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.

1 files changed, 76 insertions, 64 deletions

diff --git a/src/block/rc6/rc6.cpp b/src/block/rc6/rc6.cpp
index 3b30ea93a..8bda62259 100644
--- a/src/block/rc6/rc6.cpp
+++ b/src/block/rc6/rc6.cpp
@@ -15,85 +15,97 @@ namespace Botan {
 /*
 * RC6 Encryption
 */
-void RC6::enc(const byte in[], byte out[]) const
+void RC6::encrypt_n(const byte in[], byte out[], u32bit blocks) const
    {
-   u32bit A = load_le<u32bit>(in, 0);
-   u32bit B = load_le<u32bit>(in, 1);
-   u32bit C = load_le<u32bit>(in, 2);
-   u32bit D = load_le<u32bit>(in, 3);
+   for(u32bit i = 0; i != blocks; ++i)
+      {
+      u32bit A = load_le<u32bit>(in, 0);
+      u32bit B = load_le<u32bit>(in, 1);
+      u32bit C = load_le<u32bit>(in, 2);
+      u32bit D = load_le<u32bit>(in, 3);
 
-   B += S[0]; D += S[1];
+      B += S[0]; D += S[1];
 
-   for(u32bit j = 0; j != 20; j += 4)
-      {
-      u32bit T1, T2;
-
-      T1 = rotate_left(B*(2*B+1), 5);
-      T2 = rotate_left(D*(2*D+1), 5);
-      A = rotate_left(A ^ T1, T2 % 32) + S[2*j+2];
-      C = rotate_left(C ^ T2, T1 % 32) + S[2*j+3];
-
-      T1 = rotate_left(C*(2*C+1), 5);
-      T2 = rotate_left(A*(2*A+1), 5);
-      B = rotate_left(B ^ T1, T2 % 32) + S[2*j+4];
-      D = rotate_left(D ^ T2, T1 % 32) + S[2*j+5];
-
-      T1 = rotate_left(D*(2*D+1), 5);
-      T2 = rotate_left(B*(2*B+1), 5);
-      C = rotate_left(C ^ T1, T2 % 32) + S[2*j+6];
-      A = rotate_left(A ^ T2, T1 % 32) + S[2*j+7];
-
-      T1 = rotate_left(A*(2*A+1), 5);
-      T2 = rotate_left(C*(2*C+1), 5);
-      D = rotate_left(D ^ T1, T2 % 32) + S[2*j+8];
-      B = rotate_left(B ^ T2, T1 % 32) + S[2*j+9];
-      }
+      for(u32bit j = 0; j != 20; j += 4)
+         {
+         u32bit T1, T2;
+
+         T1 = rotate_left(B*(2*B+1), 5);
+         T2 = rotate_left(D*(2*D+1), 5);
+         A = rotate_left(A ^ T1, T2 % 32) + S[2*j+2];
+         C = rotate_left(C ^ T2, T1 % 32) + S[2*j+3];
+
+         T1 = rotate_left(C*(2*C+1), 5);
+         T2 = rotate_left(A*(2*A+1), 5);
+         B = rotate_left(B ^ T1, T2 % 32) + S[2*j+4];
+         D = rotate_left(D ^ T2, T1 % 32) + S[2*j+5];
 
-   A += S[42]; C += S[43];
+         T1 = rotate_left(D*(2*D+1), 5);
+         T2 = rotate_left(B*(2*B+1), 5);
+         C = rotate_left(C ^ T1, T2 % 32) + S[2*j+6];
+         A = rotate_left(A ^ T2, T1 % 32) + S[2*j+7];
 
-   store_le(out, A, B, C, D);
+         T1 = rotate_left(A*(2*A+1), 5);
+         T2 = rotate_left(C*(2*C+1), 5);
+         D = rotate_left(D ^ T1, T2 % 32) + S[2*j+8];
+         B = rotate_left(B ^ T2, T1 % 32) + S[2*j+9];
+         }
+
+      A += S[42]; C += S[43];
+
+      store_le(out, A, B, C, D);
+
+      in += BLOCK_SIZE;
+      out += BLOCK_SIZE;
+      }
    }
 
