aboutsummaryrefslogtreecommitdiffstats
path: root/src/block
diff options
context:
space:
mode:
authorlloyd <[email protected]>2009-10-29 01:40:55 +0000
committerlloyd <[email protected]>2009-10-29 01:40:55 +0000
commit78cb47ae1b43c51d6e91531f701ccf03fa8ea2c6 (patch)
treec71ff6ff187cc7ef15cf3c233ea7bdb43ef6f6df /src/block
parentee169027c6cd30923aa30a735eb801836ee593d2 (diff)
parent1bc4d2fb37c8f1e6e94a65ec67062826393dda7f (diff)
propagate from branch 'net.randombit.botan' (head 8fb69dd1c599ada1008c4cab2a6d502cbcc468e0)
to branch 'net.randombit.botan.general-simd' (head c05c9a6d398659891fb8cca170ed514ea7e6476d)
Diffstat (limited to 'src/block')
-rw-r--r--src/block/serpent_simd/info.txt8
-rw-r--r--src/block/serpent_simd/serp_simd.cpp (renamed from src/block/serpent_sse2/serp_sse2.cpp)154
-rw-r--r--src/block/serpent_simd/serp_simd.h (renamed from src/block/serpent_sse2/serp_sse2.h)10
-rw-r--r--src/block/serpent_simd/serp_simd_sbox.h426
-rw-r--r--src/block/serpent_sse2/info.txt6
-rw-r--r--src/block/serpent_sse2/serp_sse2_sbox.h434
-rw-r--r--src/block/xtea/xtea.h2
-rw-r--r--src/block/xtea_simd/info.txt16
-rw-r--r--src/block/xtea_simd/xtea_simd.cpp124
-rw-r--r--src/block/xtea_simd/xtea_simd.h28
10 files changed, 668 insertions, 540 deletions
diff --git a/src/block/serpent_simd/info.txt b/src/block/serpent_simd/info.txt
new file mode 100644
index 000000000..5d9115a4d
--- /dev/null
+++ b/src/block/serpent_simd/info.txt
@@ -0,0 +1,8 @@
+realname "Serpent (SIMD)"
+define SERPENT_SIMD
+
+<requires>
+serpent
+simd_32
+simd_engine
+</requires>
diff --git a/src/block/serpent_sse2/serp_sse2.cpp b/src/block/serpent_simd/serp_simd.cpp
index c51bb69ab..b394b0c26 100644
--- a/src/block/serpent_sse2/serp_sse2.cpp
+++ b/src/block/serpent_simd/serp_simd.cpp
@@ -1,99 +1,71 @@
/*
-* Serpent (SSE2)
+* Serpent (SIMD)
* (C) 2009 Jack Lloyd
*
* Distributed under the terms of the Botan license
*/
-#include <botan/serp_sse2.h>
-#include <botan/serp_sse2_sbox.h>
+#include <botan/serp_simd.h>
+#include <botan/serp_simd_sbox.h>
+#include <botan/simd_32.h>
#include <botan/loadstor.h>
-#include <emmintrin.h>
namespace Botan {
namespace {
-#define key_xor(round, B0, B1, B2, B3) \
- do { \
- __m128i key = _mm_loadu_si128(keys + round); \
- B0 = _mm_xor_si128(B0, _mm_shuffle_epi32(key, _MM_SHUFFLE(0,0,0,0))); \
- B1 = _mm_xor_si128(B1, _mm_shuffle_epi32(key, _MM_SHUFFLE(1,1,1,1))); \
- B2 = _mm_xor_si128(B2, _mm_shuffle_epi32(key, _MM_SHUFFLE(2,2,2,2))); \
- B3 = _mm_xor_si128(B3, _mm_shuffle_epi32(key, _MM_SHUFFLE(3,3,3,3))); \
+#define key_xor(round, B0, B1, B2, B3) \
+ do { \
+ B0 ^= SIMD_32(keys[4*round ]); \
+ B1 ^= SIMD_32(keys[4*round+1]); \
+ B2 ^= SIMD_32(keys[4*round+2]); \
+ B3 ^= SIMD_32(keys[4*round+3]); \
} while(0);
/*
* Serpent's linear transformations
*/
-#define rotate_left_m128(vec, rot) \
- _mm_or_si128(_mm_slli_epi32(vec, rot), _mm_srli_epi32(vec, 32-rot))
-
-#define rotate_right_m128(vec, rot) \
- _mm_or_si128(_mm_srli_epi32(vec, rot), _mm_slli_epi32(vec, 32-rot))
-
-#define transform(B0, B1, B2, B3) \
- do { \
- B0 = rotate_left_m128(B0, 13); \
- B2 = rotate_left_m128(B2, 3); \
- B1 = _mm_xor_si128(B1, _mm_xor_si128(B0, B2)); \
- B3 = _mm_xor_si128(B3, _mm_xor_si128(B2, _mm_slli_epi32(B0, 3))); \
- B1 = rotate_left_m128(B1, 1); \
- B3 = rotate_left_m128(B3, 7); \
- B0 = _mm_xor_si128(B0, _mm_xor_si128(B1, B3)); \
- B2 = _mm_xor_si128(B2, _mm_xor_si128(B3, _mm_slli_epi32(B1, 7))); \
- B0 = rotate_left_m128(B0, 5); \
- B2 = rotate_left_m128(B2, 22); \
+#define transform(B0, B1, B2, B3) \
+ do { \
+ B0.rotate_left(13); \
+ B2.rotate_left(3); \
+ B1 ^= B0 ^ B2; \
+ B3 ^= B2 ^ (B0 << 3); \
+ B1.rotate_left(1); \
+ B3.rotate_left(7); \
+ B0 ^= B1 ^ B3; \
+ B2 ^= B3 ^ (B1 << 7); \
+ B0.rotate_left(5); \
+ B2.rotate_left(22); \
} while(0);
-#define i_transform(B0, B1, B2, B3) \
