Hi,
in this ICE on invalid regression the problem happens toward the end of
lookup_destructor when adjust_result_of_qualified_name_lookup (and
possibly the callar itself) tries to set BASELINK_QUALIFIED_P of an
error_mark_node. As explained in check_dtor_name too, error_mark_node
means we already complained about this destructor, thus seems safe to
early return it back. Tested x86_64-linux.
Thanks,
Paolo.
/////////////////////
/cp
2014-01-30 Paolo Carlini <paolo.carl...@oracle.com>
PR c++/58843
* typeck.c (lookup_destructor): Check dtor_type for error_mark_node.
/testsuite
2014-01-30 Paolo Carlini <paolo.carl...@oracle.com>
PR c++/58843
* g++.dg/lookup/crash8.C: New.
Index: cp/typeck.c
===================================================================
--- cp/typeck.c (revision 207298)
+++ cp/typeck.c (working copy)
@@ -2486,6 +2486,10 @@ lookup_destructor (tree object, tree scope, tree d
tree dtor_type = TREE_OPERAND (dtor_name, 0);
tree expr;
+ /* We've already complained about this destructor. */
+ if (dtor_type == error_mark_node)
+ return error_mark_node;
+
if (scope && !check_dtor_name (scope, dtor_type))
{
if (complain & tf_error)
Index: testsuite/g++.dg/lookup/crash8.C
===================================================================
--- testsuite/g++.dg/lookup/crash8.C (revision 0)
+++ testsuite/g++.dg/lookup/crash8.C (working copy)
@@ -0,0 +1,13 @@
+// PR c++/58843
+
+struct A {};
+
+template<typename T> void foo(T t)
+{
+ t.T::~X(); // { dg-error "no type" }
+}
+
+void bar()
+{
+ foo(A());
+}
