On Fri, Mar 31, 2017 at 11:27:03AM +0200, Ondrej Mosnacek wrote:
> The gf128mul_x_ble function is currently defined in gf128mul.c, because
> it depends on the gf128mul_table_be multiplication table.
>
> However, since the function is very small and only uses two values from
> the table, it is better for it to be defined as inline function in
> gf128mul.h. That way, the function can be inlined by the compiler for
> better performance.
>
> For consistency, the other gf128mul_x_* functions are also moved to the
> header file. In addition, the code is rewritten to be constant-time.
>
> After this change, the speed of the generic 'xts(aes)' implementation
> increased from ~225 MiB/s to ~235 MiB/s (measured using 'cryptsetup
> benchmark -c aes-xts-plain64' on an Intel system with CRYPTO_AES_X86_64
> and CRYPTO_AES_NI_INTEL disabled).
>
> Signed-off-by: Ondrej Mosnacek <omosnacek@xxxxxxxxx>
> Cc: Eric Biggers <ebiggers@xxxxxxxxxx>
Reviewed-by: Eric Biggers <ebiggers@xxxxxxxxxx>
Also, I realized that for gf128mul_x_lle() now that we aren't using the table we
don't need to shift '_tt' but rather can use the constant 0xe100000000000000:
/* equivalent to (u64)gf128mul_table_le[(b << 7) & 0xff] << 48
* (see crypto/gf128mul.c): */
u64 _tt = gf128mul_mask_from_bit(b, 0) & 0xe100000000000000;
r->b = cpu_to_be64((b >> 1) | (a << 63));
r->a = cpu_to_be64((a >> 1) ^ _tt);
I think that would be better and you could send a v4 to do it that way if you
want. It's not a huge deal though.
Thanks!
- Eric