- do { \
- B2 = rotate_right_m128(B2, 22); \
- B0 = rotate_right_m128(B0, 5); \
- B2 = _mm_xor_si128(B2, _mm_xor_si128(B3, _mm_slli_epi32(B1, 7))); \
- B0 = _mm_xor_si128(B0, _mm_xor_si128(B1, B3)); \
- B3 = rotate_right_m128(B3, 7); \
- B1 = rotate_right_m128(B1, 1); \
- B3 = _mm_xor_si128(B3, _mm_xor_si128(B2, _mm_slli_epi32(B0, 3))); \
- B1 = _mm_xor_si128(B1, _mm_xor_si128(B0, B2)); \
- B2 = rotate_right_m128(B2, 3); \
- B0 = rotate_right_m128(B0, 13); \
+#define i_transform(B0, B1, B2, B3) \
+ do { \
+ B2.rotate_right(22); \
+ B0.rotate_right(5); \
+ B2 ^= B3 ^ (B1 << 7); \
+ B0 ^= B1 ^ B3; \
+ B3.rotate_right(7); \
+ B1.rotate_right(1); \
+ B3 ^= B2 ^ (B0 << 3); \
+ B1 ^= B0 ^ B2; \
+ B2.rotate_right(3); \
+ B0.rotate_right(13); \
} while(0);
/*
-* 4x4 SSE2 integer matrix transpose
-*/
-#define transpose(B0, B1, B2, B3) \
- do { \
- __m128i T0 = _mm_unpacklo_epi32(B0, B1); \
- __m128i T1 = _mm_unpacklo_epi32(B2, B3); \
- __m128i T2 = _mm_unpackhi_epi32(B0, B1); \
- __m128i T3 = _mm_unpackhi_epi32(B2, B3); \
- B0 = _mm_unpacklo_epi64(T0, T1); \
- B1 = _mm_unpackhi_epi64(T0, T1); \
- B2 = _mm_unpacklo_epi64(T2, T3); \
- B3 = _mm_unpackhi_epi64(T2, T3); \
- } while(0);
-
-/*
-* SSE2 Serpent Encryption of 4 blocks in parallel
+* SIMD Serpent Encryption of 4 blocks in parallel
*/
void serpent_encrypt_4(const byte in[64],
byte out[64],
- const u32bit keys_32[132])
+ const u32bit keys[132])
{
- const __m128i all_ones = _mm_set1_epi8(0xFF);
+ SIMD_32 B0 = SIMD_32::load_le(in);
+ SIMD_32 B1 = SIMD_32::load_le(in + 16);
+ SIMD_32 B2 = SIMD_32::load_le(in + 32);
+ SIMD_32 B3 = SIMD_32::load_le(in + 48);
- const __m128i* keys = (const __m128i*)(keys_32);
- __m128i* out_mm = (__m128i*)(out);
- __m128i* in_mm = (__m128i*)(in);
-
- __m128i B0 = _mm_loadu_si128(in_mm);
- __m128i B1 = _mm_loadu_si128(in_mm + 1);
- __m128i B2 = _mm_loadu_si128(in_mm + 2);
- __m128i B3 = _mm_loadu_si128(in_mm + 3);
-
- transpose(B0, B1, B2, B3);
+ SIMD_32::transpose(B0, B1, B2, B3);
key_xor( 0,B0,B1,B2,B3); SBoxE1(B0,B1,B2,B3); transform(B0,B1,B2,B3);
key_xor( 1,B0,B1,B2,B3); SBoxE2(B0,B1,B2,B3); transform(B0,B1,B2,B3);
@@ -131,33 +103,27 @@ void serpent_encrypt_4(const byte in[64],
key_xor(30,B0,B1,B2,B3); SBoxE7(B0,B1,B2,B3); transform(B0,B1,B2,B3);
key_xor(31,B0,B1,B2,B3); SBoxE8(B0,B1,B2,B3); key_xor(32,B0,B1,B2,B3);
- transpose(B0, B1, B2, B3);
+ SIMD_32::transpose(B0, B1, B2, B3);
- _mm_storeu_si128(out_mm , B0);
- _mm_storeu_si128(out_mm + 1, B1);
- _mm_storeu_si128(out_mm + 2, B2);
- _mm_storeu_si128(out_mm + 3, B3);
+ B0.store_le(out);
+ B1.store_le(out + 16);
+ B2.store_le(out + 32);
+ B3.store_le(out + 48);
}
/*
-* SSE2 Serpent Decryption of 4 blocks in parallel
+* SIMD Serpent Decryption of 4 blocks in parallel
*/
void serpent_decrypt_4(const byte in[64],
byte out[64],
- const u32bit keys_32[132])
+ const u32bit keys[132])
{
- const __m128i all_ones = _mm_set1_epi8(0xFF);
-
- const __m128i* keys = (const __m128i*)(keys_32);
- __m128i* out_mm = (__m128i*)(out);
- __m128i* in_mm = (__m128i*)(in);
-
- __m128i B0 = _mm_loadu_si128(in_mm);
- __m128i B1 = _mm_loadu_si128(in_mm + 1);
- __m128i B2 = _mm_loadu_si128(in_mm + 2);
- __m128i B3 = _mm_loadu_si128(in_mm + 3);
+ SIMD_32 B0 = SIMD_32::load_le(in);
+ SIMD_32 B1 = SIMD_32::load_le(in + 16);
+ SIMD_32 B2 = SIMD_32::load_le(in + 32);
+ SIMD_32 B3 = SIMD_32::load_le(in + 48);
- transpose(B0, B1, B2, B3);
+ SIMD_32::transpose(B0, B1, B2, B3);
key_xor(32,B0,B1,B2,B3); SBoxD8(B0,B1,B2,B3); key_xor(31,B0,B1,B2,B3);
i_transform(B0,B1,B2,B3); SBoxD7(B0,B1,B2,B3); key_xor(30,B0,B1,B2,B3);
@@ -195,12 +161,12 @@ void serpent_decrypt_4(const byte in[64],
i_transform(B0,B1,B2,B3); SBoxD2(B0,B1,B2,B3); key_xor( 1,B0,B1,B2,B3);
i_transform(B0,B1,B2,B3); SBoxD1(B0,B1,B2,B3); key_xor( 0,B0,B1,B2,B3);
- transpose(B0, B1, B2, B3);
