[clang] [clang] Add clang::debug_info_type attribute for bitfields (PR #69104)

[Erich Keane via cfe-commits]

================
@@ -5910,6 +5910,30 @@ static void handleBuiltinAliasAttr(Sema &S, Decl *D,
   D->addAttr(::new (S.Context) BuiltinAliasAttr(S.Context, AL, Ident));
 }
 
+static void handleDebugInfoTypeAttr(Sema &S, Decl *D, const ParsedAttr &AL) {
+  if (!AL.hasParsedType()) {
+    S.Diag(AL.getLoc(), diag::err_attribute_wrong_number_arguments) << AL << 1;
+    return;
+  }
+
+  TypeSourceInfo *ParmTSI = nullptr;
+  QualType type = S.GetTypeFromParser(AL.getTypeArg(), &ParmTSI);
+  assert(ParmTSI && "no type source info for attribute argument");
+
+  if (type->isEnumeralType()) {
+    QualType BitfieldType = llvm::cast<FieldDecl>(D)->getType();
+    QualType EnumUnderlyingType =
+        type->getAs<EnumType>()->getDecl()->getIntegerType();
+    if (EnumUnderlyingType != BitfieldType) {
----------------
erichkeane wrote:

> A few things -- I think we should look at the canonical type so that we don't run into issues with typedefs. e.g.,
> 
> ```
> enum E { Zero, One };
> typedef int Foo;
> struct S {
>   [[clang::debug_info_type(E) Foo field : 1;
> };
> ```
> 
> where the enum's underlying type is `int` but the bit-field's type is a typedef.
> 
> Another case we should consider would be whether we want to allow signed unsigned mismatches, as in:
> 
> ```
> enum E { Invalid = -1, ValidValue, OtherValidValue, YetAnotherValidValue };
> struct S {
>   [[clang::debug_info_type(E) unsigned field : 3;
> };
> ```
> 
> and only prevent situations where the specified type is not an enumeration or integral type.

This ALSO makes me wonder if forcing it to be an enum type is necessary.  I can see potential value of a "OpaqueValueOfStruct" storage type thing, that perhaps we should just 'trust' that the user is doing something sensible with it.

  I think in the signed/unsigned mismatch, we should allow it (thanks to the layout rules).  That is, consider:

```
enum SIGNED {Invalid = -1, Valid };
enum UNSIGNED {Valid, INVALID = 0xF };

struct S {
  [[attr_name(SIGNED)]]
  unsigned SignedThing : 4;
  [[attr_name(UNSIGNED)]]
  unsigned UnsignedThing : 4;
};
```
They HAVE to be the same type else they don't get packed into the same byte, but you have 1 enum that can represent a negative value.


https://github.com/llvm/llvm-project/pull/69104