fix at https://reviews.llvm.org/D133522

[MaskRay]

Since the fix was not merged, I will revert this soon.

I saw a clang segfault with clang++ libstdc++-v3/src/c++98/complex_io.cc

[mizvekov]

In D111509#3778879, @MaskRay wrote:

Since the fix was not merged, I will revert this soon.

I saw a clang segfault with clang++ libstdc++-v3/src/c++98/complex_io.cc

Would you mind if I just merge it? I was hoping there would be someone around to have a quick look, but apparently not.

[MaskRay]

In D111509#3778882, @mizvekov wrote:

In D111509#3778879, @MaskRay wrote:

Since the fix was not merged, I will revert this soon.

I saw a clang segfault with clang++ libstdc++-v3/src/c++98/complex_io.cc

Would you mind if I just merge it? I was hoping there would be someone around to have a quick look, but apparently not.

Sorry, I just reverted it:) You may check whether the fix fixes GCC libstdc++-v3/src/c++98/complex_io.cc

[mizvekov]

clang/lib/Sema/SemaExpr.cpp

1053–1119

@shafik let's continue revision of D133522 in this patch, since that one was a quick fix that I attempted in order to avoid reverting it. But now that it was reverted, I will abandon the other one.

I have highlighted with this comment the areas affected by D133522 so we can continue here.

But now I have refactored the whole thing to be simpler / more readable.

[mizvekov]

In D111509#3778883, @MaskRay wrote:

Sorry, I just reverted it:) You may check whether the fix fixes GCC libstdc++-v3/src/c++98/complex_io.cc

I confirm it does not crash on my system with this patch.

[mizvekov]

clang/lib/Sema/SemaExpr.cpp

1053–1119

Also @davrec if you can please take a look at this, this is the only part of the patch that changed, in order to fix the reason why it was reverted.

[davrec]

The lines you changed (clang/lib/Sema/SemaExpr.cpp:1091-1150) look good (IIUC you just change cast to cast_as and dyn_cast to getAs, and reorganize for clarity). I suggested some renaming and a few more comments to further clarify.

clang/lib/Sema/SemaExpr.cpp

1075–1076

Rename params for clarity, e.g.
LHS->Longer
LHSType->LongerType
RHSType->ShorterType

1078–1090

// Compute the rank of the two types, regardless of whether they are complex.

1118

// Promote the precision of the LHS if not an assignment.

1120

// Promote the precision of the RHS unless it is already the same as the LHS.

[davrec]

clang/lib/Sema/SemaExpr.cpp

1075–1076

Actually I probably have that backwards, I think LHS is the Shorter expression...in any case you see why renaming would be helpful

clang/lib/Sema/SemaExpr.cpp

 static bool handleIntegerToComplexFloatConversion(Sema &S, ExprResult &IntExpr,
                                                   ExprResult &ComplexExpr,
                                                   QualType IntTy,
                                                   QualType ComplexTy,
                                                   bool SkipCast) {
   if (IntTy->isComplexType() || IntTy->isRealFloatingType()) return true;
   if (SkipCast) return false;
   if (IntTy->isIntegerType()) {
     QualType fpTy = cast<ComplexType>(ComplexTy)->getElementType();
     IntExpr = S.ImpCastExprToType(IntExpr.get(), fpTy, CK_IntegralToFloating);
     IntExpr = S.ImpCastExprToType(IntExpr.get(), ComplexTy,
                                   CK_FloatingRealToComplex);
   } else {
     assert(IntTy->isComplexIntegerType());
     IntExpr = S.ImpCastExprToType(IntExpr.get(), ComplexTy,
                                   CK_IntegralComplexToFloatingComplex);
   }
   return false;
 }
 
 /// Handle arithmetic conversion with complex types.  Helper function of
 /// UsualArithmeticConversions()
 static QualType handleComplexFloatConversion(Sema &S, ExprResult &LHS,
                                              ExprResult &RHS, QualType LHSType,
                                              QualType RHSType,
                                              bool IsCompAssign) {
   // if we have an integer operand, the result is the complex type.
   if (!handleIntegerToComplexFloatConversion(S, RHS, LHS, RHSType, LHSType,
                                              /*skipCast*/false))
     return LHSType;
   if (!handleIntegerToComplexFloatConversion(S, LHS, RHS, LHSType, RHSType,
                                              /*skipCast*/IsCompAssign))
     return RHSType;
 
   // This handles complex/complex, complex/float, or float/complex.
   // When both operands are complex, the shorter operand is converted to the
   // type of the longer, and that is the type of the result. This corresponds
   // to what is done when combining two real floating-point operands.
   // The fun begins when size promotion occur across type domains.
   // From H&S 6.3.4: When one operand is complex and the other is a real
   // floating-point type, the less precise type is converted, within it's
   // real or complex domain, to the precision of the other type. For example,
   // when combining a "long double" with a "double _Complex", the
   // "double _Complex" is promoted to "long double _Complex".
 
