Add the logic to deal with input sizes that are not a round multiple
of the AES block size, as described by the XTS spec. This brings the
SPE implementation in line with other kernel drivers that have been
updated recently to take this into account.

Cc: Eric Biggers <ebiggers@google.com>
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>