[clang][Interp] PointerToIntegral casts

[tbaeder]

Summary

Add a new opcode doing the cast and emit a diagnostic in case we're in a constant context.

[tbaeder]

Hm, this works for pointers, but not for function pointers (which we map to FunctionPointer).

[shafik]

LGTM

[phosek]

This test is failing on our Windows builders with the following error:

Script:
--
: 'RUN: at line 1';   c:\b\s\w\ir\x\w\llvm_build\bin\clang.exe -cc1 -internal-isystem c:\b\s\w\ir\x\w\llvm_build\lib\clang\18\include -nostdsysteminc -fexperimental-new-constant-interpreter -fms-extensions -std=c++11 -verify C:\b\s\w\ir\x\w\llvm-llvm-project\clang\test\AST\Interp\literals.cpp
: 'RUN: at line 2';   c:\b\s\w\ir\x\w\llvm_build\bin\clang.exe -cc1 -internal-isystem c:\b\s\w\ir\x\w\llvm_build\lib\clang\18\include -nostdsysteminc -fexperimental-new-constant-interpreter -fms-extensions -std=c++20 -verify C:\b\s\w\ir\x\w\llvm-llvm-project\clang\test\AST\Interp\literals.cpp
: 'RUN: at line 3';   c:\b\s\w\ir\x\w\llvm_build\bin\clang.exe -cc1 -internal-isystem c:\b\s\w\ir\x\w\llvm_build\lib\clang\18\include -nostdsysteminc -std=c++11 -fms-extensions -verify=ref C:\b\s\w\ir\x\w\llvm-llvm-project\clang\test\AST\Interp\literals.cpp
: 'RUN: at line 4';   c:\b\s\w\ir\x\w\llvm_build\bin\clang.exe -cc1 -internal-isystem c:\b\s\w\ir\x\w\llvm_build\lib\clang\18\include -nostdsysteminc -std=c++20 -fms-extensions -verify=ref C:\b\s\w\ir\x\w\llvm-llvm-project\clang\test\AST\Interp\literals.cpp
--
Exit Code: 1

Command Output (stdout):
--
$ ":" "RUN: at line 1"
$ "c:\b\s\w\ir\x\w\llvm_build\bin\clang.exe" "-cc1" "-internal-isystem" "c:\b\s\w\ir\x\w\llvm_build\lib\clang\18\include" "-nostdsysteminc" "-fexperimental-new-constant-interpreter" "-fms-extensions" "-std=c++11" "-verify" "C:\b\s\w\ir\x\w\llvm-llvm-project\clang\test\AST\Interp\literals.cpp"
# command stderr:
error: 'warning' diagnostics seen but not expected: 
  File C:\b\s\w\ir\x\w\llvm-llvm-project\clang\test\AST\Interp\literals.cpp Line 946: cast to smaller integer type 'long' from 'char *'
1 error generated.

error: command failed with exit status: 1

Would it be possible to either quickly address the issue or revert the change?

[aaron.ballman]

In D150946#4536971, @phosek wrote:

This test is failing on our Windows builders with the following error:

Script:
--
: 'RUN: at line 1';   c:\b\s\w\ir\x\w\llvm_build\bin\clang.exe -cc1 -internal-isystem c:\b\s\w\ir\x\w\llvm_build\lib\clang\18\include -nostdsysteminc -fexperimental-new-constant-interpreter -fms-extensions -std=c++11 -verify C:\b\s\w\ir\x\w\llvm-llvm-project\clang\test\AST\Interp\literals.cpp
: 'RUN: at line 2';   c:\b\s\w\ir\x\w\llvm_build\bin\clang.exe -cc1 -internal-isystem c:\b\s\w\ir\x\w\llvm_build\lib\clang\18\include -nostdsysteminc -fexperimental-new-constant-interpreter -fms-extensions -std=c++20 -verify C:\b\s\w\ir\x\w\llvm-llvm-project\clang\test\AST\Interp\literals.cpp
: 'RUN: at line 3';   c:\b\s\w\ir\x\w\llvm_build\bin\clang.exe -cc1 -internal-isystem c:\b\s\w\ir\x\w\llvm_build\lib\clang\18\include -nostdsysteminc -std=c++11 -fms-extensions -verify=ref C:\b\s\w\ir\x\w\llvm-llvm-project\clang\test\AST\Interp\literals.cpp
: 'RUN: at line 4';   c:\b\s\w\ir\x\w\llvm_build\bin\clang.exe -cc1 -internal-isystem c:\b\s\w\ir\x\w\llvm_build\lib\clang\18\include -nostdsysteminc -std=c++20 -fms-extensions -verify=ref C:\b\s\w\ir\x\w\llvm-llvm-project\clang\test\AST\Interp\literals.cpp
--
Exit Code: 1

