Hi!
A recent change to check_return_expr resulted in
maybe_warn_about_returning_address_of_local being called also with
processing_template_decl. The problem with that is that the function
relies on folding (fold_for_warn) which isn't performed at all when
processing_template_decl. So, we have still ARRAY_REF for a local decl
that has pointer type, rather than the expected POINTER_PLUS_EXPR,
where ARRAY_REF would only remain if the variable was actually an array.
The following patch fixes that by not calling the function at all
when processing_template_decl like before.
Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
2017-10-19 Jakub Jelinek <ja...@redhat.com>
PR c++/82600
* typeck.c (check_return_expr): Don't call
maybe_warn_about_returning_address_of_local in templates.
* g++.dg/warn/Wreturn-local-addr-4.C: New test.
--- gcc/cp/typeck.c.jj 2017-10-10 22:04:05.000000000 +0200
+++ gcc/cp/typeck.c 2017-10-18 12:17:29.282963388 +0200
@@ -9228,7 +9228,8 @@ check_return_expr (tree retval, bool *no
&& TREE_CODE (TREE_OPERAND (retval, 1)) == AGGR_INIT_EXPR)
retval = build2 (COMPOUND_EXPR, TREE_TYPE (retval), retval,
TREE_OPERAND (retval, 0));
- else if (maybe_warn_about_returning_address_of_local (retval))
+ else if (!processing_template_decl
+ && maybe_warn_about_returning_address_of_local (retval))
retval = build2 (COMPOUND_EXPR, TREE_TYPE (retval), retval,
build_zero_cst (TREE_TYPE (retval)));
}
--- gcc/testsuite/g++.dg/warn/Wreturn-local-addr-4.C.jj 2017-10-18
12:21:08.569273452 +0200
+++ gcc/testsuite/g++.dg/warn/Wreturn-local-addr-4.C 2017-10-18
12:16:28.000000000 +0200
@@ -0,0 +1,18 @@
+// PR c++/82600
+// { dg-do compile }
+
+void *b[10];
+
+template <int N>
+void **
+foo (int x)
+{
+ void **a = b; // { dg-bogus "address of local variable 'a'
returned" }
+ return &a[x];
+}
+
+void **
+bar (int x)
+{
+ return foo <0> (x);
+}
Jakub