+ SIMD_32::transpose(B0, B1, B2, B3);
- _mm_storeu_si128(out_mm , B0);
- _mm_storeu_si128(out_mm + 1, B1);
- _mm_storeu_si128(out_mm + 2, B2);
- _mm_storeu_si128(out_mm + 3, B3);
+ B0.store_le(out);
+ B1.store_le(out + 16);
+ B2.store_le(out + 32);
+ B3.store_le(out + 48);
}
}
@@ -208,7 +174,7 @@ void serpent_decrypt_4(const byte in[64],
/*
* Serpent Encryption
*/
-void Serpent_SSE2::encrypt_n(const byte in[], byte out[], u32bit blocks) const
+void Serpent_SIMD::encrypt_n(const byte in[], byte out[], u32bit blocks) const
{
while(blocks >= 4)
{
@@ -224,7 +190,7 @@ void Serpent_SSE2::encrypt_n(const byte in[], byte out[], u32bit blocks) const
/*
* Serpent Decryption
*/
-void Serpent_SSE2::decrypt_n(const byte in[], byte out[], u32bit blocks) const
+void Serpent_SIMD::decrypt_n(const byte in[], byte out[], u32bit blocks) const
{
while(blocks >= 4)
{
diff --git a/src/block/serpent_sse2/serp_sse2.h b/src/block/serpent_simd/serp_simd.h
index f1e5c2028..1ecb70159 100644
--- a/src/block/serpent_sse2/serp_sse2.h
+++ b/src/block/serpent_simd/serp_simd.h
@@ -1,12 +1,12 @@
/*
-* Serpent (SSE2)
+* Serpent (SIMD)
* (C) 2009 Jack Lloyd
*
* Distributed under the terms of the Botan license
*/
-#ifndef BOTAN_SERPENT_SSE2_H__
-#define BOTAN_SERPENT_SSE2_H__
+#ifndef BOTAN_SERPENT_SIMD_H__
+#define BOTAN_SERPENT_SIMD_H__
#include <botan/serpent.h>
@@ -15,13 +15,13 @@ namespace Botan {
/*
* Serpent
*/
-class BOTAN_DLL Serpent_SSE2 : public Serpent
+class BOTAN_DLL Serpent_SIMD : public Serpent
{
public:
void encrypt_n(const byte in[], byte out[], u32bit blocks) const;
void decrypt_n(const byte in[], byte out[], u32bit blocks) const;
- BlockCipher* clone() const { return new Serpent_SSE2; }
+ BlockCipher* clone() const { return new Serpent_SIMD; }
};
}
diff --git a/src/block/serpent_simd/serp_simd_sbox.h b/src/block/serpent_simd/serp_simd_sbox.h
new file mode 100644
index 000000000..6e3da7359
--- /dev/null
+++ b/src/block/serpent_simd/serp_simd_sbox.h
@@ -0,0 +1,426 @@
+/*
+* Serpent Sboxes in SIMD form
+* (C) 2009 Jack Lloyd
+*
+* Distributed under the terms of the Botan license
+*/
+
+#ifndef SERPENT_SIMD_SBOXES_H__
+#define SERPENT_SIMD_SBOXES_H__
+
+#define SBoxE1(B0, B1, B2, B3) \
+ do { \
+ B3 ^= B0; \
+ SIMD_32 B4 = B1; \
+ B1 &= B3; \
+ B4 ^= B2; \
+ B1 ^= B0; \
+ B0 |= B3; \
+ B0 ^= B4; \
+ B4 ^= B3; \
+ B3 ^= B2; \
+ B2 |= B1; \
+ B2 ^= B4; \
+ B4 = ~B4; \
+ B4 |= B1; \
+ B1 ^= B3; \
+ B1 ^= B4; \
+ B3 |= B0; \
+ B1 ^= B3; \
+ B4 ^= B3; \
+ B3 = B0; \
+ B0 = B1; \
+ B1 = B4; \
+ } while(0);
+
+#define SBoxE2(B0, B1, B2, B3) \
+ do { \
+ B0 = ~B0; \
+ B2 = ~B2; \
+ SIMD_32 B4 = B0; \
+ B0 &= B1; \
+ B2 ^= B0; \
+ B0 |= B3; \
+ B3 ^= B2; \
+ B1 ^= B0; \
+ B0 ^= B4; \
+ B4 |= B1; \
+ B1 ^= B3; \
+ B2 |= B0; \
+ B2 &= B4; \
+ B0 ^= B1; \
+ B1 &= B2; \
+ B1 ^= B0; \
+ B0 &= B2; \
+ B4 ^= B0; \
+ B0 = B2; \
+ B2 = B3; \
+ B3 = B1; \
+ B1 = B4; \
+ } while(0);
+
+#define SBoxE3(B0, B1, B2, B3) \
+ do { \
+ SIMD_32 B4 = B0; \
+ B0 &= B2; \
+ B0 ^= B3; \
+ B2 ^= B1; \
+ B2 ^= B0; \
+ B3 |= B4; \
+ B3 ^= B1; \
+ B4 ^= B2; \
+ B1 = B3; \
+ B3 |= B4; \
+ B3 ^= B0; \
+ B0 &= B1; \
+ B4 ^= B0; \
+ B1 ^= B3; \
+ B1 ^= B4; \
+ B4 = ~B4; \
+ B0 = B2; \
+ B2 = B1; \
+ B1 = B3; \
+ B3 = B4; \
+ } while(0);
+
+#define SBoxE4(B0, B1, B2, B3) \
+ do { \
+ SIMD_32 B4 = B0; \
+ B0 |= B3; \
+ B3 ^= B1; \
+ B1 &= B4; \
+ B4 ^= B2; \
+ B2 ^= B3; \
+ B3 &= B0; \
+ B4 |= B1; \
+ B3 ^= B4; \
+ B0 ^= B1; \
+ B4 &= B0; \
+ B1 ^= B3; \
+ B4 ^= B2; \
+ B1 |= B0; \
+ B1 ^= B2; \
+ B0 ^= B3; \
+ B2 = B1; \
+ B1 |= B3; \
+ B0 ^= B1; \
+ B1 = B2; \
+ B2 = B3; \
+ B3 = B4; \
+ } while(0);
+
+#define SBoxE5(B0, B1, B2, B3) \
+ do { \
+ B1 ^= B3; \
+ B3 = ~B3; \
+ B2 ^= B3; \
+ B3 ^= B0; \
+ SIMD_32 B4 = B1; \
+ B1 &= B3; \
+ B1 ^= B2; \
+ B4 ^= B3; \
+ B0 ^= B4; \
+ B2 &= B4; \
+ B2 ^= B0; \