   // Compute the rank of the two types, regardless of whether they are complex.
   int Order = S.Context.getFloatingTypeOrder(LHSType, RHSType);
 
   auto *LHSComplexType = dyn_cast<ComplexType>(LHSType);
   auto *RHSComplexType = dyn_cast<ComplexType>(RHSType);
   QualType LHSElementType =
       LHSComplexType ? LHSComplexType->getElementType() : LHSType;
   QualType RHSElementType =
       RHSComplexType ? RHSComplexType->getElementType() : RHSType;
 
   QualType ResultType = S.Context.getComplexType(LHSElementType);
   if (Order < 0) {
     // Promote the precision of the LHS if not an assignment.
     ResultType = S.Context.getComplexType(RHSElementType);
     if (!IsCompAssign) {
       if (LHSComplexType)
         LHS =
             S.ImpCastExprToType(LHS.get(), ResultType, CK_FloatingComplexCast);
       else
         LHS = S.ImpCastExprToType(LHS.get(), RHSElementType, CK_FloatingCast);
     }
   } else if (Order > 0) {
     // Promote the precision of the RHS.
     if (RHSComplexType)
       RHS = S.ImpCastExprToType(RHS.get(), ResultType, CK_FloatingComplexCast);
     else
       RHS = S.ImpCastExprToType(RHS.get(), LHSElementType, CK_FloatingCast);
   }
   return ResultType;
 }
 
 /// Handle arithmetic conversion from integer to float.  Helper function
 /// of UsualArithmeticConversions()
 static QualType handleIntToFloatConversion(Sema &S, ExprResult &FloatExpr,
                                            ExprResult &IntExpr,
                                            QualType FloatTy, QualType IntTy,
                                            bool ConvertFloat, bool ConvertInt) {
   if (IntTy->isIntegerType()) {
     if (ConvertInt)
QualType Sema::UsualArithmeticConversions(ExprResult &LHS, ExprResult &RHS,
   }
 
   RHS = UsualUnaryConversions(RHS.get());
   if (RHS.isInvalid())
     return QualType();
 
   // For conversion purposes, we ignore any qualifiers.
   // For example, "const float" and "float" are equivalent.
-  QualType LHSType =
-    Context.getCanonicalType(LHS.get()->getType()).getUnqualifiedType();
-  QualType RHSType =
-    Context.getCanonicalType(RHS.get()->getType()).getUnqualifiedType();
+  QualType LHSType = LHS.get()->getType().getUnqualifiedType();
+  QualType RHSType = RHS.get()->getType().getUnqualifiedType();
 
   // For conversion purposes, we ignore any atomic qualifier on the LHS.
   if (const AtomicType *AtomicLHS = LHSType->getAs<AtomicType>())
     LHSType = AtomicLHS->getValueType();
 
   // If both types are identical, no conversion is needed.
-  if (LHSType == RHSType)
-    return LHSType;
+  if (Context.hasSameType(LHSType, RHSType))
+    return Context.getCommonSugar(LHSType, RHSType);
 
   // If either side is a non-arithmetic type (e.g. a pointer), we are done.
   // The caller can deal with this (e.g. pointer + int).
   if (!LHSType->isArithmeticType() || !RHSType->isArithmeticType())
     return QualType();
 
   // Apply unary and bitfield promotions to the LHS's type.
   QualType LHSUnpromotedType = LHSType;
   if (LHSType->isPromotableIntegerType())
     LHSType = Context.getPromotedIntegerType(LHSType);
   QualType LHSBitfieldPromoteTy = Context.isPromotableBitField(LHS.get());
   if (!LHSBitfieldPromoteTy.isNull())
     LHSType = LHSBitfieldPromoteTy;
   if (LHSType != LHSUnpromotedType && ACK != ACK_CompAssign)
     LHS = ImpCastExprToType(LHS.get(), LHSType, CK_IntegralCast);
 
   // If both types are identical, no conversion is needed.
-  if (LHSType == RHSType)
-    return LHSType;
+  if (Context.hasSameType(LHSType, RHSType))
+    return Context.getCommonSugar(LHSType, RHSType);
 
   // At this point, we have two different arithmetic types.
 
