Change the BlockCipher interface to support multi-block encryption and

decryption. Currently only used for counter mode. Doesn't offer much
advantage as-is (though might help slightly, in terms of cache effects),
but allows for SIMD implementations to process multiple blocks in parallel
when possible. Particularly thinking here of Serpent; TEA/XTEA also seem
promising in this sense, as is Threefish once that is implemented as a
standalone block cipher.

2 files changed, 108 insertions, 88 deletions

diff --git a/src/block/square/square.cpp b/src/block/square/square.cpp
index cb226542d..fdd47d3b2 100644
--- a/src/block/square/square.cpp
+++ b/src/block/square/square.cpp
@@ -14,103 +14,123 @@ namespace Botan {
 /*
 * Square Encryption
 */
-void Square::enc(const byte in[], byte out[]) const
+void Square::encrypt_n(const byte in[], byte out[], u32bit blocks) const
    {
-   u32bit T0, T1, T2, T3, B0, B1, B2, B3;
-   B0 = TE0[in[ 0] ^ ME[ 0]] ^ TE1[in[ 4] ^ ME[ 4]] ^
-        TE2[in[ 8] ^ ME[ 8]] ^ TE3[in[12] ^ ME[12]] ^ EK[0];
-   B1 = TE0[in[ 1] ^ ME[ 1]] ^ TE1[in[ 5] ^ ME[ 5]] ^
-        TE2[in[ 9] ^ ME[ 9]] ^ TE3[in[13] ^ ME[13]] ^ EK[1];
-   B2 = TE0[in[ 2] ^ ME[ 2]] ^ TE1[in[ 6] ^ ME[ 6]] ^
-        TE2[in[10] ^ ME[10]] ^ TE3[in[14] ^ ME[14]] ^ EK[2];
-   B3 = TE0[in[ 3] ^ ME[ 3]] ^ TE1[in[ 7] ^ ME[ 7]] ^
-        TE2[in[11] ^ ME[11]] ^ TE3[in[15] ^ ME[15]] ^ EK[3];
-   for(u32bit j = 1; j != 7; j += 2)
+   for(u32bit i = 0; i != blocks; ++i)
       {
-      T0 = TE0[get_byte(0, B0)] ^ TE1[get_byte(0, B1)] ^
-           TE2[get_byte(0, B2)] ^ TE3[get_byte(0, B3)] ^ EK[4*j+0];
-      T1 = TE0[get_byte(1, B0)] ^ TE1[get_byte(1, B1)] ^
-           TE2[get_byte(1, B2)] ^ TE3[get_byte(1, B3)] ^ EK[4*j+1];
-      T2 = TE0[get_byte(2, B0)] ^ TE1[get_byte(2, B1)] ^
-           TE2[get_byte(2, B2)] ^ TE3[get_byte(2, B3)] ^ EK[4*j+2];
-      T3 = TE0[get_byte(3, B0)] ^ TE1[get_byte(3, B1)] ^
-           TE2[get_byte(3, B2)] ^ TE3[get_byte(3, B3)] ^ EK[4*j+3];
-      B0 = TE0[get_byte(0, T0)] ^ TE1[get_byte(0, T1)] ^
-           TE2[get_byte(0, T2)] ^ TE3[get_byte(0, T3)] ^ EK[4*j+4];
-      B1 = TE0[get_byte(1, T0)] ^ TE1[get_byte(1, T1)] ^
-           TE2[get_byte(1, T2)] ^ TE3[get_byte(1, T3)] ^ EK[4*j+5];
-      B2 = TE0[get_byte(2, T0)] ^ TE1[get_byte(2, T1)] ^
-           TE2[get_byte(2, T2)] ^ TE3[get_byte(2, T3)] ^ EK[4*j+6];
-      B3 = TE0[get_byte(3, T0)] ^ TE1[get_byte(3, T1)] ^
-           TE2[get_byte(3, T2)] ^ TE3[get_byte(3, T3)] ^ EK[4*j+7];
+      u32bit T0, T1, T2, T3, B0, B1, B2, B3;
+
+      B0 = TE0[in[ 0] ^ ME[ 0]] ^ TE1[in[ 4] ^ ME[ 4]] ^
+           TE2[in[ 8] ^ ME[ 8]] ^ TE3[in[12] ^ ME[12]] ^ EK[0];
+      B1 = TE0[in[ 1] ^ ME[ 1]] ^ TE1[in[ 5] ^ ME[ 5]] ^
+           TE2[in[ 9] ^ ME[ 9]] ^ TE3[in[13] ^ ME[13]] ^ EK[1];
+      B2 = TE0[in[ 2] ^ ME[ 2]] ^ TE1[in[ 6] ^ ME[ 6]] ^