+ B0 &= B1; \
+ B3 ^= B0; \
+ B4 |= B1; \
+ B4 ^= B0; \
+ B0 |= B3; \
+ B0 ^= B2; \
+ B2 &= B3; \
+ B0 = ~B0; \
+ B4 ^= B2; \
+ B2 = B0; \
+ B0 = B1; \
+ B1 = B4; \
+ } while(0);
+
+#define SBoxE6(B0, B1, B2, B3) \
+ do { \
+ B0 ^= B1; \
+ B1 ^= B3; \
+ B3 = ~B3; \
+ SIMD_32 B4 = B1; \
+ B1 &= B0; \
+ B2 ^= B3; \
+ B1 ^= B2; \
+ B2 |= B4; \
+ B4 ^= B3; \
+ B3 &= B1; \
+ B3 ^= B0; \
+ B4 ^= B1; \
+ B4 ^= B2; \
+ B2 ^= B0; \
+ B0 &= B3; \
+ B2 = ~B2; \
+ B0 ^= B4; \
+ B4 |= B3; \
+ B4 ^= B2; \
+ B2 = B0; \
+ B0 = B1; \
+ B1 = B3; \
+ B3 = B4; \
+ } while(0);
+
+#define SBoxE7(B0, B1, B2, B3) \
+ do { \
+ B2 = ~B2; \
+ SIMD_32 B4 = B3; \
+ B3 &= B0; \
+ B0 ^= B4; \
+ B3 ^= B2; \
+ B2 |= B4; \
+ B1 ^= B3; \
+ B2 ^= B0; \
+ B0 |= B1; \
+ B2 ^= B1; \
+ B4 ^= B0; \
+ B0 |= B3; \
+ B0 ^= B2; \
+ B4 ^= B3; \
+ B4 ^= B0; \
+ B3 = ~B3; \
+ B2 &= B4; \
+ B3 ^= B2; \
+ B2 = B4; \
+ } while(0);
+
+#define SBoxE8(B0, B1, B2, B3) \
+ do { \
+ SIMD_32 B4 = B1; \
+ B1 |= B2; \
+ B1 ^= B3; \
+ B4 ^= B2; \
+ B2 ^= B1; \
+ B3 |= B4; \
+ B3 &= B0; \
+ B4 ^= B2; \
+ B3 ^= B1; \
+ B1 |= B4; \
+ B1 ^= B0; \
+ B0 |= B4; \
+ B0 ^= B2; \
+ B1 ^= B4; \
+ B2 ^= B1; \
+ B1 &= B0; \
+ B1 ^= B4; \
+ B2 = ~B2; \
+ B2 |= B0; \
+ B4 ^= B2; \
+ B2 = B1; \
+ B1 = B3; \
+ B3 = B0; \
+ B0 = B4; \
+ } while(0);
+
+#define SBoxD1(B0, B1, B2, B3) \
+ do { \
+ B2 = ~B2; \
+ SIMD_32 B4 = B1; \
+ B1 |= B0; \
+ B4 = ~B4; \
+ B1 ^= B2; \
+ B2 |= B4; \
+ B1 ^= B3; \
+ B0 ^= B4; \
+ B2 ^= B0; \
+ B0 &= B3; \
+ B4 ^= B0; \
+ B0 |= B1; \
+ B0 ^= B2; \
+ B3 ^= B4; \
+ B2 ^= B1; \
+ B3 ^= B0; \
+ B3 ^= B1; \
+ B2 &= B3; \
+ B4 ^= B2; \
+ B2 = B1; \
+ B1 = B4; \
+ } while(0);
+
+#define SBoxD2(B0, B1, B2, B3) \
+ do { \
+ SIMD_32 B4 = B1; \
+ B1 ^= B3; \
+ B3 &= B1; \
+ B4 ^= B2; \
+ B3 ^= B0; \
+ B0 |= B1; \
+ B2 ^= B3; \
+ B0 ^= B4; \
+ B0 |= B2; \
+ B1 ^= B3; \
+ B0 ^= B1; \
+ B1 |= B3; \
+ B1 ^= B0; \
+ B4 = ~B4; \
+ B4 ^= B1; \
+ B1 |= B0; \
+ B1 ^= B0; \
+ B1 |= B4; \
+ B3 ^= B1; \
+ B1 = B0; \
+ B0 = B4; \
+ B4 = B2; \
+ B2 = B3; \
+ B3 = B4; \
+ } while(0);
+
+#define SBoxD3(B0, B1, B2, B3) \
+ do { \
+ B2 ^= B3; \
+ B3 ^= B0; \
+ SIMD_32 B4 = B3; \
+ B3 &= B2; \
+ B3 ^= B1; \
+ B1 |= B2; \
+ B1 ^= B4; \
+ B4 &= B3; \
+ B2 ^= B3; \
+ B4 &= B0; \
+ B4 ^= B2; \
+ B2 &= B1; \
+ B2 |= B0; \
+ B3 = ~B3; \
+ B2 ^= B3; \
+ B0 ^= B3; \
+ B0 &= B1; \
+ B3 ^= B4; \
+ B3 ^= B0; \
+ B0 = B1; \
+ B1 = B4; \
+ } while(0);
+
+#define SBoxD4(B0, B1, B2, B3) \
+ do { \
+ SIMD_32 B4 = B2; \
+ B2 ^= B1; \
+ B0 ^= B2; \
+ B4 &= B2; \
+ B4 ^= B0; \
+ B0 &= B1; \
+ B1 ^= B3; \
+ B3 |= B4; \
+ B2 ^= B3; \
+ B0 ^= B3; \
+ B1 ^= B4; \
+ B3 &= B2; \
+ B3 ^= B1; \
+ B1 ^= B0; \
+ B1 |= B2; \
+ B0 ^= B3; \
+ B1 ^= B4; \
+ B0 ^= B1; \
+ B4 = B0; \
+ B0 = B2; \
+ B2 = B3; \
+ B3 = B4; \
+ } while(0);
+
+#define SBoxD5(B0, B1, B2, B3) \
+ do { \
+ SIMD_32 B4 = B2; \
+ B2 &= B3; \
+ B2 ^= B1; \
+ B1 |= B3; \
+ B1 &= B0; \
+ B4 ^= B2; \
+ B4 ^= B1; \
+ B1 &= B2; \
+ B0 = ~B0; \
+ B3 ^= B4; \
+ B1 ^= B3; \
+ B3 &= B0; \
+ B3 ^= B2; \
+ B0 ^= B1; \
+ B2 &= B0; \
+ B3 ^= B0; \
+ B2 ^= B4; \
+ B2 |= B3; \
+ B3 ^= B0; \
+ B2 ^= B1; \
+ B1 = B3; \
+ B3 = B4; \
+ } while(0);
+
+#define SBoxD6(B0, B1, B2, B3) \
+ do { \
+ B1 = ~B1; \
+ SIMD_32 B4 = B3; \
+ B2 ^= B1; \
+ B3 |= B0; \
+ B3 ^= B2; \
+ B2 |= B1; \
+ B2 &= B0; \
+ B4 ^= B3; \
+ B2 ^= B4; \
+ B4 |= B0; \
+ B4 ^= B1; \
+ B1 &= B2; \
+ B1 ^= B3; \
+ B4 ^= B2; \
+ B3 &= B4; \
+ B4 ^= B1; \
+ B3 ^= B4; \
+ B4 = ~B4; \
+ B3 ^= B0; \
+ B0 = B1; \
+ B1 = B4; \
+ B4 = B3; \
+ B3 = B2; \
+ B2 = B4; \
+ } while(0);
+
+#define SBoxD7(B0, B1, B2, B3) \
+ do { \
+ B0 ^= B2; \
+ SIMD_32 B4 = B2; \
+ B2 &= B0; \
+ B4 ^= B3; \
+ B2 = ~B2; \
+ B3 ^= B1; \
+ B2 ^= B3; \
+ B4 |= B0; \
+ B0 ^= B2; \
+ B3 ^= B4; \
+ B4 ^= B1; \
+ B1 &= B3; \
+ B1 ^= B0; \
+ B0 ^= B3; \
+ B0 |= B2; \