   // Diagnose attempts to convert between __ibm128, __float128 and long double
   // where such conversions currently can't be handled.
   if (unsupportedTypeConversion(*this, LHSType, RHSType))
     return QualType();
 
static bool checkCondition(Sema &S, Expr *Cond, SourceLocation QuestionLoc) {
   // C99 6.5.15p2
   if (CondTy->isScalarType()) return false;
 
   S.Diag(QuestionLoc, diag::err_typecheck_cond_expect_scalar)
     << CondTy << Cond->getSourceRange();
   return true;
 }
 
-/// Handle when one or both operands are void type.
-static QualType checkConditionalVoidType(Sema &S, ExprResult &LHS,
-                                         ExprResult &RHS) {
-    Expr *LHSExpr = LHS.get();
-    Expr *RHSExpr = RHS.get();
-
-    if (!LHSExpr->getType()->isVoidType())
-      S.Diag(RHSExpr->getBeginLoc(), diag::ext_typecheck_cond_one_void)
-          << RHSExpr->getSourceRange();
-    if (!RHSExpr->getType()->isVoidType())
-      S.Diag(LHSExpr->getBeginLoc(), diag::ext_typecheck_cond_one_void)
-          << LHSExpr->getSourceRange();
-    LHS = S.ImpCastExprToType(LHS.get(), S.Context.VoidTy, CK_ToVoid);
-    RHS = S.ImpCastExprToType(RHS.get(), S.Context.VoidTy, CK_ToVoid);
-    return S.Context.VoidTy;
-}
-
 /// Return false if the NullExpr can be promoted to PointerTy,
 /// true otherwise.
 static bool checkConditionalNullPointer(Sema &S, ExprResult &NullExpr,
                                         QualType PointerTy) {
   if ((!PointerTy->isAnyPointerType() && !PointerTy->isBlockPointerType()) ||
       !NullExpr.get()->isNullPointerConstant(S.Context,
                                             Expr::NPC_ValueDependentIsNull))
     return true;
 
   NullExpr = S.ImpCastExprToType(NullExpr.get(), PointerTy, CK_NullToPointer);
   return false;
 }
 
 /// Checks compatibility between two pointers and return the resulting
 /// type.
 static QualType checkConditionalPointerCompatibility(Sema &S, ExprResult &LHS,
                                                      ExprResult &RHS,
                                                      SourceLocation Loc) {
   QualType LHSTy = LHS.get()->getType();
   QualType RHSTy = RHS.get()->getType();
 
   if (S.Context.hasSameType(LHSTy, RHSTy)) {
     // Two identical pointers types are always compatible.
-    return LHSTy;
+    return S.Context.getCommonSugar(LHSTy, RHSTy);
   }
 
   QualType lhptee, rhptee;
 
   // Get the pointee types.
   bool IsBlockPointer = false;
   if (const BlockPointerType *LHSBTy = LHSTy->getAs<BlockPointerType>()) {
     lhptee = LHSBTy->getPointeeType();
if (LHSTy->isArithmeticType() && RHSTy->isArithmeticType()) {
     LHS = ImpCastExprToType(LHS.get(), ResTy, PrepareScalarCast(LHS, ResTy));
     RHS = ImpCastExprToType(RHS.get(), ResTy, PrepareScalarCast(RHS, ResTy));
 
     return ResTy;
   }
 
   // And if they're both bfloat (which isn't arithmetic), that's fine too.
   if (LHSTy->isBFloat16Type() && RHSTy->isBFloat16Type()) {
-    return LHSTy;
+    return Context.getCommonSugar(LHSTy, RHSTy);
   }
 
   // If both operands are the same structure or union type, the result is that
   // type.
   if (const RecordType *LHSRT = LHSTy->getAs<RecordType>()) {    // C99 6.5.15p3
     if (const RecordType *RHSRT = RHSTy->getAs<RecordType>())
       if (LHSRT->getDecl() == RHSRT->getDecl())
         // "If both the operands have structure or union type, the result has
         // that type."  This implies that CV qualifiers are dropped.
-        return LHSTy.getUnqualifiedType();
+        return Context.getCommonSugar(LHSTy.getUnqualifiedType(),
+                                      RHSTy.getUnqualifiedType());
     // FIXME: Type of conditional expression must be complete in C mode.
   }
 
   // C99 6.5.15p5: "If both operands have void type, the result has void type."
   // The following || allows only one side to be void (a GCC-ism).
   if (LHSTy->isVoidType() || RHSTy->isVoidType()) {
-    return checkConditionalVoidType(*this, LHS, RHS);
+    QualType ResTy;
+    if (LHSTy->isVoidType() && RHSTy->isVoidType()) {
+      ResTy = Context.getCommonSugar(LHSTy, RHSTy);
+    } else if (RHSTy->isVoidType()) {
+      ResTy = RHSTy;
+      Diag(RHS.get()->getBeginLoc(), diag::ext_typecheck_cond_one_void)
+          << RHS.get()->getSourceRange();
+    } else {
+      ResTy = LHSTy;
+      Diag(LHS.get()->getBeginLoc(), diag::ext_typecheck_cond_one_void)
+          << LHS.get()->getSourceRange();
+    }
+    LHS = ImpCastExprToType(LHS.get(), ResTy, CK_ToVoid);
+    RHS = ImpCastExprToType(RHS.get(), ResTy, CK_ToVoid);
+    return ResTy;
   }
 