+           TE2[in[10] ^ ME[10]] ^ TE3[in[14] ^ ME[14]] ^ EK[2];
+      B3 = TE0[in[ 3] ^ ME[ 3]] ^ TE1[in[ 7] ^ ME[ 7]] ^
+           TE2[in[11] ^ ME[11]] ^ TE3[in[15] ^ ME[15]] ^ EK[3];
+
+      for(u32bit j = 1; j != 7; j += 2)
+         {
+         T0 = TE0[get_byte(0, B0)] ^ TE1[get_byte(0, B1)] ^
+              TE2[get_byte(0, B2)] ^ TE3[get_byte(0, B3)] ^ EK[4*j+0];
+         T1 = TE0[get_byte(1, B0)] ^ TE1[get_byte(1, B1)] ^
+              TE2[get_byte(1, B2)] ^ TE3[get_byte(1, B3)] ^ EK[4*j+1];
+         T2 = TE0[get_byte(2, B0)] ^ TE1[get_byte(2, B1)] ^
+              TE2[get_byte(2, B2)] ^ TE3[get_byte(2, B3)] ^ EK[4*j+2];
+         T3 = TE0[get_byte(3, B0)] ^ TE1[get_byte(3, B1)] ^
+              TE2[get_byte(3, B2)] ^ TE3[get_byte(3, B3)] ^ EK[4*j+3];
+
+         B0 = TE0[get_byte(0, T0)] ^ TE1[get_byte(0, T1)] ^
+              TE2[get_byte(0, T2)] ^ TE3[get_byte(0, T3)] ^ EK[4*j+4];
+         B1 = TE0[get_byte(1, T0)] ^ TE1[get_byte(1, T1)] ^
+              TE2[get_byte(1, T2)] ^ TE3[get_byte(1, T3)] ^ EK[4*j+5];
+         B2 = TE0[get_byte(2, T0)] ^ TE1[get_byte(2, T1)] ^
+              TE2[get_byte(2, T2)] ^ TE3[get_byte(2, T3)] ^ EK[4*j+6];
+         B3 = TE0[get_byte(3, T0)] ^ TE1[get_byte(3, T1)] ^
+              TE2[get_byte(3, T2)] ^ TE3[get_byte(3, T3)] ^ EK[4*j+7];
+         }
+
+      out[ 0] = SE[get_byte(0, B0)] ^ ME[16];
+      out[ 1] = SE[get_byte(0, B1)] ^ ME[17];
+      out[ 2] = SE[get_byte(0, B2)] ^ ME[18];
+      out[ 3] = SE[get_byte(0, B3)] ^ ME[19];
+      out[ 4] = SE[get_byte(1, B0)] ^ ME[20];
+      out[ 5] = SE[get_byte(1, B1)] ^ ME[21];
+      out[ 6] = SE[get_byte(1, B2)] ^ ME[22];
+      out[ 7] = SE[get_byte(1, B3)] ^ ME[23];
+      out[ 8] = SE[get_byte(2, B0)] ^ ME[24];
+      out[ 9] = SE[get_byte(2, B1)] ^ ME[25];
+      out[10] = SE[get_byte(2, B2)] ^ ME[26];
+      out[11] = SE[get_byte(2, B3)] ^ ME[27];
+      out[12] = SE[get_byte(3, B0)] ^ ME[28];
+      out[13] = SE[get_byte(3, B1)] ^ ME[29];
+      out[14] = SE[get_byte(3, B2)] ^ ME[30];
+      out[15] = SE[get_byte(3, B3)] ^ ME[31];
+
+      in += BLOCK_SIZE;
+      out += BLOCK_SIZE;
       }
-   out[ 0] = SE[get_byte(0, B0)] ^ ME[16];
-   out[ 1] = SE[get_byte(0, B1)] ^ ME[17];
-   out[ 2] = SE[get_byte(0, B2)] ^ ME[18];
-   out[ 3] = SE[get_byte(0, B3)] ^ ME[19];
-   out[ 4] = SE[get_byte(1, B0)] ^ ME[20];
-   out[ 5] = SE[get_byte(1, B1)] ^ ME[21];
-   out[ 6] = SE[get_byte(1, B2)] ^ ME[22];
-   out[ 7] = SE[get_byte(1, B3)] ^ ME[23];
-   out[ 8] = SE[get_byte(2, B0)] ^ ME[24];
-   out[ 9] = SE[get_byte(2, B1)] ^ ME[25];
-   out[10] = SE[get_byte(2, B2)] ^ ME[26];
-   out[11] = SE[get_byte(2, B3)] ^ ME[27];
-   out[12] = SE[get_byte(3, B0)] ^ ME[28];
-   out[13] = SE[get_byte(3, B1)] ^ ME[29];
-   out[14] = SE[get_byte(3, B2)] ^ ME[30];
-   out[15] = SE[get_byte(3, B3)] ^ ME[31];
