From: Bernd Edlinger Date: Thu, 23 Apr 2020 23:28:07 +0000 (+0200) Subject: Remove AES bitsliced S-box implementation from Boyar and Peralta X-Git-Url: https://git.librecmc.org/?a=commitdiff_plain;h=6b057f75074a9061147d7e64fc77db85e310db47;p=oweals%2Fopenssl.git Remove AES bitsliced S-box implementation from Boyar and Peralta [extended tests] Reviewed-by: Richard Levitte Reviewed-by: Tomas Mraz (Merged from https://github.com/openssl/openssl/pull/11626) --- diff --git a/crypto/aes/aes_core.c b/crypto/aes/aes_core.c index 687dd5829b..ad00c729e7 100644 --- a/crypto/aes/aes_core.c +++ b/crypto/aes/aes_core.c @@ -673,357 +673,6 @@ void AES_decrypt(const unsigned char *in, unsigned char *out, InvCipher(in, out, rk, key->rounds); } - -# ifndef OPENSSL_SMALL_FOOTPRINT -void AES_ctr32_encrypt(const unsigned char *in, unsigned char *out, - size_t blocks, const AES_KEY *key, - const unsigned char *ivec); - -static void RawToBits(const u8 raw[64], u64 bits[8]) -{ - int i, j; - u64 in, out; - - memset(bits, 0, 64); - for (i = 0; i < 8; i++) { - in = 0; - for (j = 0; j < 8; j++) - in |= ((u64)raw[i * 8 + j]) << (8 * j); - out = in & 0xF0F0F0F00F0F0F0FuLL; - out |= (in & 0x0F0F0F0F00000000uLL) >> 28; - out |= (in & 0x00000000F0F0F0F0uLL) << 28; - in = out & 0xCCCC3333CCCC3333uLL; - in |= (out & 0x3333000033330000uLL) >> 14; - in |= (out & 0x0000CCCC0000CCCCuLL) << 14; - out = in & 0xAA55AA55AA55AA55uLL; - out |= (in & 0x5500550055005500uLL) >> 7; - out |= (in & 0x00AA00AA00AA00AAuLL) << 7; - for (j = 0; j < 8; j++) { - bits[j] |= (out & 0xFFuLL) << (8 * i); - out = out >> 8; - } - } -} - -static void BitsToRaw(const u64 bits[8], u8 raw[64]) -{ - int i, j; - u64 in, out; - - for (i = 0; i < 8; i++) { - in = 0; - for (j = 0; j < 8; j++) - in |= ((bits[j] >> (8 * i)) & 0xFFuLL) << (8 * j); - out = in & 0xF0F0F0F00F0F0F0FuLL; - out |= (in & 0x0F0F0F0F00000000uLL) >> 28; - out |= (in & 0x00000000F0F0F0F0uLL) << 28; - in = out & 0xCCCC3333CCCC3333uLL; - in |= (out & 0x3333000033330000uLL) >> 14; - in |= (out & 0x0000CCCC0000CCCCuLL) << 14; - out = in & 0xAA55AA55AA55AA55uLL; - out |= (in & 0x5500550055005500uLL) >> 7; - out |= (in & 0x00AA00AA00AA00AAuLL) << 7; - for (j = 0; j < 8; j++) { - raw[i * 8 + j] = (u8)out; - out = out >> 8; - } - } -} - -static void BitsXtime(u64 state[8]) -{ - u64 b; - - b = state[7]; - state[7] = state[6]; - state[6] = state[5]; - state[5] = state[4]; - state[4] = state[3] ^ b; - state[3] = state[2] ^ b; - state[2] = state[1]; - state[1] = state[0] ^ b; - state[0] = b; -} - -/* - * This S-box implementation follows a circuit described in - * Boyar and Peralta: "A new combinational logic minimization - * technique with applications to cryptology." - * https://eprint.iacr.org/2009/191.pdf - * - * The math is similar to above, in that it uses - * a tower field of GF(2^2^2^2) but with a different - * basis representation, that is better suited to - * logic designs. - */ -static void BitsSub(u64 state[8]) -{ - u64 x0, x1, x2, x3, x4, x5, x6, x7; - u64 y1, y2, y3, y4, y5, y6, y7, y8, y9, y10, y11; - u64 y12, y13, y14, y15, y16, y17, y18, y19, y20, y21; - u64 t0, t1, t2, t3, t4, t5, t6, t7, t8, t9, t10, t11; - u64 