Hi,
in this error-recovery regression we ICE after a sensible diagnostic
emitted by cp_build_unary_op, called by finish_unary_op_expr via
build_x_unary_op. In principle we could dig deeper, but I don't think it
makes sense for finish_unary_op_expr to go on having seen the
error_mark_node returned by build_x_unary_op given that the purpose of
its second half is only issuing warnings.
Tested x86_64-linux.
Thanks, Paolo.
//////////////////////
/cp
2018-09-25 Paolo Carlini <paolo.carl...@oracle.com>
PR c++/84940
* semantics.c (finish_unary_op_expr): Check return value of
build_x_unary_op for error_mark_node.
/testsuite
2018-09-25 Paolo Carlini <paolo.carl...@oracle.com>
PR c++/84940
* g++.dg/expr/unary4.C: New.
Index: cp/semantics.c
===================================================================
--- cp/semantics.c (revision 264578)
+++ cp/semantics.c (working copy)
@@ -2727,13 +2727,14 @@ finish_unary_op_expr (location_t op_loc, enum tree
/* TODO: build_x_unary_op doesn't always honor the location. */
result.set_location (combined_loc);
- tree result_ovl, expr_ovl;
+ if (result == error_mark_node)
+ return result;
if (!(complain & tf_warning))
return result;
- result_ovl = result;
- expr_ovl = expr;
+ tree result_ovl = result;
+ tree expr_ovl = expr;
if (!processing_template_decl)
expr_ovl = cp_fully_fold (expr_ovl);
Index: testsuite/g++.dg/expr/unary4.C
===================================================================
--- testsuite/g++.dg/expr/unary4.C (nonexistent)
+++ testsuite/g++.dg/expr/unary4.C (working copy)
@@ -0,0 +1,8 @@
+// PR c++/84940
+// { dg-additional-options -Wno-vla }
+
+void
+foo (int x)
+{
+ struct {} a[1][x](-a[0]); // { dg-error "wrong type argument to unary minus"
}
+}