 /*
 * RC6 Decryption
 */
-void RC6::dec(const byte in[], byte out[]) const
+void RC6::decrypt_n(const byte in[], byte out[], u32bit blocks) const
    {
-   u32bit A = load_le<u32bit>(in, 0);
-   u32bit B = load_le<u32bit>(in, 1);
-   u32bit C = load_le<u32bit>(in, 2);
-   u32bit D = load_le<u32bit>(in, 3);
+   for(u32bit i = 0; i != blocks; ++i)
+      {
+      u32bit A = load_le<u32bit>(in, 0);
+      u32bit B = load_le<u32bit>(in, 1);
+      u32bit C = load_le<u32bit>(in, 2);
+      u32bit D = load_le<u32bit>(in, 3);
 
-   C -= S[43]; A -= S[42];
+      C -= S[43]; A -= S[42];
 
-   for(u32bit j = 0; j != 20; j += 4)
-      {
-      u32bit T1, T2;
-
-      T1 = rotate_left(A*(2*A+1), 5);
-      T2 = rotate_left(C*(2*C+1), 5);
-      B = rotate_right(B - S[41 - 2*j], T1 % 32) ^ T2;
-      D = rotate_right(D - S[40 - 2*j], T2 % 32) ^ T1;
-
-      T1 = rotate_left(D*(2*D+1), 5);
-      T2 = rotate_left(B*(2*B+1), 5);
-      A = rotate_right(A - S[39 - 2*j], T1 % 32) ^ T2;
-      C = rotate_right(C - S[38 - 2*j], T2 % 32) ^ T1;
-
-      T1 = rotate_left(C*(2*C+1), 5);
-      T2 = rotate_left(A*(2*A+1), 5);
-      D = rotate_right(D - S[37 - 2*j], T1 % 32) ^ T2;
-      B = rotate_right(B - S[36 - 2*j], T2 % 32) ^ T1;
-
-      T1 = rotate_left(B*(2*B+1), 5);
-      T2 = rotate_left(D*(2*D+1), 5);
-      C = rotate_right(C - S[35 - 2*j], T1 % 32) ^ T2;
-      A = rotate_right(A - S[34 - 2*j], T2 % 32) ^ T1;
-      }
+      for(u32bit j = 0; j != 20; j += 4)
+         {
+         u32bit T1, T2;
+
+         T1 = rotate_left(A*(2*A+1), 5);
+         T2 = rotate_left(C*(2*C+1), 5);
+         B = rotate_right(B - S[41 - 2*j], T1 % 32) ^ T2;
+         D = rotate_right(D - S[40 - 2*j], T2 % 32) ^ T1;
+
+         T1 = rotate_left(D*(2*D+1), 5);
+         T2 = rotate_left(B*(2*B+1), 5);
+         A = rotate_right(A - S[39 - 2*j], T1 % 32) ^ T2;
+         C = rotate_right(C - S[38 - 2*j], T2 % 32) ^ T1;
 
-   D -= S[1]; B -= S[0];
+         T1 = rotate_left(C*(2*C+1), 5);
+         T2 = rotate_left(A*(2*A+1), 5);
+         D = rotate_right(D - S[37 - 2*j], T1 % 32) ^ T2;
+         B = rotate_right(B - S[36 - 2*j], T2 % 32) ^ T1;
 
-   store_le(out, A, B, C, D);
+         T1 = rotate_left(B*(2*B+1), 5);
+         T2 = rotate_left(D*(2*D+1), 5);
+         C = rotate_right(C - S[35 - 2*j], T1 % 32) ^ T2;
+         A = rotate_right(A - S[34 - 2*j], T2 % 32) ^ T1;
+         }
+
+      D -= S[1]; B -= S[0];
+
+      store_le(out, A, B, C, D);
+
+      in += BLOCK_SIZE;
+      out += BLOCK_SIZE;
+      }
    }
 
 /*