t12, t13, t14, t15, t16, t17, t18, t19, t20, t21; - u64 t22, t23, t24, t25, t26, t27, t28, t29, t30, t31; - u64 t32, t33, t34, t35, t36, t37, t38, t39, t40, t41; - u64 t42, t43, t44, t45, t46, t47, t48, t49, t50, t51; - u64 t52, t53, t54, t55, t56, t57, t58, t59, t60, t61; - u64 t62, t63, t64, t65, t66, t67; - u64 z0, z1, z2, z3, z4, z5, z6, z7, z8, z9, z10, z11; - u64 z12, z13, z14, z15, z16, z17; - u64 s0, s1, s2, s3, s4, s5, s6, s7; - - x7 = state[0]; - x6 = state[1]; - x5 = state[2]; - x4 = state[3]; - x3 = state[4]; - x2 = state[5]; - x1 = state[6]; - x0 = state[7]; - y14 = x3 ^ x5; - y13 = x0 ^ x6; - y9 = x0 ^ x3; - y8 = x0 ^ x5; - t0 = x1 ^ x2; - y1 = t0 ^ x7; - y4 = y1 ^ x3; - y12 = y13 ^ y14; - y2 = y1 ^ x0; - y5 = y1 ^ x6; - y3 = y5 ^ y8; - t1 = x4 ^ y12; - y15 = t1 ^ x5; - y20 = t1 ^ x1; - y6 = y15 ^ x7; - y10 = y15 ^ t0; - y11 = y20 ^ y9; - y7 = x7 ^ y11; - y17 = y10 ^ y11; - y19 = y10 ^ y8; - y16 = t0 ^ y11; - y21 = y13 ^ y16; - y18 = x0 ^ y16; - t2 = y12 & y15; - t3 = y3 & y6; - t4 = t3 ^ t2; - t5 = y4 & x7; - t6 = t5 ^ t2; - t7 = y13 & y16; - t8 = y5 & y1; - t9 = t8 ^ t7; - t10 = y2 & y7; - t11 = t10 ^ t7; - t12 = y9 & y11; - t13 = y14 & y17; - t14 = t13 ^ t12; - t15 = y8 & y10; - t16 = t15 ^ t12; - t17 = t4 ^ t14; - t18 = t6 ^ t16; - t19 = t9 ^ t14; - t20 = t11 ^ t16; - t21 = t17 ^ y20; - t22 = t18 ^ y19; - t23 = t19 ^ y21; - t24 = t20 ^ y18; - t25 = t21 ^ t22; - t26 = t21 & t23; - t27 = t24 ^ t26; - t28 = t25 & t27; - t29 = t28 ^ t22; - t30 = t23 ^ t24; - t31 = t22 ^ t26; - t32 = t31 & t30; - t33 = t32 ^ t24; - t34 = t23 ^ t33; - t35 = t27 ^ t33; - t36 = t24 & t35; - t37 = t36 ^ t34; - t38 = t27 ^ t36; - t39 = t29 & t38; - t40 = t25 ^ t39; - t41 = t40 ^ t37; - t42 = t29 ^ t33; - t43 = t29 ^ t40; - t44 = t33 ^ t37; - t45 = t42 ^ t41; - z0 = t44 & y15; - z1 = t37 & y6; - z2 = t33 & x7; - z3 = t43 & y16; - z4 = t40 & y1; - z5 = t29 & y7; - z6 = t42 & y11; - z7 = t45 & y17; - z8 = t41 & y10; - z9 = t44 & y12; - z10 = t37 & y3; - z11 = t33 & y4; - z12 = t43 & y13; - z13 = t40 & y5; - z14 = t29 & y2; - z15 = t42 & y9; - z16 = t45 & y14; - z17 = t41 & y8; - t46 = z15 ^ z16; - t47 = z10 ^ z11; - t48 = z5 ^ z13; - t49 = z9 ^ z10; - t50 = z2 ^ z12; - t51 = z2 ^ z5; - t52 = z7 ^ z8; - t53 = z0 ^ z3; - t54 = z6 ^ z7; - t55 = z16 ^ z17; - t56 = z12 ^ t48; - t57 = t50 ^ t53; - t58 = z4 ^ t46; - t59 = z3 ^ t54; - t60 = t46 ^ t57; - t61 = z14 ^ t57; - t62 = t52 ^ t58; - t63 = t49 ^ t58; - t64 = z4 ^ t59; - t65 = t61 ^ t62; - t66 = z1 ^ t63; - s0 = t59 ^ t63; - s6 = ~(t56 ^ t62); - s7 = ~(t48 ^ t60); - t67 = t64 ^ t65; - s3 = t53 ^ t66; - s4 = t51 ^ t66; - s5 = t47 ^ t65; - s1 = ~(t64 ^ s3); - s2 = ~(t55 ^ t67); - state[0] = s7; - state[1] = s6; - state[2] = s5; - state[3] = s4; - state[4] = s3; - state[5] = s2; - state[6] = s1; - state[7] = s0; -} - -static void BitsShiftRows(u64 state[8]) -{ - u64 s, s0; - int i; - - for (i = 0; i < 8; i++) { - s = state[i]; - s0 = s & 0x1111111111111111uLL; - s0 |= ((s & 0x2220222022202220uLL) >> 4) | ((s & 0x0002000200020002uLL) << 12); - s0 |= ((s & 0x4400440044004400uLL) >> 8) | ((s & 0x0044004400440044uLL) << 8); - s0 |= ((s & 