[Bug middle-end/51071] [4.7 Regression] ICE in gimple_has_side_effects, at gimple.c:2513

["rguenth at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51071

Richard Guenther <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2011-11-10
         AssignedTo|unassigned at gcc dot       |rguenth at gcc dot gnu.org
                   |gnu.org                     |
   Target Milestone|---                         |4.7.0
     Ever Confirmed|0                           |1

--- Comment #1 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-11-10 10:51:51 UTC ---
That asserts are sort-of bogus (I'd wanted to get rid of them at some point).

We have:

(gdb) call debug_gimple_stmt (s)
__here:
(gdb) call debug_tree (op)
 <label_decl 0x7ffff5b55c00 __here
    type <void_type 0x7ffff5a34bd0 void VOID
        align 8 symtab 0 alias set -1 canonical type 0x7ffff5a34bd0
        pointer_to_this <pointer_type 0x7ffff5a34c78>>
    side-effects addressable used VOID file t2.c line 10 col 10
    align 1 context <function_decl 0x7ffff5b39c00 compress_file_range> initial
<error_mark 0x7ffff5a29c60> abstract_origin <label_decl 0x7ffff5b55080 __here>>

TREE_SIDE_EFFECTS on a LABEL_DECL is FORCED_LABEL ...

The loops in gimple_has_side_effects are never triggered unless they
hit the assert (if they ever cause us to return true we'd have returned
true already because of the gimple_has_volatile_ops check).

Mine.