    }
 
 /*
 * Square Decryption
 */
-void Square::dec(const byte in[], byte out[]) const
+void Square::decrypt_n(const byte in[], byte out[], u32bit blocks) const
    {
-   u32bit T0, T1, T2, T3, B0, B1, B2, B3;
-   B0 = TD0[in[ 0] ^ MD[ 0]] ^ TD1[in[ 4] ^ MD[ 4]] ^
-        TD2[in[ 8] ^ MD[ 8]] ^ TD3[in[12] ^ MD[12]] ^ DK[0];
-   B1 = TD0[in[ 1] ^ MD[ 1]] ^ TD1[in[ 5] ^ MD[ 5]] ^
-        TD2[in[ 9] ^ MD[ 9]] ^ TD3[in[13] ^ MD[13]] ^ DK[1];
-   B2 = TD0[in[ 2] ^ MD[ 2]] ^ TD1[in[ 6] ^ MD[ 6]] ^
-        TD2[in[10] ^ MD[10]] ^ TD3[in[14] ^ MD[14]] ^ DK[2];
-   B3 = TD0[in[ 3] ^ MD[ 3]] ^ TD1[in[ 7] ^ MD[ 7]] ^
-        TD2[in[11] ^ MD[11]] ^ TD3[in[15] ^ MD[15]] ^ DK[3];
-   for(u32bit j = 1; j != 7; j += 2)
+   for(u32bit i = 0; i != blocks; ++i)
       {
-      T0 = TD0[get_byte(0, B0)] ^ TD1[get_byte(0, B1)] ^
-           TD2[get_byte(0, B2)] ^ TD3[get_byte(0, B3)] ^ DK[4*j+0];
-      T1 = TD0[get_byte(1, B0)] ^ TD1[get_byte(1, B1)] ^
-           TD2[get_byte(1, B2)] ^ TD3[get_byte(1, B3)] ^ DK[4*j+1];
-      T2 = TD0[get_byte(2, B0)] ^ TD1[get_byte(2, B1)] ^
-           TD2[get_byte(2, B2)] ^ TD3[get_byte(2, B3)] ^ DK[4*j+2];
-      T3 = TD0[get_byte(3, B0)] ^ TD1[get_byte(3, B1)] ^
-           TD2[get_byte(3, B2)] ^ TD3[get_byte(3, B3)] ^ DK[4*j+3];
-      B0 = TD0[get_byte(0, T0)] ^ TD1[get_byte(0, T1)] ^
-           TD2[get_byte(0, T2)] ^ TD3[get_byte(0, T3)] ^ DK[4*j+4];
-      B1 = TD0[get_byte(1, T0)] ^ TD1[get_byte(1, T1)] ^
-           TD2[get_byte(1, T2)] ^ TD3[get_byte(1, T3)] ^ DK[4*j+5];
-      B2 = TD0[get_byte(2, T0)] ^ TD1[get_byte(2, T1)] ^
-           TD2[get_byte(2, T2)] ^ TD3[get_byte(2, T3)] ^ DK[4*j+6];
-      B3 = TD0[get_byte(3, T0)] ^ TD1[get_byte(3, T1)] ^
-           TD2[get_byte(3, T2)] ^ TD3[get_byte(3, T3)] ^ DK[4*j+7];
+      u32bit T0, T1, T2, T3, B0, B1, B2, B3;
+
+      B0 = TD0[in[ 0] ^ MD[ 0]] ^ TD1[in[ 4] ^ MD[ 4]] ^
+           TD2[in[ 8] ^ MD[ 8]] ^ TD3[in[12] ^ MD[12]] ^ DK[0];
+      B1 = TD0[in[ 1] ^ MD[ 1]] ^ TD1[in[ 5] ^ MD[ 5]] ^
+           TD2[in[ 9] ^ MD[ 9]] ^ TD3[in[13] ^ MD[13]] ^ DK[1];
+      B2 = TD0[in[ 2] ^ MD[ 2]] ^ TD1[in[ 6] ^ MD[ 6]] ^
+           TD2[in[10] ^ MD[10]] ^ TD3[in[14] ^ MD[14]] ^ DK[2];
+      B3 = TD0[in[ 3] ^ MD[ 3]] ^ TD1[in[ 7] ^ MD[ 7]] ^
+           TD2[in[11] ^ MD[11]] ^ TD3[in[15] ^ MD[15]] ^ DK[3];
+
+      for(u32bit j = 1; j != 7; j += 2)
+         {
+         T0 = TD0[get_byte(0, B0)] ^ TD1[get_byte(0, B1)] ^
+              TD2[get_byte(0, B2)] ^ TD3[get_byte(0, B3)] ^ DK[4*j+0];
+         T1 = TD0[get_byte(1, B0)] ^ TD1[get_byte(1, B1)] ^
+              TD2[get_byte(1, B2)] ^ TD3[get_byte(1, B3)] ^ DK[4*j+1];
+         T2 = TD0[get_byte(2, B0)] ^ TD1[get_byte(2, B1)] ^
+              TD2[get_byte(2, B2)] ^ TD3[get_byte(2, B3)] ^ DK[4*j+2];
+         T3 = TD0[get_byte(3, B0)] ^ TD1[get_byte(3, B1)] ^
