Merge branch 'alpha-reformulation' into 'master'

Reformulations des considérations d'implémentation sur le tweakey schedule

See merge request paclido/sp3!11

1 files changed, 17 insertions, 53 deletions

diff --git a/src/ref/tweakey.c b/src/ref/tweakey.c
index 319ca49..7019037 100644
--- a/src/ref/tweakey.c
+++ b/src/ref/tweakey.c
@@ -14,7 +14,7 @@ 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 functions expressing the exponentiated matrices with shifts and XORs.
+by applying functions for M and M_R n times.
 */
 
 #include <stdint.h>
@@ -62,26 +62,6 @@ void tweakey_state_extract(
 }
 
 
-static uint8_t _M1(uint8_t x)
-{
-    return x<<3 ^ x>>3;
-}
-
-static uint8_t _M2(uint8_t x)
-{
-    return x<<6 ^ (x & 0xf8) ^ 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];
@@ -96,26 +76,18 @@ static void _multiply_M(const uint8_t X[LANE_BYTES], uint8_t Y[LANE_BYTES])
 
 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];
+    uint8_t M_X[LANE_BYTES];
+    _multiply_M(X, M_X);
+    _multiply_M(M_X, Y);
 }
 
 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];
+    uint8_t M_X[LANE_BYTES];
+    uint8_t M2_X[LANE_BYTES];
+    _multiply_M(X, M_X);
+    _multiply_M(M_X, M2_X);
+    _multiply_M(M2_X, Y);
 }
 
 static void _multiply_MR(const uint8_t X[LANE_BYTES], uint8_t Y[LANE_BYTES])
@@ -132,26 +104,18 @@ static void _multiply_MR(const uint8_t X[LANE_BYTES], uint8_t Y[LANE_BYTES])
 
 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];
+    uint8_t MR_X[LANE_BYTES];
+    _multiply_MR(X, MR_X);
+    _multiply_MR(MR_X, Y);
 }
 
 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];
+    uint8_t MR_X[LANE_BYTES];
+    uint8_t MR2_X[LANE_BYTES];
+    _multiply_MR(X, MR_X);
+    _multiply_MR(MR_X, MR2_X);
+    _multiply_MR(MR2_X, Y);
 }
 
 typedef void (*matrix_multiplication)(const uint8_t X[LANE_BYTES], uint8_t Y[LANE_BYTES]);