+ B3 ^= B1; \
+ B4 ^= B0; \
+ B0 = B1; \
+ B1 = B2; \
+ B2 = B4; \
+ } while(0);
+
+#define SBoxD8(B0, B1, B2, B3) \
+ do { \
+ SIMD_32 B4 = B2; \
+ B2 ^= B0; \
+ B0 &= B3; \
+ B4 |= B3; \
+ B2 = ~B2; \
+ B3 ^= B1; \
+ B1 |= B0; \
+ B0 ^= B2; \
+ B2 &= B4; \
+ B3 &= B4; \
+ B1 ^= B2; \
+ B2 ^= B0; \
+ B0 |= B2; \
+ B4 ^= B1; \
+ B0 ^= B3; \
+ B3 ^= B4; \
+ B4 |= B0; \
+ B3 ^= B2; \
+ B4 ^= B2; \
+ B2 = B1; \
+ B1 = B0; \
+ B0 = B3; \
+ B3 = B4; \
+ } while(0);
+
+#endif
diff --git a/src/block/serpent_sse2/info.txt b/src/block/serpent_sse2/info.txt
deleted file mode 100644
index a4ec561a8..000000000
--- a/src/block/serpent_sse2/info.txt
+++ /dev/null
@@ -1,6 +0,0 @@
-define SERPENT_SSE2
-
-<requires>
-serpent
-sse2_eng
-</requires>
diff --git a/src/block/serpent_sse2/serp_sse2_sbox.h b/src/block/serpent_sse2/serp_sse2_sbox.h
deleted file mode 100644
index 40c552e87..000000000
--- a/src/block/serpent_sse2/serp_sse2_sbox.h
+++ /dev/null
@@ -1,434 +0,0 @@
-/*
-* Serpent Sboxes in SSE2 form
-* (C) 2009 Jack Lloyd
-*
-* Distributed under the terms of the Botan license
-*/
-
-#ifndef SERPENT_SSE2_SBOXES_H__
-#define SERPENT_SSE2_SBOXES_H__
-
-#define SBoxE1(B0, B1, B2, B3) \
- do { \
- B3 = _mm_xor_si128(B3, B0); \
- __m128i B4 = B1; \
- B1 = _mm_and_si128(B1, B3); \
- B4 = _mm_xor_si128(B4, B2); \
- B1 = _mm_xor_si128(B1, B0); \
- B0 = _mm_or_si128(B0, B3); \
- B0 = _mm_xor_si128(B0, B4); \
- B4 = _mm_xor_si128(B4, B3); \
- B3 = _mm_xor_si128(B3, B2); \
- B2 = _mm_or_si128(B2, B1); \
- B2 = _mm_xor_si128(B2, B4); \
- B4 = _mm_xor_si128(B4, all_ones); \
- B4 = _mm_or_si128(B4, B1); \
- B1 = _mm_xor_si128(B1, B3); \
- B1 = _mm_xor_si128(B1, B4); \
- B3 = _mm_or_si128(B3, B0); \
- B1 = _mm_xor_si128(B1, B3); \
- B4 = _mm_xor_si128(B4, B3); \
- B3 = B0; \
- B0 = B1; \
- B1 = B4; \
- } while(0);
-
-#define SBoxE2(B0, B1, B2, B3) \
- do { \
- B0 = _mm_xor_si128(B0, all_ones); \
- B2 = _mm_xor_si128(B2, all_ones); \
- __m128i B4 = B0; \
- B0 = _mm_and_si128(B0, B1); \
- B2 = _mm_xor_si128(B2, B0); \
- B0 = _mm_or_si128(B0, B3); \
- B3 = _mm_xor_si128(B3, B2); \
- B1 = _mm_xor_si128(B1, B0); \
- B0 = _mm_xor_si128(B0, B4); \
- B4 = _mm_or_si128(B4, B1); \
- B1 = _mm_xor_si128(B1, B3); \
- B2 = _mm_or_si128(B2, B0); \
- B2 = _mm_and_si128(B2, B4); \
- B0 = _mm_xor_si128(B0, B1); \
- B1 = _mm_and_si128(B1, B2); \
- B1 = _mm_xor_si128(B1, B0); \
- B0 = _mm_and_si128(B0, B2); \
- B4 = _mm_xor_si128(B4, B0); \
- B0 = B2; \
- B2 = B3; \
- B3 = B1; \
- B1 = B4; \
- } while(0);
-
-#define SBoxE3(B0, B1, B2, B3) \
- do { \
- __m128i B4 = B0; \
- B0 = _mm_and_si128(B0, B2); \
- B0 = _mm_xor_si128(B0, B3); \
- B2 = _mm_xor_si128(B2, B1); \
- B2 = _mm_xor_si128(B2, B0); \
- B3 = _mm_or_si128(B3, B4); \
- B3 = _mm_xor_si128(B3, B1); \
- B4 = _mm_xor_si128(B4, B2); \
- B1 = B3; \
- B3 = _mm_or_si128(B3, B4); \
- B3 = _mm_xor_si128(B3, B0); \
- B0 = _mm_and_si128(B0, B1); \
- B4 = _mm_xor_si128(B4, B0); \
- B1 = _mm_xor_si128(B1, B3); \
- B1 = _mm_xor_si128(B1, B4); \
- B4 = _mm_xor_si128(B4, all_ones); \
- B0 = B2; \
- B2 = B1; \
- B1 = B3; \
- B3 = B4; \
- } while(0);
-
-#define SBoxE4(B0, B1, B2, B3) \
- do { \
- __m128i B4 = B0; \
- B0 = _mm_or_si128(B0, B3); \
- B3 = _mm_xor_si128(B3, B1); \
- B1 = _mm_and_si128(B1, B4); \
- B4 = _mm_xor_si128(B4, B2); \
- B2 = _mm_xor_si128(B2, B3); \
- B3 = _mm_and_si128(B3, B0); \
- B4 = _mm_or_si128(B4, B1); \
- B3 = _mm_xor_si128(B3, B4); \
- B0 = _mm_xor_si128(B0, B1); \
- B4 = _mm_and_si128(B4, B0); \
- B1 = _mm_xor_si128(B1, B3); \
- B4 = _mm_xor_si128(B4, B2); \
- B1 = _mm_or_si128(B1, B0); \
- B1 = _mm_xor_si128(B1, B2); \
- B0 = _mm_xor_si128(B0, B3); \
- B2 = B1; \
- B1 = _mm_or_si128(B1, B3); \
- B0 = _mm_xor_si128(B0, B1); \
- B1 = B2; \
- B2 = B3; \
- B3 = B4; \
- } while(0);
-
-#define SBoxE5(B0, B1, B2, B3) \
- do { \