   // C99 6.5.15p6 - "if one operand is a null pointer constant, the result has
   // the type of the other operand."
   if (!checkConditionalNullPointer(*this, RHS, LHSTy)) return LHSTy;
   if (!checkConditionalNullPointer(*this, LHS, RHSTy)) return RHSTy;
 
   // All objective-c pointer type analysis is done here.
if (checkPointerIntegerMismatch(*this, LHS, RHS.get(), QuestionLoc,
     return RHSTy;
   if (checkPointerIntegerMismatch(*this, RHS, LHS.get(), QuestionLoc,
       /*IsIntFirstExpr=*/false))
     return LHSTy;
 
   // Allow ?: operations in which both operands have the same
   // built-in sizeless type.
   if (LHSTy->isSizelessBuiltinType() && Context.hasSameType(LHSTy, RHSTy))
-    return LHSTy;
+    return Context.getCommonSugar(LHSTy, RHSTy);
 
   // Emit a better diagnostic if one of the expressions is a null pointer
   // constant and the other is not a pointer type. In this case, the user most
   // likely forgot to take the address of the other expression.
   if (DiagnoseConditionalForNull(LHS.get(), RHS.get(), QuestionLoc))
     return QualType();
 
   // Otherwise, the operands are not compatible.
QualType Sema::CheckVectorOperands(ExprResult &LHS, ExprResult &RHS,
   // for some operators but not others.
   if (!AllowBothBool &&
       LHSVecType && LHSVecType->getVectorKind() == VectorType::AltiVecBool &&
       RHSVecType && RHSVecType->getVectorKind() == VectorType::AltiVecBool)
     return InvalidOperands(Loc, LHS, RHS);
 
   // If the vector types are identical, return.
   if (Context.hasSameType(LHSType, RHSType))
-    return LHSType;
+    return Context.getCommonSugar(LHSType, RHSType);
 
   // If we have compatible AltiVec and GCC vector types, use the AltiVec type.
   if (LHSVecType && RHSVecType &&
       Context.areCompatibleVectorTypes(LHSType, RHSType)) {
     if (isa<ExtVectorType>(LHSVecType)) {
       RHS = ImpCastExprToType(RHS.get(), LHSType, CK_BitCast);
       return LHSType;
     }
QualType Sema::CheckMatrixElementwiseOperands(ExprResult &LHS, ExprResult &RHS,
   QualType LHSType = LHS.get()->getType().getUnqualifiedType();
   QualType RHSType = RHS.get()->getType().getUnqualifiedType();
 
   const MatrixType *LHSMatType = LHSType->getAs<MatrixType>();
   const MatrixType *RHSMatType = RHSType->getAs<MatrixType>();
   assert((LHSMatType || RHSMatType) && "At least one operand must be a matrix");
 
   if (Context.hasSameType(LHSType, RHSType))
-    return LHSType;
+    return Context.getCommonSugar(LHSType, RHSType);
 
   // Type conversion may change LHS/RHS. Keep copies to the original results, in
   // case we have to return InvalidOperands.
   ExprResult OriginalLHS = LHS;
   ExprResult OriginalRHS = RHS;
   if (LHSMatType && !RHSMatType) {
     RHS = tryConvertExprToType(RHS.get(), LHSMatType->getElementType());
     if (!RHS.isInvalid())
QualType Sema::CheckMatrixMultiplyOperands(ExprResult &LHS, ExprResult &RHS,
   if (RHS.isInvalid())
     return QualType();
 
   auto *LHSMatType = LHS.get()->getType()->getAs<ConstantMatrixType>();
   auto *RHSMatType = RHS.get()->getType()->getAs<ConstantMatrixType>();
   assert((LHSMatType || RHSMatType) && "At least one operand must be a matrix");
 
   if (LHSMatType && RHSMatType) {
+    if (LHSMatType->getCanonicalTypeInternal() ==
+        RHSMatType->getCanonicalTypeInternal()) {
+      return Context.getCommonSugar(LHS.get()->getType().getUnqualifiedType(),
+                                    RHS.get()->getType().getUnqualifiedType());
+    }
+
     if (LHSMatType->getNumColumns() != RHSMatType->getNumRows())
       return InvalidOperands(Loc, LHS, RHS);
 
     if (!Context.hasSameType(LHSMatType->getElementType(),
                              RHSMatType->getElementType()))
       return InvalidOperands(Loc, LHS, RHS);
 
