This should massively improve performance for (at least) MIPS targets:

    * poly1305: optimize unaligned access

    This is a very appreciated fix from René van Dorst, adjusting the
    arithmetic in Poly1305 to work fast on platforms with slow unaligned
    access, such as MIPS. According to his calculation, this gives a 50%
    improvement on small MIPS boxes.

Signed-off-by: Baptiste Jonglez <git@bitsofnetworks.org>