Unaligned block moves and MEM_ALIGN re-broken

[law@redhat.com]

In message <10201251226.AA10447@vlsi1.ultra.nyu.edu>, Richard Kenner writes:
 >     You previously said that DImode was the right mode
 >     (http://gcc.gnu.org/ml/gcc/2001-11/msg00399.html) because "GCC should
 >     choose an integral mode for a struct as often as it can since that will
 >     allow objects of that type to be put in registers."
 >     (http://gcc.gnu.org/ml/gcc/2001-11/msg00381.html)
 > 
 > Right.  But it can only put it in registers if the alignment permits it.
 > 
 >     DImode represents the size of the object, not the alignment. 
 > 
 > No, it represents *both*.  It refers to a 64-bit object whose
 > alignment is MIN (64, BIGGEST_ALIGNMENT).
It seems to me this only true on a STRICT_ALIGNMENT target.
jeff