-    return Context.getConstantMatrixType(LHSMatType->getElementType(),
-                                         LHSMatType->getNumRows(),
-                                         RHSMatType->getNumColumns());
+    return Context.getConstantMatrixType(
+        Context.getCommonSugar(LHSMatType->getElementType(),
+                               RHSMatType->getElementType()),
+        LHSMatType->getNumRows(), RHSMatType->getNumColumns());
   }
   return CheckMatrixElementwiseOperands(LHS, RHS, Loc, IsCompAssign);
 }
 
 inline QualType Sema::CheckBitwiseOperands(ExprResult &LHS, ExprResult &RHS,
                                            SourceLocation Loc,
                                            BinaryOperatorKind Opc) {
   checkArithmeticNull(*this, LHS, RHS, Loc, /*IsCompare=*/false);

clang/lib/Sema/SemaExprCXX.cpp

if (LHSVT && RHSVT) {
     }
 
     // If both are vector types, they must be the same type.
     if (!Context.hasSameType(LHSType, RHSType)) {
       Diag(QuestionLoc, diag::err_conditional_vector_mismatched)
           << LHSType << RHSType;
       return {};
     }
-    ResultType = LHSType;
+    ResultType = Context.getCommonSugar(LHSType, RHSType);
   } else if (LHSVT || RHSVT) {
     ResultType = CheckVectorOperands(
         LHS, RHS, QuestionLoc, /*isCompAssign*/ false, /*AllowBothBool*/ true,
         /*AllowBoolConversions*/ false);
     if (ResultType.isNull())
       return {};
   } else {
     // Both are scalar.
-    QualType ResultElementTy;
-    LHSType = LHSType.getCanonicalType().getUnqualifiedType();
-    RHSType = RHSType.getCanonicalType().getUnqualifiedType();
-
-    if (Context.hasSameType(LHSType, RHSType))
-      ResultElementTy = LHSType;
-    else
-      ResultElementTy =
-          UsualArithmeticConversions(LHS, RHS, QuestionLoc, ACK_Conditional);
+    LHSType = LHSType.getUnqualifiedType();
+    RHSType = RHSType.getUnqualifiedType();
+    QualType ResultElementTy =
+        Context.hasSameType(LHSType, RHSType)
+            ? Context.getCommonSugar(LHSType, RHSType)
+            : UsualArithmeticConversions(LHS, RHS, QuestionLoc,
+                                         ACK_Conditional);
 
     if (ResultElementTy->isEnumeralType()) {
       Diag(QuestionLoc, diag::err_conditional_vector_operand_type)
           << ResultElementTy;
       return {};
     }
     if (CondType->isExtVectorType())
       ResultType =
if (LThrow != RThrow) {
       // non-throw-expression operand is a bit-field.
       OK = NonThrow->getObjectKind();
       return NonThrow->getType();
     }
 
     //   -- Both the second and third operands have type void; the result is of
     //      type void and is a prvalue.
     if (LVoid && RVoid)
-      return Context.VoidTy;
+      return Context.getCommonSugar(LTy, RTy);
 
     // Neither holds, error.
     Diag(QuestionLoc, diag::err_conditional_void_nonvoid)
       << (LVoid ? RTy : LTy) << (LVoid ? 0 : 1)
       << LHS.get()->getSourceRange() << RHS.get()->getSourceRange();
     return QualType();
   }
 
QualType Sema::CXXCheckConditionalOperands(ExprResult &Cond, ExprResult &LHS,
   if (Same && LVK == RVK && LVK != VK_PRValue &&
       LHS.get()->isOrdinaryOrBitFieldObject() &&
       RHS.get()->isOrdinaryOrBitFieldObject()) {
     VK = LHS.get()->getValueKind();
     if (LHS.get()->getObjectKind() == OK_BitField ||
         RHS.get()->getObjectKind() == OK_BitField)
       OK = OK_BitField;
 
     // If we have function pointer types, unify them anyway to unify their
     // exception specifications, if any.
     if (LTy->isFunctionPointerType() || LTy->isMemberFunctionPointerType()) {
       Qualifiers Qs = LTy.getQualifiers();
       LTy = FindCompositePointerType(QuestionLoc, LHS, RHS,
                                      /*ConvertArgs*/false);
       LTy = Context.getQualifiedType(LTy, Qs);
 
       assert(!LTy.isNull() && "failed to find composite pointer type for "
                               "canonically equivalent function ptr types");
       assert(Context.hasSameType(LTy, RTy) && "bad composite pointer type");
     }
 