- B1 = _mm_xor_si128(B1, B3); \
- B3 = _mm_xor_si128(B3, all_ones); \
- B2 = _mm_xor_si128(B2, B3); \
- B3 = _mm_xor_si128(B3, B0); \
- __m128i B4 = B1; \
- B1 = _mm_and_si128(B1, B3); \
- B1 = _mm_xor_si128(B1, B2); \
- B4 = _mm_xor_si128(B4, B3); \
- B0 = _mm_xor_si128(B0, B4); \
- B2 = _mm_and_si128(B2, B4); \
- B2 = _mm_xor_si128(B2, B0); \
- B0 = _mm_and_si128(B0, B1); \
- B3 = _mm_xor_si128(B3, B0); \
- B4 = _mm_or_si128(B4, B1); \
- B4 = _mm_xor_si128(B4, B0); \
- B0 = _mm_or_si128(B0, B3); \
- B0 = _mm_xor_si128(B0, B2); \
- B2 = _mm_and_si128(B2, B3); \
- B0 = _mm_xor_si128(B0, all_ones); \
- B4 = _mm_xor_si128(B4, B2); \
- B2 = B0; \
- B0 = B1; \
- B1 = B4; \
- } while(0);
-
-#define SBoxE6(B0, B1, B2, B3) \
- do { \
- B0 = _mm_xor_si128(B0, B1); \
- B1 = _mm_xor_si128(B1, B3); \
- B3 = _mm_xor_si128(B3, all_ones); \
- __m128i B4 = B1; \
- B1 = _mm_and_si128(B1, B0); \
- B2 = _mm_xor_si128(B2, B3); \
- B1 = _mm_xor_si128(B1, B2); \
- B2 = _mm_or_si128(B2, B4); \
- B4 = _mm_xor_si128(B4, B3); \
- B3 = _mm_and_si128(B3, B1); \
- B3 = _mm_xor_si128(B3, B0); \
- B4 = _mm_xor_si128(B4, B1); \
- B4 = _mm_xor_si128(B4, B2); \
- B2 = _mm_xor_si128(B2, B0); \
- B0 = _mm_and_si128(B0, B3); \
- B2 = _mm_xor_si128(B2, all_ones); \
- B0 = _mm_xor_si128(B0, B4); \
- B4 = _mm_or_si128(B4, B3); \
- B4 = _mm_xor_si128(B4, B2); \
- B2 = B0; \
- B0 = B1; \
- B1 = B3; \
- B3 = B4; \
- } while(0);
-
-#define SBoxE7(B0, B1, B2, B3) \
- do { \
- B2 = _mm_xor_si128(B2, all_ones); \
- __m128i B4 = B3; \
- B3 = _mm_and_si128(B3, B0); \
- B0 = _mm_xor_si128(B0, B4); \
- B3 = _mm_xor_si128(B3, B2); \
- B2 = _mm_or_si128(B2, B4); \
- B1 = _mm_xor_si128(B1, B3); \
- B2 = _mm_xor_si128(B2, B0); \
- B0 = _mm_or_si128(B0, B1); \
- B2 = _mm_xor_si128(B2, B1); \
- B4 = _mm_xor_si128(B4, B0); \
- B0 = _mm_or_si128(B0, B3); \
- B0 = _mm_xor_si128(B0, B2); \
- B4 = _mm_xor_si128(B4, B3); \
- B4 = _mm_xor_si128(B4, B0); \
- B3 = _mm_xor_si128(B3, all_ones); \
- B2 = _mm_and_si128(B2, B4); \
- B3 = _mm_xor_si128(B3, B2); \
- B2 = B4; \
- } while(0);
-
-#define SBoxE8(B0, B1, B2, B3) \
- do { \
- __m128i B4 = B1; \
- B1 = _mm_or_si128(B1, B2); \
- B1 = _mm_xor_si128(B1, B3); \
- B4 = _mm_xor_si128(B4, B2); \
- B2 = _mm_xor_si128(B2, B1); \
- B3 = _mm_or_si128(B3, B4); \
- B3 = _mm_and_si128(B3, B0); \
- B4 = _mm_xor_si128(B4, B2); \
- B3 = _mm_xor_si128(B3, B1); \
- B1 = _mm_or_si128(B1, B4); \
- B1 = _mm_xor_si128(B1, B0); \
- B0 = _mm_or_si128(B0, B4); \
- B0 = _mm_xor_si128(B0, B2); \
- B1 = _mm_xor_si128(B1, B4); \
- B2 = _mm_xor_si128(B2, B1); \
- B1 = _mm_and_si128(B1, B0); \
- B1 = _mm_xor_si128(B1, B4); \
- B2 = _mm_xor_si128(B2, all_ones); \
- B2 = _mm_or_si128(B2, B0); \
- B4 = _mm_xor_si128(B4, B2); \
- B2 = B1; \
- B1 = B3; \
- B3 = B0; \
- B0 = B4; \
- } while(0);
-
-#define SBoxD1(B0, B1, B2, B3) \
- do \
- { \
- B2 = _mm_xor_si128(B2, all_ones); \
- __m128i B4 = B1; \
- B1 = _mm_or_si128(B1, B0); \
- B4 = _mm_xor_si128(B4, all_ones); \
- B1 = _mm_xor_si128(B1, B2); \
- B2 = _mm_or_si128(B2, B4); \
- B1 = _mm_xor_si128(B1, B3); \
- B0 = _mm_xor_si128(B0, B4); \
- B2 = _mm_xor_si128(B2, B0); \
- B0 = _mm_and_si128(B0, B3); \
- B4 = _mm_xor_si128(B4, B0); \
- B0 = _mm_or_si128(B0, B1); \
- B0 = _mm_xor_si128(B0, B2); \
- B3 = _mm_xor_si128(B3, B4); \
- B2 = _mm_xor_si128(B2, B1); \
- B3 = _mm_xor_si128(B3, B0); \
- B3 = _mm_xor_si128(B3, B1); \
- B2 = _mm_and_si128(B2, B3); \
- B4 = _mm_xor_si128(B4, B2); \
- B2 = B1; \
- B1 = B4; \
- } while(0);
-
-#define SBoxD2(B0, B1, B2, B3) \
- do \
- { \
- __m128i B4 = B1; \
- B1 = _mm_xor_si128(B1, B3); \
- B3 = _mm_and_si128(B3, B1); \
- B4 = _mm_xor_si128(B4, B2); \
- B3 = _mm_xor_si128(B3, B0); \
- B0 = _mm_or_si128(B0, B1); \
- B2 = _mm_xor_si128(B2, B3); \
- B0 = _mm_xor_si128(B0, B4); \
- B0 = _mm_or_si128(B0, B2); \
