diff options
-rw-r--r-- | src/block/xtea_sse2/xtea_sse2.cpp | 58 |
1 files changed, 47 insertions, 11 deletions
diff --git a/src/block/xtea_sse2/xtea_sse2.cpp b/src/block/xtea_sse2/xtea_sse2.cpp index db7775103..5d9c7dd05 100644 --- a/src/block/xtea_sse2/xtea_sse2.cpp +++ b/src/block/xtea_sse2/xtea_sse2.cpp @@ -1,5 +1,5 @@ /* -* XTEA SSE2 +* XTEA in SIMD * (C) 2009 Jack Lloyd * * Distributed under the terms of the Botan license @@ -24,7 +24,10 @@ void xtea_encrypt_8(const byte in[64], byte out[64], const u32bit EK[64]) for(u32bit i = 0; i != 32; i += 2) { - SIMD_32 K0(EK[2*i]), K1(EK[2*i+1]), K2(EK[2*i+2]), K3(EK[2*i+3]); + SIMD_32 K0(EK[2*i ]); + SIMD_32 K1(EK[2*i+1]); + SIMD_32 K2(EK[2*i+2]); + SIMD_32 K3(EK[2*i+3]); L0 += (((R0 << 4) ^ (R0 >> 5)) + R0) ^ K0; L1 += (((R1 << 4) ^ (R1 >> 5)) + R1) ^ K0; @@ -47,6 +50,43 @@ void xtea_encrypt_8(const byte in[64], byte out[64], const u32bit EK[64]) R1.store_be(out + 48); } +void xtea_decrypt_8(const byte in[64], byte out[64], const u32bit EK[64]) + { + SIMD_32 L0 = SIMD_32::load_be(in ); + SIMD_32 R0 = SIMD_32::load_be(in + 16); + SIMD_32 L1 = SIMD_32::load_be(in + 32); + SIMD_32 R1 = SIMD_32::load_be(in + 48); + + SIMD_32::transpose(L0, R0, L1, R1); + + for(u32bit i = 0; i != 32; i += 2) + { + SIMD_32 K0(EK[63 - 2*i]); + SIMD_32 K1(EK[62 - 2*i]); + SIMD_32 K2(EK[61 - 2*i]); + SIMD_32 K3(EK[60 - 2*i]); + + R0 -= (((L0 << 4) ^ (L0 >> 5)) + L0) ^ K0; + R1 -= (((L1 << 4) ^ (L1 >> 5)) + L1) ^ K0; + + L0 -= (((R0 << 4) ^ (R0 >> 5)) + R0) ^ K1; + L1 -= (((R1 << 4) ^ (R1 >> 5)) + R1) ^ K1; + + R0 -= (((L0 << 4) ^ (L0 >> 5)) + L0) ^ K2; + R1 -= (((L1 << 4) ^ (L1 >> 5)) + L1) ^ K2; + + L0 -= (((R0 << 4) ^ (R0 >> 5)) + R0) ^ K3; + L1 -= (((R1 << 4) ^ (R1 >> 5)) + R1) ^ K3; + } + + SIMD_32::transpose(L0, R0, L1, R1); + + L0.store_be(out); + R0.store_be(out + 16); + L1.store_be(out + 32); + R1.store_be(out + 48); + } + } /* @@ -54,8 +94,6 @@ void xtea_encrypt_8(const byte in[64], byte out[64], const u32bit EK[64]) */ void XTEA_SSE2::encrypt_n(const byte in[], byte out[], u32bit blocks) const { - memset(out, 0, blocks * BLOCK_SIZE); - while(blocks >= 8) { xtea_encrypt_8(in, out, this->EK); @@ -72,15 +110,13 @@ void XTEA_SSE2::encrypt_n(const byte in[], byte out[], u32bit blocks) const */ void XTEA_SSE2::decrypt_n(const byte in[], byte out[], u32bit blocks) const { -#if 0 - while(blocks >= 4) + while(blocks >= 8) { - xtea_decrypt_4(in, out, this->EK); - in += 4 * BLOCK_SIZE; - out += 4 * BLOCK_SIZE; - blocks -= 4; + xtea_decrypt_8(in, out, this->EK); + in += 8 * BLOCK_SIZE; + out += 8 * BLOCK_SIZE; + blocks -= 8; } -#endif XTEA::decrypt_n(in, out, blocks); } |