Command Output (stdout):
--
$ ":" "RUN: at line 1"
$ "c:\b\s\w\ir\x\w\llvm_build\bin\clang.exe" "-cc1" "-internal-isystem" "c:\b\s\w\ir\x\w\llvm_build\lib\clang\18\include" "-nostdsysteminc" "-fexperimental-new-constant-interpreter" "-fms-extensions" "-std=c++11" "-verify" "C:\b\s\w\ir\x\w\llvm-llvm-project\clang\test\AST\Interp\literals.cpp"
# command stderr:
error: 'warning' diagnostics seen but not expected: 
  File C:\b\s\w\ir\x\w\llvm-llvm-project\clang\test\AST\Interp\literals.cpp Line 946: cast to smaller integer type 'long' from 'char *'
1 error generated.

error: command failed with exit status: 1

Would it be possible to either quickly address the issue or revert the change?

Did the changes in 2a6cfc4baa784ff299a81f8c5e7163207c93eafe address the issue for you?

[phosek]

In D150946#4538469, @aaron.ballman wrote:

In D150946#4536971, @phosek wrote:

This test is failing on our Windows builders with the following error:

Script:
--
: 'RUN: at line 1';   c:\b\s\w\ir\x\w\llvm_build\bin\clang.exe -cc1 -internal-isystem c:\b\s\w\ir\x\w\llvm_build\lib\clang\18\include -nostdsysteminc -fexperimental-new-constant-interpreter -fms-extensions -std=c++11 -verify C:\b\s\w\ir\x\w\llvm-llvm-project\clang\test\AST\Interp\literals.cpp
: 'RUN: at line 2';   c:\b\s\w\ir\x\w\llvm_build\bin\clang.exe -cc1 -internal-isystem c:\b\s\w\ir\x\w\llvm_build\lib\clang\18\include -nostdsysteminc -fexperimental-new-constant-interpreter -fms-extensions -std=c++20 -verify C:\b\s\w\ir\x\w\llvm-llvm-project\clang\test\AST\Interp\literals.cpp
: 'RUN: at line 3';   c:\b\s\w\ir\x\w\llvm_build\bin\clang.exe -cc1 -internal-isystem c:\b\s\w\ir\x\w\llvm_build\lib\clang\18\include -nostdsysteminc -std=c++11 -fms-extensions -verify=ref C:\b\s\w\ir\x\w\llvm-llvm-project\clang\test\AST\Interp\literals.cpp
: 'RUN: at line 4';   c:\b\s\w\ir\x\w\llvm_build\bin\clang.exe -cc1 -internal-isystem c:\b\s\w\ir\x\w\llvm_build\lib\clang\18\include -nostdsysteminc -std=c++20 -fms-extensions -verify=ref C:\b\s\w\ir\x\w\llvm-llvm-project\clang\test\AST\Interp\literals.cpp
--
Exit Code: 1

Command Output (stdout):
--
$ ":" "RUN: at line 1"
$ "c:\b\s\w\ir\x\w\llvm_build\bin\clang.exe" "-cc1" "-internal-isystem" "c:\b\s\w\ir\x\w\llvm_build\lib\clang\18\include" "-nostdsysteminc" "-fexperimental-new-constant-interpreter" "-fms-extensions" "-std=c++11" "-verify" "C:\b\s\w\ir\x\w\llvm-llvm-project\clang\test\AST\Interp\literals.cpp"
# command stderr:
error: 'warning' diagnostics seen but not expected: 
  File C:\b\s\w\ir\x\w\llvm-llvm-project\clang\test\AST\Interp\literals.cpp Line 946: cast to smaller integer type 'long' from 'char *'
1 error generated.

error: command failed with exit status: 1

Would it be possible to either quickly address the issue or revert the change?