- B1 = _mm_xor_si128(B1, B3); \
- B0 = _mm_xor_si128(B0, B1); \
- B1 = _mm_or_si128(B1, B3); \
- B1 = _mm_xor_si128(B1, B0); \
- B4 = _mm_xor_si128(B4, all_ones); \
- B4 = _mm_xor_si128(B4, B1); \
- B1 = _mm_or_si128(B1, B0); \
- B1 = _mm_xor_si128(B1, B0); \
- B1 = _mm_or_si128(B1, B4); \
- B3 = _mm_xor_si128(B3, B1); \
- B1 = B0; \
- B0 = B4; \
- B4 = B2; \
- B2 = B3; \
- B3 = B4; \
- } while(0);
-
-#define SBoxD3(B0, B1, B2, B3) \
- do \
- { \
- B2 = _mm_xor_si128(B2, B3); \
- B3 = _mm_xor_si128(B3, B0); \
- __m128i B4 = B3; \
- B3 = _mm_and_si128(B3, B2); \
- B3 = _mm_xor_si128(B3, B1); \
- B1 = _mm_or_si128(B1, B2); \
- B1 = _mm_xor_si128(B1, B4); \
- B4 = _mm_and_si128(B4, B3); \
- B2 = _mm_xor_si128(B2, B3); \
- B4 = _mm_and_si128(B4, B0); \
- B4 = _mm_xor_si128(B4, B2); \
- B2 = _mm_and_si128(B2, B1); \
- B2 = _mm_or_si128(B2, B0); \
- B3 = _mm_xor_si128(B3, all_ones); \
- B2 = _mm_xor_si128(B2, B3); \
- B0 = _mm_xor_si128(B0, B3); \
- B0 = _mm_and_si128(B0, B1); \
- B3 = _mm_xor_si128(B3, B4); \
- B3 = _mm_xor_si128(B3, B0); \
- B0 = B1; \
- B1 = B4; \
- } while(0);
-
-#define SBoxD4(B0, B1, B2, B3) \
- do \
- { \
- __m128i B4 = B2; \
- B2 = _mm_xor_si128(B2, B1); \
- B0 = _mm_xor_si128(B0, B2); \
- B4 = _mm_and_si128(B4, B2); \
- B4 = _mm_xor_si128(B4, B0); \
- B0 = _mm_and_si128(B0, B1); \
- B1 = _mm_xor_si128(B1, B3); \
- B3 = _mm_or_si128(B3, B4); \
- B2 = _mm_xor_si128(B2, B3); \
- B0 = _mm_xor_si128(B0, B3); \
- B1 = _mm_xor_si128(B1, B4); \
- B3 = _mm_and_si128(B3, B2); \
- B3 = _mm_xor_si128(B3, B1); \
- B1 = _mm_xor_si128(B1, B0); \
- B1 = _mm_or_si128(B1, B2); \
- B0 = _mm_xor_si128(B0, B3); \
- B1 = _mm_xor_si128(B1, B4); \
- B0 = _mm_xor_si128(B0, B1); \
- B4 = B0; \
- B0 = B2; \
- B2 = B3; \
- B3 = B4; \
- } while(0);
-
-#define SBoxD5(B0, B1, B2, B3) \
- do \
- { \
- __m128i B4 = B2; \
- B2 = _mm_and_si128(B2, B3); \
- B2 = _mm_xor_si128(B2, B1); \
- B1 = _mm_or_si128(B1, B3); \
- B1 = _mm_and_si128(B1, B0); \
- B4 = _mm_xor_si128(B4, B2); \
- B4 = _mm_xor_si128(B4, B1); \
- B1 = _mm_and_si128(B1, B2); \
- B0 = _mm_xor_si128(B0, all_ones); \
- B3 = _mm_xor_si128(B3, B4); \
- B1 = _mm_xor_si128(B1, B3); \
- B3 = _mm_and_si128(B3, B0); \
- B3 = _mm_xor_si128(B3, B2); \
- B0 = _mm_xor_si128(B0, B1); \
- B2 = _mm_and_si128(B2, B0); \
- B3 = _mm_xor_si128(B3, B0); \
- B2 = _mm_xor_si128(B2, B4); \
- B2 = _mm_or_si128(B2, B3); \
- B3 = _mm_xor_si128(B3, B0); \
- B2 = _mm_xor_si128(B2, B1); \
- B1 = B3; \
- B3 = B4; \
- } while(0);
-
-#define SBoxD6(B0, B1, B2, B3) \
- do \
- { \
- B1 = _mm_xor_si128(B1, all_ones); \
- __m128i B4 = B3; \
- B2 = _mm_xor_si128(B2, B1); \
- B3 = _mm_or_si128(B3, B0); \
- B3 = _mm_xor_si128(B3, B2); \
- B2 = _mm_or_si128(B2, B1); \
- B2 = _mm_and_si128(B2, B0); \
- B4 = _mm_xor_si128(B4, B3); \
- B2 = _mm_xor_si128(B2, B4); \
- B4 = _mm_or_si128(B4, B0); \
- B4 = _mm_xor_si128(B4, B1); \
- B1 = _mm_and_si128(B1, B2); \
- B1 = _mm_xor_si128(B1, B3); \
- B4 = _mm_xor_si128(B4, B2); \
- B3 = _mm_and_si128(B3, B4); \
- B4 = _mm_xor_si128(B4, B1); \
- B3 = _mm_xor_si128(B3, B4); \
- B4 = _mm_xor_si128(B4, all_ones); \
- B3 = _mm_xor_si128(B3, B0); \
- B0 = B1; \
- B1 = B4; \
- B4 = B3; \
- B3 = B2; \
- B2 = B4; \
- } while(0);
-
-#define SBoxD7(B0, B1, B2, B3) \
- do \
- { \
- B0 = _mm_xor_si128(B0, B2); \
- __m128i B4 = B2; \
- B2 = _mm_and_si128(B2, B0); \
- B4 = _mm_xor_si128(B4, B3); \
- B2 = _mm_xor_si128(B2, all_ones); \
- B3 = _mm_xor_si128(B3, B1); \
- B2 = _mm_xor_si128(B2, B3); \
- B4 = _mm_or_si128(B4, B0); \
- B0 = _mm_xor_si128(B0, B2); \
- B3 = _mm_xor_si128(B3, B4); \
- B4 = _mm_xor_si128(B4, B1); \
- B1 = _mm_and_si128(B1, B3); \
- B1 = _mm_xor_si128(B1, B0); \
- B0 = _mm_xor_si128(B0, B3); \
- B0 = _mm_or_si128(B0, B2); \
- B3 = _mm_xor_si128(B3, B1); \
- B4 = _mm_xor_si128(B4, B0); \
