================
@@ -1668,6 +1656,25 @@ static bool checkCommonAttributeFeatures(Sema &S, const
Ty *Node,
return false;
}
+template <typename Ty>
+static bool checkCommonAttributeFeatures(Sema &S, const Ty *Node,
+ const ParsedAttr &A,
+ bool SkipArgCountCheck) {
+ // Run the checks shared with type attributes (language options, target, and
+ // argument count). This also bails on unknown and ignored attributes.
+ if (checkCommonAttributeFeatures(S, A, SkipArgCountCheck))
+ return true;
+ // Check whether the attribute appertains to the given subject.
+ if (!A.diagnoseAppertainsTo(S, Node))
+ return true;
+ // Check whether the attribute is mutually exclusive with other attributes
+ // that have already been applied to the declaration.
+ if (!A.diagnoseMutualExclusion(S, Node))
+ return true;
----------------
elizabethandrews wrote:
> The type is essentially mid-construction when we process attributes. I am not
> sure how that affects appertainment checks. For example, does
> SubjectList<[Pointer]> mean the chunk that the attribute is applied to is a
> pointer, or the overall type being qualified is a pointer?
I might be overthinking this. I just remembered we have individual checks
inside handlers for appertainment for some handlers now. So we can probably
just use whatever `Type` (or `QualType` ?) is at the site.
https://github.com/llvm/llvm-project/pull/227421
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
