[Bug c++/41144] [4.5 Regression] ice for legal code with -O2 in get_alias_set

[rguenth at gcc dot gnu dot org]

------- Comment #4 from rguenth at gcc dot gnu dot org  2009-08-24 12:53 -------
We're layouting an array with incomplete element type and get into

        if (!TYPE_SIZE (element))
          /* We don't know the size of the underlying element type, so
             our alignment calculations will be wrong, forcing us to
             fall back on structural equality. */
          SET_TYPE_STRUCTURAL_EQUALITY (type);

later re-using this type when building the same array type with complete
element type as the canonical variant.

IMHO the above is simply bogus.

To test patch:

Index: stor-layout.c
===================================================================
--- stor-layout.c       (revision 151050)
+++ stor-layout.c       (working copy)
@@ -2020,11 +2020,6 @@ layout_type (tree type)
 #else
        TYPE_ALIGN (type) = MAX (TYPE_ALIGN (element), BITS_PER_UNIT);
 #endif
-       if (!TYPE_SIZE (element))
-         /* We don't know the size of the underlying element type, so
-            our alignment calculations will be wrong, forcing us to
-            fall back on structural equality. */
-         SET_TYPE_STRUCTURAL_EQUALITY (type);
        TYPE_USER_ALIGN (type) = TYPE_USER_ALIGN (element);
        SET_TYPE_MODE (type, BLKmode);
        if (TYPE_SIZE (type) != 0


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=41144