================
@@ -98,3 +98,6 @@ enum E8 { A8 = -128, B8 = 127 } x8;
// CHECK-NOT: DIFlagEnumClass
// CHECK: !DIEnumerator(name: "A8", value: -128)
+// Forward declaration of an enum class.
+enum class Color : int;
+// CHECK-NOT: !DICompositeType(tag: DW_TAG_enumeration_type, name: "Color"
----------------
tahonermann wrote:
I spent some time experimenting and debugging today but have not been able to
find a way for `CGDebugInfo::CreateTypeDefinition(const EnumType *)` to be
invoked without an available definition. I tried various uses of non-standard
enumeration forward declarations, but non-suppressible errors were issued for
every case I tried that I thought might lead to generation of a debug info type.
The most interesting scenario I played with involved use of the enumeration
type within its own definition. In that context, the type is incomplete, but a
definition is (being made) available. I wouldn't expect generation of debug
info to ever be sensitive to such a context though. I know little about how
debug information is generated, but seeing as the function under discussion is
part of code generation, I wouldn't expect a not-yet-completed definition to
ever be observed.
I think adding an assert as proposed in
https://github.com/llvm/llvm-project/pull/105556 suffices to address the static
analysis concern.
https://github.com/llvm/llvm-project/pull/97105
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
