https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101350
--- Comment #2 from kargl at gcc dot gnu.org ---
(In reply to kargl from comment #1)
> diff --git a/gcc/fortran/resolve.c b/gcc/fortran/resolve.c
> index 45c3ad387ac..897cfde3ed3 100644
> --- a/gcc/fortran/resolve.c
> +++ b/gcc/fortran/resolve.c
> @@ -15307,6 +15308,15 @@ resolve_symbol (gfc_symbol *sym)
> return;
> sym->resolve_symbol_called = 1;
>
> + /* Symbol cannot have an external attribute and be a contained procedure.
> */
> + if (sym->attr.external && sym->attr.contained)
> + {
> + gfc_error ("Contained procedure %qs at %L conflicts with "
> + "a previously declared EXTERNAL procedure",
> + sym->name, &sym->declared_at);
> + return;
> + }
> +
> /* No symbol will ever have union type; only components can be unions.
> Union type declaration symbols have type BT_UNKNOWN but flavor FL_UNION
> (just like derived type declaration symbols have flavor FL_DERIVED). */
This patch works to fix the reported problem. It, however, causes
a number regressions, which can be traced to confusing and conflicting
settings of attributes. In particular, attr.external=1 is often set
when attr.proc=PROC_MODULE or attr.module_procedure=1 occur. This
should not happen as an external procedure can never be a module
procedure.
