------- Comment #1 from rguenth at gcc dot gnu dot org 2009-11-20 15:33 -------
Confirmed. Sth as simple as
Index: gcc/expr.c
===================================================================
--- gcc/expr.c (revision 154364)
+++ gcc/expr.c (working copy)
@@ -6834,7 +6834,8 @@ expand_expr_addr_expr_1 (tree exp, rtx t
/* ??? This should be considered a front-end bug. We should not be
generating ADDR_EXPR of something that isn't an LVALUE. The only
exception here is STRING_CST. */
- if (CONSTANT_CLASS_P (exp))
+ if (CONSTANT_CLASS_P (exp)
+ || (TREE_CODE (exp) == ADDR_EXPR && TREE_CONSTANT (exp)))
return XEXP (expand_expr_constant (exp, 0, modifier), 0);
/* Everything must be something allowed by is_gimple_addressable. */
might fix it, though the recursion in this function for CONST_DECLs
make it a bit convoluted to call it obvious.
--
rguenth at gcc dot gnu dot org changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Component|fortran |middle-end
Ever Confirmed|0 |1
Keywords| |ice-on-valid-code
Known to fail| |4.5.0
Last reconfirmed|0000-00-00 00:00:00 |2009-11-20 15:33:46
date| |
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42119