Did the changes in 2a6cfc4baa784ff299a81f8c5e7163207c93eafe address the issue for you?

That fixed it, thank you.

clang/lib/AST/Interp/ByteCodeExprGen.cpp

case CK_FloatingToIntegral: {
     return this->emitCastFloatingIntegral(*ToT, CE);
   }
 
   case CK_NullToPointer:
     if (DiscardResult)
       return true;
     return this->emitNull(classifyPrim(CE->getType()), CE);
 
+  case CK_PointerToIntegral: {
+    // TODO: Discard handling.
+    if (!this->visit(SubExpr))
+      return false;
+
+    PrimType T = classifyPrim(CE->getType());
+    return this->emitCastPointerIntegral(T, CE);
+  }
+
   case CK_ArrayToPointerDecay:
   case CK_AtomicToNonAtomic:
   case CK_ConstructorConversion:
   case CK_FunctionToPointerDecay:
   case CK_NonAtomicToAtomic:
   case CK_NoOp:
   case CK_UserDefinedConversion:
     return this->visit(SubExpr);

clang/lib/AST/Interp/Interp.h

 bool CheckThis(InterpState &S, CodePtr OpPC, const Pointer &This);
 
 /// Checks if a method is pure virtual.
 bool CheckPure(InterpState &S, CodePtr OpPC, const CXXMethodDecl *MD);
 
 /// Checks that all fields are initialized after a constructor call.
 bool CheckCtorCall(InterpState &S, CodePtr OpPC, const Pointer &This);
 
+/// Checks if reinterpret casts are legal in the current context.
+bool CheckPotentialReinterpretCast(InterpState &S, CodePtr OpPC,
+                                   const Pointer &Ptr);
+
 /// Checks if the shift operation is legal.
 template <typename LT, typename RT>
 bool CheckShift(InterpState &S, CodePtr OpPC, const LT &LHS, const RT &RHS,
                 unsigned Bits) {
   if (RHS.isNegative()) {
     const SourceInfo &Loc = S.Current->getSource(OpPC);
     S.CCEDiag(Loc, diag::note_constexpr_negative_shift) << RHS.toAPSInt();
     return false;
if ((Status & APFloat::opStatus::opInvalidOp) && F.isFinite()) {
       return S.noteUndefinedBehavior();
     }
 
     S.Stk.push<T>(T(Result));
     return CheckFloatResult(S, OpPC, Status);
   }
 }
 
+template <PrimType Name, class T = typename PrimConv<Name>::T>
+bool CastPointerIntegral(InterpState &S, CodePtr OpPC) {
+  const Pointer &Ptr = S.Stk.pop<Pointer>();
+
+  if (!CheckPotentialReinterpretCast(S, OpPC, Ptr))
+    return false;
+
+  S.Stk.push<T>(T::from(Ptr.getIntegerRepresentation()));
+  return true;
+}
+
 //===----------------------------------------------------------------------===//
 // Zero, Nullptr
 //===----------------------------------------------------------------------===//
 
 template <PrimType Name, class T = typename PrimConv<Name>::T>
 bool Zero(InterpState &S, CodePtr OpPC) {
   S.Stk.push<T>(T::zero());
   return true;

clang/lib/AST/Interp/Interp.cpp

bool CheckCtorCall(InterpState &S, CodePtr OpPC, const Pointer &This) {
   assert(!This.isZero());
   if (const Record *R = This.getRecord())
     return CheckFieldsInitialized(S, OpPC, This, R);
   const auto *CAT =
       cast<ConstantArrayType>(This.getType()->getAsArrayTypeUnsafe());
   return CheckArrayInitialized(S, OpPC, This, CAT);
 }
 
