[Bug middle-end/39954] [4.5 Regression] Unaligned access in gcc.dg/torture/pr26565.c

[rguenth at gcc dot gnu dot org]

------- Comment #3 from rguenth at gcc dot gnu dot org  2009-04-28 22:04 -------
Indeed.  We likely fold the memcpy, but that should better preserve the
known alignment.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39954