The LO_ARG define needs to consider endianness also for 32 bit builds.

The "bpf_fancy" test case didn't work on s390 in 32 bit and compat mode
because the LO_ARG define resulted in a BPF program which read the upper
halve of the 64 bit system call arguments instead of the lower halves.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Acked-by: Kees Cook <keescook@chromium.org>
Signed-off-by: James Morris <james.l.morris@oracle.com>