- B0 = B1; \
- B1 = B2; \
- B2 = B4; \
- } while(0);
-
-#define SBoxD8(B0, B1, B2, B3) \
- do \
- { \
- __m128i B4 = B2; \
- B2 = _mm_xor_si128(B2, B0); \
- B0 = _mm_and_si128(B0, B3); \
- B4 = _mm_or_si128(B4, B3); \
- B2 = _mm_xor_si128(B2, all_ones); \
- B3 = _mm_xor_si128(B3, B1); \
- B1 = _mm_or_si128(B1, B0); \
- B0 = _mm_xor_si128(B0, B2); \
- B2 = _mm_and_si128(B2, B4); \
- B3 = _mm_and_si128(B3, B4); \
- B1 = _mm_xor_si128(B1, B2); \
- B2 = _mm_xor_si128(B2, B0); \
- B0 = _mm_or_si128(B0, B2); \
- B4 = _mm_xor_si128(B4, B1); \
- B0 = _mm_xor_si128(B0, B3); \
- B3 = _mm_xor_si128(B3, B4); \
- B4 = _mm_or_si128(B4, B0); \
- B3 = _mm_xor_si128(B3, B2); \
- B4 = _mm_xor_si128(B4, B2); \
- B2 = B1; \
- B1 = B0; \
- B0 = B3; \
- B3 = B4; \
- } while(0);
-
-#endif
diff --git a/src/block/xtea/xtea.h b/src/block/xtea/xtea.h
index f3b554edb..9982d0712 100644
--- a/src/block/xtea/xtea.h
+++ b/src/block/xtea/xtea.h
@@ -26,7 +26,7 @@ class BOTAN_DLL XTEA : public BlockCipher
BlockCipher* clone() const { return new XTEA; }
XTEA() : BlockCipher(8, 16) {}
- private:
+ protected:
void key_schedule(const byte[], u32bit);
SecureBuffer<u32bit, 64> EK;
};
diff --git a/src/block/xtea_simd/info.txt b/src/block/xtea_simd/info.txt
new file mode 100644
index 000000000..0cdce14ef
--- /dev/null
+++ b/src/block/xtea_simd/info.txt
@@ -0,0 +1,16 @@
+realname "XTEA (SIMD)"
+
+define XTEA_SIMD
+
+load_on auto
+
+<add>
+xtea_simd.cpp
+xtea_simd.h
+</add>
+
+<requires>
+xtea
+simd_32
+simd_engine
+</requires>
diff --git a/src/block/xtea_simd/xtea_simd.cpp b/src/block/xtea_simd/xtea_simd.cpp
new file mode 100644
index 000000000..6151c355c
--- /dev/null
+++ b/src/block/xtea_simd/xtea_simd.cpp
@@ -0,0 +1,124 @@
+/*
+* XTEA in SIMD
+* (C) 2009 Jack Lloyd
+*
+* Distributed under the terms of the Botan license
+*/
+
+#include <botan/xtea_simd.h>
+#include <botan/loadstor.h>
+#include <botan/simd_32.h>
+
+namespace Botan {
+
+namespace {
+
+void xtea_encrypt_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[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;
+
+ R0 += (((L0 << 4) ^ (L0 >> 5)) + L0) ^ K1;
+ R1 += (((L1 << 4) ^ (L1 >> 5)) + L1) ^ K1;
+
+ L0 += (((R0 << 4) ^ (R0 >> 5)) + R0) ^ K2;
+ L1 += (((R1 << 4) ^ (R1 >> 5)) + R1) ^ K2;
+
+ R0 += (((L0 << 4) ^ (L0 >> 5)) + L0) ^ K3;
+ R1 += (((L1 << 4) ^ (L1 >> 5)) + L1) ^ 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);
+ }
+
+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);
+ }
+
+}
+
+/*
+* XTEA Encryption
+*/
+void XTEA_SIMD::encrypt_n(const byte in[], byte out[], u32bit blocks) const
+ {
+ while(blocks >= 8)
+ {
+ xtea_encrypt_8(in, out, this->EK);
+ in += 8 * BLOCK_SIZE;
+ out += 8 * BLOCK_SIZE;
+ blocks -= 8;
+ }
+
+ XTEA::encrypt_n(in, out, blocks);
+ }
+
+/*
+* XTEA Decryption
+*/
+void XTEA_SIMD::decrypt_n(const byte in[], byte out[], u32bit blocks) const
+ {
+ while(blocks >= 8)
+ {
+ xtea_decrypt_8(in, out, this->EK);
+ in += 8 * BLOCK_SIZE;
+ out += 8 * BLOCK_SIZE;
+ blocks -= 8;
+ }
+
+ XTEA::decrypt_n(in, out, blocks);
+ }
+
+}
diff --git a/src/block/xtea_simd/xtea_simd.h b/src/block/xtea_simd/xtea_simd.h
new file mode 100644
index 000000000..e4ce734ed
--- /dev/null
+++ b/src/block/xtea_simd/xtea_simd.h
@@ -0,0 +1,28 @@
+/*
+* XTEA in SIMD
+* (C) 2009 Jack Lloyd
+*
+* Distributed under the terms of the Botan license
+*/
+
+#ifndef BOTAN_XTEA_SIMD_H__
+#define BOTAN_XTEA_SIMD_H__
+
+#include <botan/xtea.h>
+
+namespace Botan {
+
+/*
+* XTEA (SIMD variant)
+*/
+class BOTAN_DLL XTEA_SIMD : public XTEA
+ {
+ public:
+ void encrypt_n(const byte in[], byte out[], u32bit blocks) const;
+ void decrypt_n(const byte in[], byte out[], u32bit blocks) const;
+ BlockCipher* clone() const { return new XTEA_SIMD; }
+ };
+
+}
+
+#endif
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-15 22:33:47 +0000