0x8000800080008000uLL) >> 12) | ((s & 0x0888088808880888uLL) << 4); - state[i] = s0; - } -} - -static void BitsMixColumns(u64 state[8]) -{ - u64 s1, s; - u64 s0[8]; - int i; - - for (i = 0; i < 8; i++) { - s1 = state[i]; - s = s1; - s ^= ((s & 0xCCCCCCCCCCCCCCCCuLL) >> 2) | ((s & 0x3333333333333333uLL) << 2); - s ^= ((s & 0xAAAAAAAAAAAAAAAAuLL) >> 1) | ((s & 0x5555555555555555uLL) << 1); - s ^= s1; - s0[i] = s; - } - BitsXtime(state); - for (i = 0; i < 8; i++) { - s1 = state[i]; - s = s0[i]; - s ^= s1; - s ^= ((s1 & 0xEEEEEEEEEEEEEEEEuLL) >> 1) | ((s1 & 0x1111111111111111uLL) << 3); - state[i] = s; - } -} - -static void BitsAddRoundKey(u64 state[8], const u64 key[8]) -{ - int i; - - for (i = 0; i < 8; i++) - state[i] ^= key[i]; -} - -void AES_ctr32_encrypt(const unsigned char *in, unsigned char *out, - size_t blocks, const AES_KEY *key, - const unsigned char *ivec) -{ - struct { - u8 cipher[64]; - u64 state[8]; - u64 rd_key[AES_MAXNR + 1][8]; - } *bs; - u32 ctr32; - int i; - - ctr32 = GETU32(ivec + 12); - if (blocks >= 4 - && (bs = OPENSSL_malloc(sizeof(*bs)))) { - for (i = 0; i < key->rounds + 1; i++) { - memcpy(bs->cipher + 0, &key->rd_key[4 * i], 16); - memcpy(bs->cipher + 16, bs->cipher, 16); - memcpy(bs->cipher + 32, bs->cipher, 32); - RawToBits(bs->cipher, bs->rd_key[i]); - } - while (blocks) { - memcpy(bs->cipher, ivec, 12); - PUTU32(bs->cipher + 12, ctr32); - ctr32++; - memcpy(bs->cipher + 16, ivec, 12); - PUTU32(bs->cipher + 28, ctr32); - ctr32++; - memcpy(bs->cipher + 32, ivec, 12); - PUTU32(bs->cipher + 44, ctr32); - ctr32++; - memcpy(bs->cipher + 48, ivec, 12); - PUTU32(bs->cipher + 60, ctr32); - ctr32++; - RawToBits(bs->cipher, bs->state); - BitsAddRoundKey(bs->state, bs->rd_key[0]); - for (i = 1; i < key->rounds; i++) { - BitsSub(bs->state); - BitsShiftRows(bs->state); - BitsMixColumns(bs->state); - BitsAddRoundKey(bs->state, bs->rd_key[i]); - } - BitsSub(bs->state); - BitsShiftRows(bs->state); - BitsAddRoundKey(bs->state, bs->rd_key[key->rounds]); - BitsToRaw(bs->state, bs->cipher); - for (i = 0; i < 64 && blocks; i++) { - out[i] = in[i] ^ bs->cipher[i]; - if ((i & 15) == 15) - blocks--; - } - in += i; - out += i; - } - OPENSSL_clear_free(bs, sizeof(*bs)); - } else { - unsigned char cipher[16]; - - while (blocks) { - memcpy(cipher, ivec, 12); - PUTU32(cipher + 12, ctr32); - AES_encrypt(cipher, cipher, key); - for (i = 0; i < 16; i++) - out[i] = in[i] ^ cipher[i]; - in += 16; - out += 16; - ctr32++; - blocks--; - } - } -} -# endif #elif !defined(AES_ASM) /*- Te0[x] = S [x].[02, 01, 01, 03]; diff --git a/crypto/evp/e_aes.c b/crypto/evp/e_aes.c index a1b7d50bbf..405ddbf9bf 100644 --- a/crypto/evp/e_aes.c +++ b/crypto/evp/e_aes.c @@ -130,11 +130,6 @@ void bsaes_xts_decrypt(const unsigned char *inp, unsigned char *out, size_t len, const AES_KEY *key1, const AES_KEY *key2, const unsigned char iv[16]); #endif -#if !defined(AES_ASM) && !defined(AES_CTR_ASM) \ - && defined(OPENSSL_AES_CONST_TIME) \ - && !defined(OPENSSL_SMALL_FOOTPRINT) -# define AES_CTR_ASM -#endif #ifdef AES_CTR_ASM void AES_ctr32_encrypt(const unsigned char *in, unsigned char *out, size_t blocks, const AES_KEY *key,