+              TD2[get_byte(3, B2)] ^ TD3[get_byte(3, B3)] ^ DK[4*j+3];
+
+         B0 = TD0[get_byte(0, T0)] ^ TD1[get_byte(0, T1)] ^
+              TD2[get_byte(0, T2)] ^ TD3[get_byte(0, T3)] ^ DK[4*j+4];
+         B1 = TD0[get_byte(1, T0)] ^ TD1[get_byte(1, T1)] ^
+              TD2[get_byte(1, T2)] ^ TD3[get_byte(1, T3)] ^ DK[4*j+5];
+         B2 = TD0[get_byte(2, T0)] ^ TD1[get_byte(2, T1)] ^
+              TD2[get_byte(2, T2)] ^ TD3[get_byte(2, T3)] ^ DK[4*j+6];
+         B3 = TD0[get_byte(3, T0)] ^ TD1[get_byte(3, T1)] ^
+              TD2[get_byte(3, T2)] ^ TD3[get_byte(3, T3)] ^ DK[4*j+7];
+         }
+
+      out[ 0] = SD[get_byte(0, B0)] ^ MD[16];
+      out[ 1] = SD[get_byte(0, B1)] ^ MD[17];
+      out[ 2] = SD[get_byte(0, B2)] ^ MD[18];
+      out[ 3] = SD[get_byte(0, B3)] ^ MD[19];
+      out[ 4] = SD[get_byte(1, B0)] ^ MD[20];
+      out[ 5] = SD[get_byte(1, B1)] ^ MD[21];
+      out[ 6] = SD[get_byte(1, B2)] ^ MD[22];
+      out[ 7] = SD[get_byte(1, B3)] ^ MD[23];
+      out[ 8] = SD[get_byte(2, B0)] ^ MD[24];
+      out[ 9] = SD[get_byte(2, B1)] ^ MD[25];
+      out[10] = SD[get_byte(2, B2)] ^ MD[26];
+      out[11] = SD[get_byte(2, B3)] ^ MD[27];
+      out[12] = SD[get_byte(3, B0)] ^ MD[28];
+      out[13] = SD[get_byte(3, B1)] ^ MD[29];
+      out[14] = SD[get_byte(3, B2)] ^ MD[30];
+      out[15] = SD[get_byte(3, B3)] ^ MD[31];
+
+      in += BLOCK_SIZE;
+      out += BLOCK_SIZE;
       }
-   out[ 0] = SD[get_byte(0, B0)] ^ MD[16];
-   out[ 1] = SD[get_byte(0, B1)] ^ MD[17];
-   out[ 2] = SD[get_byte(0, B2)] ^ MD[18];
-   out[ 3] = SD[get_byte(0, B3)] ^ MD[19];
-   out[ 4] = SD[get_byte(1, B0)] ^ MD[20];
-   out[ 5] = SD[get_byte(1, B1)] ^ MD[21];
-   out[ 6] = SD[get_byte(1, B2)] ^ MD[22];
-   out[ 7] = SD[get_byte(1, B3)] ^ MD[23];
-   out[ 8] = SD[get_byte(2, B0)] ^ MD[24];
-   out[ 9] = SD[get_byte(2, B1)] ^ MD[25];
-   out[10] = SD[get_byte(2, B2)] ^ MD[26];
-   out[11] = SD[get_byte(2, B3)] ^ MD[27];
-   out[12] = SD[get_byte(3, B0)] ^ MD[28];
-   out[13] = SD[get_byte(3, B1)] ^ MD[29];
-   out[14] = SD[get_byte(3, B2)] ^ MD[30];
-   out[15] = SD[get_byte(3, B3)] ^ MD[31];
    }
 
 /*
diff --git a/src/block/square/square.h b/src/block/square/square.h
index 94a1fc370..0de4c20bd 100644
--- a/src/block/square/square.h
+++ b/src/block/square/square.h
@@ -23,8 +23,8 @@ class BOTAN_DLL Square : public BlockCipher
       BlockCipher* clone() const { return new Square; }
       Square() : BlockCipher(16, 16) {}
    private:
-      void enc(const byte[], byte[]) const;
-      void dec(const byte[], byte[]) const;
+      void encrypt_n(const byte in[], byte out[], u32bit blocks) const;
+      void decrypt_n(const byte in[], byte out[], u32bit blocks) const;
       void key_schedule(const byte[], u32bit);
 
       static void transform(u32bit[4]);