-    return LTy;
+    return Context.getCommonSugar(LTy, RTy);
   }
 
   // C++11 [expr.cond]p5
   //   Otherwise, the result is a prvalue. If the second and third operands
   //   do not have the same type, and either has (cv) class type, ...
   if (!Same && (LTy->isRecordType() || RTy->isRecordType())) {
     //   ... overload resolution is used to determine the conversions (if any)
     //   to be applied to the operands. If the overload resolution fails, the
if (LTy->isRecordType()) {
                                                      RHS);
       if (RHSCopy.isInvalid())
         return QualType();
 
       LHS = LHSCopy;
       RHS = RHSCopy;
     }
 
     // If we have function pointer types, unify them anyway to unify their
     // exception specifications, if any.
     if (LTy->isFunctionPointerType() || LTy->isMemberFunctionPointerType()) {
       LTy = FindCompositePointerType(QuestionLoc, LHS, RHS);
       assert(!LTy.isNull() && "failed to find composite pointer type for "
                               "canonically equivalent function ptr types");
+      return LTy;
     }
 
-    return LTy;
+    return Context.getCommonSugar(LTy, RTy);
   }
 
   // Extension: conditional operator involving vector types.
   if (LTy->isVectorType() || RTy->isVectorType())
     return CheckVectorOperands(LHS, RHS, QuestionLoc, /*isCompAssign*/false,
                                /*AllowBothBool*/true,
                                /*AllowBoolConversions*/false);
 

clang/test/Sema/matrix-type-operators.c

 
 void matrix_matrix_multiply(sx10x10_t a, sx5x10_t b, ix10x5_t c, ix10x10_t d, float sf, char *p) {
   // Check dimension mismatches.
   a = a * b;
   // expected-error@-1 {{invalid operands to binary expression ('sx10x10_t' (aka 'float __attribute__((matrix_type(10, 10)))') and 'sx5x10_t' (aka 'float __attribute__((matrix_type(5, 10)))'))}}
   a *= b;
   // expected-error@-1 {{invalid operands to binary expression ('sx10x10_t' (aka 'float __attribute__((matrix_type(10, 10)))') and 'sx5x10_t' (aka 'float __attribute__((matrix_type(5, 10)))'))}}
   b = a * a;
-  // expected-error@-1 {{assigning to 'sx5x10_t' (aka 'float __attribute__((matrix_type(5, 10)))') from incompatible type 'float __attribute__((matrix_type(10, 10)))'}}
+  // expected-error@-1 {{assigning to 'sx5x10_t' (aka 'float __attribute__((matrix_type(5, 10)))') from incompatible type 'sx10x10_t' (aka 'float __attribute__((matrix_type(10, 10)))')}}
 
   // Check element type mismatches.
   a = b * c;
   // expected-error@-1 {{invalid operands to binary expression ('sx5x10_t' (aka 'float __attribute__((matrix_type(5, 10)))') and 'ix10x5_t' (aka 'int __attribute__((matrix_type(10, 5)))'))}}
   b *= c;
   // expected-error@-1 {{invalid operands to binary expression ('sx5x10_t' (aka 'float __attribute__((matrix_type(5, 10)))') and 'ix10x5_t' (aka 'int __attribute__((matrix_type(10, 5)))'))}}
   d = a * a;
-  // expected-error@-1 {{assigning to 'ix10x10_t' (aka 'int __attribute__((matrix_type(10, 10)))') from incompatible type 'float __attribute__((matrix_type(10, 10)))'}}
+  // expected-error@-1 {{assigning to 'ix10x10_t' (aka 'int __attribute__((matrix_type(10, 10)))') from incompatible type 'sx10x10_t' (aka 'float __attribute__((matrix_type(10, 10)))')}}
 
   p = a * a;
-  // expected-error@-1 {{assigning to 'char *' from incompatible type 'float __attribute__((matrix_type(10, 10)))'}}
+  // expected-error@-1 {{assigning to 'char *' from incompatible type 'sx10x10_t' (aka 'float __attribute__((matrix_type(10, 10)))')}}
 }
 
 void mat_scalar_multiply(sx10x10_t a, sx5x10_t b, float sf, char *p) {
   // Shape of multiplication result does not match the type of b.
   b = a * sf;
   // expected-error@-1 {{assigning to 'sx5x10_t' (aka 'float __attribute__((matrix_type(5, 10)))') from incompatible type 'sx10x10_t' (aka 'float __attribute__((matrix_type(10, 10)))')}}
   b = sf * a;
   // expected-error@-1 {{assigning to 'sx5x10_t' (aka 'float __attribute__((matrix_type(5, 10)))') from incompatible type 'sx10x10_t' (aka 'float __attribute__((matrix_type(10, 10)))')}}