+bool CheckPotentialReinterpretCast(InterpState &S, CodePtr OpPC,
+                                   const Pointer &Ptr) {
+  if (!S.inConstantContext())
+    return true;
+
+  const SourceInfo &E = S.Current->getSource(OpPC);
+  S.CCEDiag(E, diag::note_constexpr_invalid_cast)
+      << 2 << S.getLangOpts().CPlusPlus;
+  return false;
+}
+
 bool CheckFloatResult(InterpState &S, CodePtr OpPC, APFloat::opStatus Status) {
   // In a constant context, assume that any dynamic rounding mode or FP
   // exception state matches the default floating-point environment.
   if (S.inConstantContext())
     return true;
 
   const SourceInfo &E = S.Current->getSource(OpPC);
   FPOptions FPO = E.asExpr()->getFPFeaturesInEffect(S.Ctx.getLangOpts());

clang/lib/AST/Interp/Opcodes.td

 
 // Cast a floating to an integer type
 def CastFloatingIntegral : Opcode {
   let Types = [AluTypeClass];
   let Args = [];
   let HasGroup = 1;
 }
 
+def CastPointerIntegral : Opcode {
+  let Types = [AluTypeClass];
+  let Args = [];
+  let HasGroup = 1;
+}
+
 //===----------------------------------------------------------------------===//
 // Comparison opcodes.
 //===----------------------------------------------------------------------===//
 
 class EqualityOpcode : Opcode {
   let Types = [AllTypeClass];
   let HasGroup = 1;
 }

clang/lib/AST/Interp/Pointer.h

public:
   ~Pointer();
 
   void operator=(const Pointer &P);
   void operator=(Pointer &&P);
 
   /// Converts the pointer to an APValue.
   APValue toAPValue() const;
 
+  /// Converts the pointer to a string usable in diagnostics.
+  std::string toDiagnosticString(const ASTContext &Ctx) const;
+
+  unsigned getIntegerRepresentation() const {
+    return reinterpret_cast<uintptr_t>(Pointee) + Offset;
+  }
+
   /// Offsets a pointer inside an array.
   Pointer atIndex(unsigned Idx) const {
     if (Base == RootPtrMark)
       return Pointer(Pointee, RootPtrMark, getDeclDesc()->getSize());
     unsigned Off = Idx * elemSize();
     if (getFieldDesc()->ElemDesc)
       Off += sizeof(InlineDescriptor);
     else

clang/test/AST/Interp/literals.cpp

 // RUN: %clang_cc1 -fexperimental-new-constant-interpreter -fms-extensions -std=c++11 -verify %s
 // RUN: %clang_cc1 -fexperimental-new-constant-interpreter -fms-extensions -std=c++20 -verify %s
 // RUN: %clang_cc1 -std=c++11 -fms-extensions -verify=ref %s
 // RUN: %clang_cc1 -std=c++20 -fms-extensions -verify=ref %s
 
 #define INT_MIN (~__INT_MAX__)
 #define INT_MAX __INT_MAX__
 
+typedef __INTPTR_TYPE__ intptr_t;
+
 
 static_assert(true, "");
 static_assert(false, ""); // expected-error{{failed}} ref-error{{failed}}
 static_assert(nullptr == nullptr, "");
 static_assert(1 == 1, "");
 static_assert(1 == 3, ""); // expected-error{{failed}} ref-error{{failed}}
 
 constexpr int number = 10;
constexpr int a() {
     (void)noexcept(++b); // expected-warning {{expression with side effects has no effect in an unevaluated context}} \
                          // ref-warning {{expression with side effects has no effect in an unevaluated context}}
 
     return b;
   }
   static_assert(a() == 0, "");
 #endif
 }
+
+namespace PointerCasts {
+  constexpr int M = 10;
+  constexpr const int *P = &M;
+  constexpr intptr_t A = (intptr_t)P; // ref-error {{must be initialized by a constant expression}} \
+                                      // ref-note {{cast that performs the conversions of a reinterpret_cast}} \
+                                      // expected-error {{must be initialized by a constant expression}} \
+                                      // expected-note {{cast that performs the conversions of a reinterpret_cast}}
+
+  int array[(long)(char*)0]; // ref-warning {{variable length array folded to constant array}} \
+                             // expected-warning {{variable length array folded to constant array}}
+}