Align ARM Neon blocks to 16 bytes

When available, OCB algorithms use ARM Neon instructions to efficiently
move and manipulate data. However, ARM requires the data used by these
instructions to be aligned to 16 bytes. Executing Neon instructions
without proper alignment causes an alignment fault - which is fatal.

Add aligned keyword and use memalign to force 16 byte alignment where
required.

Change-Id: I0d972d77d521fd57007771a361910d467a821c2c
1 file changed