Here's a small cleanup discovered by modules. The alias instantiation
machinery would setup template_info, and then sometime later overwrite
that with equivalent info. The template info, once set, is logically
immutable. Let's just not do that.
pushed to master
nathan
--
Nathan Sidwell
2020-05-19 Nathan Sidwell <nat...@acm.org>
* pt.c (lookup_template_class_1): Do not reinit template_info of an
alias here.
diff --git i/gcc/cp/pt.c w/gcc/cp/pt.c
index 4d9651acee6..c17a038c6d0 100644
--- i/gcc/cp/pt.c
+++ w/gcc/cp/pt.c
@@ -10062,8 +10062,21 @@ lookup_template_class_1 (tree d1, tree arglist, tree in_decl, tree context,
}
}
- // Build template info for the new specialization.
- SET_TYPE_TEMPLATE_INFO (t, build_template_info (found, arglist));
+ /* Build template info for the new specialization. This can
+ overwrite the existing TEMPLATE_INFO for T (that points to
+ its instantiated TEMPLATE_DECL), with this one that points to
+ the most general template, but that's what we want. */
+
+ if (TYPE_ALIAS_P (t))
+ {
+ /* This should already have been constructed during
+ instantiation of the alias decl. */
+ tree ti = DECL_TEMPLATE_INFO (TYPE_NAME (t));
+ gcc_checking_assert (template_args_equal (TI_ARGS (ti), arglist)
+ && TI_TEMPLATE (ti) == found);
+ }
+ else
+ SET_TYPE_TEMPLATE_INFO (t, build_template_info (found, arglist));
elt.spec = t;
slot = type_specializations->find_slot_with_hash (&elt, hash, INSERT);
