* do a "get_unaligned()" if this helps and is sufficiently
* fast.
*
- * - Little-endian machines (so that we can generate the mask
- * of low bytes efficiently). Again, we *could* do a byte
- * swapping load on big-endian architectures if that is not
- * expensive enough to make the optimization worthless.
- *
* - non-CONFIG_DEBUG_PAGEALLOC configurations (so that we
* do not trap on the (extremely unlikely) case of a page
* crossing operation.
if (!len)
goto done;
}
- mask = ~(~0ul << len*8);
+ mask = bytemask_from_count(len);
hash += mask & a;
done:
return fold_hash(hash);