clang/test/Sema/nullability.c

void conditional_expr(int c) {
   typedef NonnullIntP0 _Nonnull NonnullIntP1;
   typedef IntP _Nullable NullableIntP0;
   typedef NullableIntP0 _Nullable NullableIntP1;
   NonnullIntP1 nonnullP2;
   NullableIntP1 nullableP2;
 
   p = c ? nonnullP2 : nonnullP2;
   p = c ? nonnullP2 : nullableP2; // expected-warning{{implicit conversion from nullable pointer 'IntP _Nullable' (aka 'int *') to non-nullable pointer type 'int * _Nonnull'}}
-  p = c ? nullableP2 : nonnullP2; // expected-warning{{implicit conversion from nullable pointer 'NullableIntP1' (aka 'int *') to non-nullable pointer type 'int * _Nonnull'}}
+  p = c ? nullableP2 : nonnullP2; // expected-warning{{implicit conversion from nullable pointer 'IntP _Nullable' (aka 'int *') to non-nullable pointer type 'int * _Nonnull'}}
   p = c ? nullableP2 : nullableP2; // expected-warning{{implicit conversion from nullable pointer 'NullableIntP1' (aka 'int *') to non-nullable pointer type 'int * _Nonnull'}}
 }
 
 // Check nullability of binary conditional expressions.
 void binary_conditional_expr() {
   int * _Nonnull p;
   int * _Nonnull nonnullP;
   int * _Nullable nullableP;

clang/test/Sema/sugar-common-types.c

+// RUN: %clang_cc1 -fsyntax-only -verify %s -std=c99 -triple aarch64-arm-none-eabi -target-feature +bf16 -target-feature +sve
+
+typedef struct N {} N;
+
+typedef int B1;
+typedef B1 X1;
+typedef B1 Y1;
+
+typedef void B2;
+typedef B2 X2;
+typedef B2 Y2;
+
+typedef struct B3 {} B3;
+typedef B3 X3;
+typedef B3 Y3;
+
+typedef struct B4 {} *B4;
+typedef B4 X4;
+typedef B4 Y4;
+
+typedef __bf16 B5;
+typedef B5 X5;
+typedef B5 Y5;
+
+typedef __SVInt8_t B6;
+typedef B6 X6;
+typedef B6 Y6;
+
+N t1 = 0 ? (X1)0 : (Y1)0;   // expected-error {{incompatible type 'B1'}}
+N t2 = 0 ? (X2)0 : 0;       // expected-error {{incompatible type 'X2'}}
+N t3 = 0 ? 0 : (Y2)0;       // expected-error {{incompatible type 'Y2'}}
+N t4 = 0 ? (X2)0 : (Y2)0;   // expected-error {{incompatible type 'B2'}}
+N t5 = 0 ? (X3){} : (B3){}; // expected-error {{incompatible type 'B3'}}
+N t6 = 0 ? (X4)0 : (Y4)0;   // expected-error {{incompatible type 'B4'}}
+
+X5 x5;
+Y5 y5;
+N t7 = 0 ? x5 : y5; // expected-error {{incompatible type 'B5'}}
+
+void f8() {
+  X6 x6;
+  Y6 y6;
+  N t8 = 0 ? x6 : y6; // expected-error {{incompatible type 'B6'}}
+}

clang/test/SemaCXX/matrix-type-operators.cpp

void test_subtract_template(unsigned *Ptr1, float *Ptr2) {
 
   Mat1.value = subtract<unsigned, 3, 3, float, 2, 2, unsigned, 2, 2>(Mat2, Mat3);
   // expected-note@-1 {{in instantiation of function template specialization 'subtract<unsigned int, 3U, 3U, float, 2U, 2U, unsigned int, 2U, 2U>' requested here}}
 }
 
 template <typename EltTy0, unsigned R0, unsigned C0, typename EltTy1, unsigned R1, unsigned C1, typename EltTy2, unsigned R2, unsigned C2>
 typename MyMatrix<EltTy2, R2, C2>::matrix_t multiply(MyMatrix<EltTy0, R0, C0> &A, MyMatrix<EltTy1, R1, C1> &B) {
   char *v1 = A.value * B.value;
-  // expected-error@-1 {{cannot initialize a variable of type 'char *' with an rvalue of type 'unsigned int __attribute__((matrix_type(2, 2)))'}}
+  // expected-error@-1 {{cannot initialize a variable of type 'char *' with an rvalue of type 'MyMatrix<unsigned int, 2, 2>::matrix_t' (aka 'unsigned int __attribute__((matrix_type(2, 2)))')}}
   // expected-error@-2 {{invalid operands to binary expression ('MyMatrix<unsigned int, 3, 2>::matrix_t' (aka 'unsigned int __attribute__((matrix_type(3, 2)))') and 'MyMatrix<unsigned int, 3, 3>::matrix_t' (aka 'unsigned int __attribute__((matrix_type(3, 3)))'))}}
   // expected-error@-3 {{invalid operands to binary expression ('MyMatrix<float, 2, 2>::matrix_t' (aka 'float __attribute__((matrix_type(2, 2)))') and 'MyMatrix<unsigned int, 2, 2>::matrix_t' (aka 'unsigned int __attribute__((matrix_type(2, 2)))'))}}
 
