Utilisation du tweakey schedule "itératif" comme référence

Au final, il n'est pas moins performant que l'autre ; cf. issue #4.

1 files changed, 182 insertions, 0 deletions

diff --git a/src/add_tweakeyunrolled/tweakey.c b/src/add_tweakeyunrolled/tweakey.c
new file mode 100644
index 0000000..d0a4eb7
--- /dev/null
+++ b/src/add_tweakeyunrolled/tweakey.c
@@ -0,0 +1,182 @@
+/*
+Implementation of the Lilliput-AE tweakable block cipher.
+
+Author: Kévin Le Gouguec, 2019.
+
+For more information, feedback or questions, refer to our website:
+https://paclido.fr/lilliput-ae
+
+To the extent possible under law, the implementer has waived all copyright
+and related or neighboring rights to the source code in this file.
+http://creativecommons.org/publicdomain/zero/1.0/
+
+---
+
+This file provides an implementation of Lilliput-TBC's tweakey schedule,
+where multiplications by matrices M and M_R to the power n are performed
+by applying functions for M and M_R n times.
+*/
+
+#include <stdint.h>
+#include <string.h>
+
+#include "constants.h"
+#include "tweakey.h"
+
+
+#define LANE_BITS  64
+#define LANE_BYTES (LANE_BITS/8)
+#define LANES_NB   (TWEAKEY_BYTES/LANE_BYTES)
+
+
+void tweakey_state_init(
+    uint8_t TK[TWEAKEY_BYTES],
+    const uint8_t key[KEY_BYTES],
+    const uint8_t tweak[TWEAK_BYTES]
+)
+{
+    memcpy(TK,             tweak, TWEAK_BYTES);
+    memcpy(TK+TWEAK_BYTES, key,   KEY_BYTES);
+}
+
+
+void tweakey_state_extract(
+    const uint8_t TK[TWEAKEY_BYTES],
+    uint8_t round_constant,
+    uint8_t round_tweakey[ROUND_TWEAKEY_BYTES]
+)
+{
+    memset(round_tweakey, 0, ROUND_TWEAKEY_BYTES);
+
+    for (size_t j=0; j<LANES_NB; j++)
+    {
+        const uint8_t *TKj = TK + j*LANE_BYTES;
+
+        for (size_t k=0; k<LANE_BYTES; k++)
+        {
+            round_tweakey[k] ^= TKj[k];
+        }
+    }
+
+    round_tweakey[0] ^= round_constant;
+}
+
+
+static uint8_t _M1(uint8_t x)
+{
+    return x<<3 ^ x>>3;
+}
+
+static uint8_t _M2(uint8_t x)
+{
+    return x<<6 ^ (x>>3)<<3 ^ x>>6;
+}
+
+static uint8_t _M3(uint8_t x)
+{
+    return (uint8_t)(x<<3) >> 3;
+}
+
+static uint8_t _M4(uint8_t x)
+{
+    return (uint8_t)(x<<2) >> 3;
+}
+
+static void _multiply_M(const uint8_t X[LANE_BYTES], uint8_t Y[LANE_BYTES])
+{
+    Y[7] = X[6];
+    Y[6] = X[5];
+    Y[5] = X[5]<<3 ^ X[4];
+    Y[4] = X[4]>>3 ^ X[3];
+    Y[3] = X[2];
+    Y[2] = X[6]<<2 ^ X[1];
+    Y[1] = X[0];
+    Y[0] = X[7];
+}
+
+static void _multiply_M2(const uint8_t X[LANE_BYTES], uint8_t Y[LANE_BYTES])
+{
+    Y[7] = X[5];
+    Y[6] = X[5]<<3 ^ X[4];
+    Y[5] = X[5]<<6 ^ _M1(X[4]) ^ X[3];
+    Y[4] = X[4]>>6 ^ X[3]>>3   ^ X[2];
+    Y[3] = X[6]<<2 ^ X[1];
+    Y[2] = X[5]<<2 ^ X[0];
+    Y[1] = X[7];
+    Y[0] = X[6];
+}
+
+static void _multiply_M3(const uint8_t X[LANE_BYTES], uint8_t Y[LANE_BYTES])
+{
+    Y[7] = X[5]<<3   ^ X[4];
+    Y[6] = X[5]<<6   ^ _M1(X[4]) ^ X[3];
+    Y[5] = _M2(X[4]) ^ _M1(X[3]) ^ X[2];
+    Y[4] = X[6]<<2   ^ X[3]>>6   ^ X[2]>>3 ^ X[1];
+    Y[3] = X[5]<<2   ^ X[0];
+    Y[2] = X[7]      ^ X[5]<<5   ^ X[4]<<2;
+    Y[1] = X[6];
+    Y[0] = X[5];
+}
+
+static void _multiply_MR(const uint8_t X[LANE_BYTES], uint8_t Y[LANE_BYTES])
+{
+    Y[0] = X[1];
+    Y[1] = X[2];
+    Y[2] = X[3]    ^ X[4]>>3;
+    Y[3] = X[4];
+    Y[4] = X[5]    ^ X[6]<<3;
+    Y[5] = X[3]<<2 ^ X[6];
+    Y[6] = X[7];
+    Y[7] = X[0];
+}
+
+static void _multiply_MR2(const uint8_t X[LANE_BYTES], uint8_t Y[LANE_BYTES])
+{
+    Y[0] = X[2];
+    Y[1] = X[3]    ^ X[4]>>3;
+    Y[2] = X[4]    ^ X[5]>>3 ^ _M3(X[6]);
+    Y[3] = X[5]    ^ X[6]<<3;
+    Y[4] = X[3]<<2 ^ X[6]    ^ X[7]<<3;
+    Y[5] = X[4]<<2 ^ X[7];
+    Y[6] = X[0];
+    Y[7] = X[1];
+}
+
+static void _multiply_MR3(const uint8_t X[LANE_BYTES], uint8_t Y[LANE_BYTES])
+{
+    Y[0] = X[3]      ^ X[4]>>3;
+    Y[1] = X[4]      ^ X[5]>>3 ^ _M3(X[6]);
+    Y[2] = _M4(X[3]) ^ X[5]    ^ _M1(X[6]) ^ _M3(X[7]);
+    Y[3] = X[3]<<2   ^ X[6]    ^ X[7]<<3;
+    Y[4] = X[0]<<3   ^ X[4]<<2 ^ X[7];
+    Y[5] = X[0]      ^ X[5]<<2 ^ X[6]<<5;
+    Y[6] = X[1];
+    Y[7] = X[2];
+}
+
+typedef void (*matrix_multiplication)(const uint8_t X[LANE_BYTES], uint8_t Y[LANE_BYTES]);
+
+static const matrix_multiplication ALPHAS[6] = {
+    _multiply_M,
+    _multiply_M2,
+    _multiply_M3,
+    _multiply_MR,
+    _multiply_MR2,
+    _multiply_MR3
+};
+
+
+void tweakey_state_update(uint8_t TK[TWEAKEY_BYTES])
+{
+    /* Skip lane 0, as it is multiplied by the identity matrix. */
+
+    for (size_t j=1; j<LANES_NB; j++)
+    {
+        uint8_t *TKj = TK + j*LANE_BYTES;
+
+        uint8_t TKj_old[LANE_BYTES];
+        memcpy(TKj_old, TKj, LANE_BYTES);
+
+        ALPHAS[j-1](TKj_old, TKj);
+    }
+}