Implémentation de M₄

1 files changed, 16 insertions, 5 deletions

diff --git a/src/add_developedtweakey/tweakey.c b/src/add_developedtweakey/tweakey.c
index a602191..605098b 100644
--- a/src/add_developedtweakey/tweakey.c
+++ b/src/add_developedtweakey/tweakey.c
@@ -58,6 +58,14 @@ static uint8_t _M3(uint8_t x)
     return x & 0x1f;
 }
 
+static uint8_t _M4(uint8_t x)
+{
+    /* (x<<2)>>3 would keep bit x6 due to integer promotion.
+     * Side-step this by writing each shift on its own statement. */
+    x <<= 2;
+    return x>>3;
+}
+
 static void _multiply_M(const uint8_t X[LANE_BYTES], uint8_t Y[LANE_BYTES])
 {
     Y[7] = X[6];
@@ -120,11 +128,14 @@ static void _multiply_MR2(const uint8_t X[LANE_BYTES], uint8_t Y[LANE_BYTES])
 
 static void _multiply_MR3(const uint8_t X[LANE_BYTES], uint8_t Y[LANE_BYTES])
 {
-    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);
+    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]);