   MyMatrix<int, 5, 6> m;
   B.value = m.value * A.value;
   // expected-error@-1 {{invalid operands to binary expression ('MyMatrix<int, 5, 6>::matrix_t' (aka 'int __attribute__((matrix_type(5, 6)))') and 'MyMatrix<unsigned int, 2, 2>::matrix_t' (aka 'unsigned int __attribute__((matrix_type(2, 2)))'))}}
   // expected-error@-2 {{invalid operands to binary expression ('MyMatrix<int, 5, 6>::matrix_t' (aka 'int __attribute__((matrix_type(5, 6)))') and 'MyMatrix<unsigned int, 3, 2>::matrix_t' (aka 'unsigned int __attribute__((matrix_type(3, 2)))'))}}
   // expected-error@-3 {{invalid operands to binary expression ('MyMatrix<int, 5, 6>::matrix_t' (aka 'int __attribute__((matrix_type(5, 6)))') and 'MyMatrix<float, 2, 2>::matrix_t' (aka 'float __attribute__((matrix_type(2, 2)))'))}}

clang/test/SemaCXX/sugar-common-types.cpp

+// RUN: %clang_cc1 -fsyntax-only -verify %s -std=c++20 -fenable-matrix
+
+enum class N {};
+
+using B1 = int;
+using X1 = B1;
+using Y1 = B1;
+
+using B2 = void;
+using X2 = B2;
+using Y2 = B2;
+
+using A3 = char __attribute__((vector_size(4)));
+using B3 = A3;
+using X3 = B3;
+using Y3 = B3;
+
+using A4 = float;
+using B4 = A4 __attribute__((matrix_type(4, 4)));
+using X4 = B4;
+using Y4 = B4;
+
+using X5 = A4 __attribute__((matrix_type(3, 4)));
+using Y5 = A4 __attribute__((matrix_type(4, 3)));
+
+N t1 = 0 ? X1() : Y1(); // expected-error {{rvalue of type 'B1'}}
+N t2 = 0 ? X2() : Y2(); // expected-error {{rvalue of type 'B2'}}
+
+const X1 &xt3 = 0;
+const Y1 &yt3 = 0;
+N t3 = 0 ? xt3 : yt3; // expected-error {{lvalue of type 'const B1'}}
+
+N t4 = X3() + Y3();   // expected-error {{rvalue of type 'B3'}}
+
+N t5 = A3() ? X3() : Y3(); // expected-error {{rvalue of type 'B3'}}
+N t6 = A3() ? X1() : Y1(); // expected-error {{vector condition type 'A3' (vector of 4 'char' values) and result type '__attribute__((__vector_size__(4 * sizeof(B1)))) B1' (vector of 4 'B1' values) do not have elements of the same size}}
+
+N t7 = X4() + Y4(); // expected-error {{rvalue of type 'B4'}}
+N t8 = X4() * Y4(); // expected-error {{rvalue of type 'B4'}}
+N t9 = X5() * Y5(); // expected-error {{rvalue of type 'A4 __attribute__((matrix_type(3, 3)))'}}

clang/test/SemaCXX/sugared-auto.cpp

 
 using fp3 = SARS (*)() throw(Man);
 using fp4 = Ebola (*)() throw(Vibrio);
 auto t7(fp3 a, fp4 b) {
   if (false)
     return true ? a : b;
   if (false)
     return a;
-  return N(); // expected-error {{but deduced as 'SARS (*)() throw(Man, Vibrio)' (aka 'void (*)() throw(Man, Vibrio)')}}
+  return N(); // expected-error {{but deduced as 'Virus (*)() throw(Man, Vibrio)' (aka 'void (*)() throw(Man, Vibrio)')}}
 }
 
 using fp5 = void (*)(const Man);
 using fp6 = void (*)(Dog);
 TEST_AUTO(t8, fp5, fp6); // expected-error {{but deduced as 'void (*)(const Animal)' (aka 'void (*)(const int)')}}
 
 using fp6 = void (*)(ConstMan);
 using fp7 = void (*)(ConstDog);