/************************************************************************* * Copyright (C) 2021 Intel Corporation * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. *************************************************************************/ #include "owndefs.h" #if (_IPP32E >= _IPP32E_K1) #include "gfpec/sm2/ifma_arith_nsm2.h" #include "gfpec/sm2/ifma_arith_psm2.h" #include "gfpec/sm2/ifma_defs_sm2.h" /* * EC SM2 base point order * in 2^52 radix */ static const __ALIGN64 Ipp64u nsm2_x1[PSM2_LEN52] = { 0x000bf40939d54123, 0x0006b21c6052b53b, 0x000fffffff7203df, 0x000fffffffffffff, 0x0000fffffffeffff }; /* k0 = -( (1/nsm2) mod 2^DIGIT_SIZE_52 ) mod 2^DIGIT_SIZE_52 */ static const Ipp64u nsm2_k0 = 0x000f9e8872350975; /* r = 2^((52*8)) mod nsm2 */ static const __ALIGN64 Ipp64u nsm2_r[PSM2_LEN52] = { 0x00062abedd000000, 0x00006cb726ecad3c, 0x00056c361b698a7e, 0x0000000008dfc209, 0x0000010000000000 }; /* To Montgomery conversion constant * rr = 2^((52*6)*2) mod nsm2 */ static const __ALIGN64 Ipp64u nsm2_rr[PSM2_LEN52] = { 0x000643f5455b3830, 0x000c453935b521eb, 0x0009ec21d42b082e, 0x0004d74634238016, 0x0000ca6ea35cbdde }; static const __ALIGN64 Ipp64u ones[PSM2_LEN52] = { 0x1, 0x0, 0x0, 0x0 }; #define MULT_ROUND(R, A, B, IDX) \ const fesm2 Bi##R##IDX = permutexvar_i8(idx##IDX, (B)); \ const fesm2 amBiLo##R##IDX = madd52lo_i64(zero, (A), Bi##R##IDX); \ fesm2 tr##R##IDX = madd52hi_i64(zero, (A), Bi##R##IDX); \ { \ /* low */ \ (R) = add_i64((R), amBiLo##R##IDX); \ const fesm2 R0 = permutexvar_i8(idx0, (R)); \ const fesm2 u = madd52lo_i64(zero, R0, K0); /* u = R0 * K0 */ \ tr##R##IDX = madd52hi_i64(tr##R##IDX, N, u); \ (R) = madd52lo_i64((R), N, u); \ /* shift */ \ const fesm2 carryone = maskz_srai_i64(maskone, (R), DIGIT_SIZE_52); \ tr##R##IDX = add_i64(tr##R##IDX, carryone); \ (R) = maskz_permutexvar_i8(mask_sr64, idx_sr64, (R)); \ /* hi */ \ (R) = add_i64((R), tr##R##IDX); \ } /* R = (A*B) - no normalization (in radix 2^52) */ IPP_OWN_DEFN(fesm2, fesm2_mul_norder, (const fesm2 a, const fesm2 b)) { const fesm2 N = FESM2_LOADU(nsm2_x1); /* p */ const fesm2 K0 = set1_i64(nsm2_k0); const fesm2 zero = setzero_i64(); const mask8 maskone = 0x1; /* index broadcast */ const fesm2 idx0 = set_i64(REPL8(0x0706050403020100)); // 7, 6, 5, 4, 3, 2, 1, 0 const fesm2 idx1 = set_i64(REPL8(0x0f0e0d0c0b0a0908)); // 15, 14, 13, 12, 11, 10, 9, 8 const fesm2 idx2 = set_i64(REPL8(0x1716151413121110)); // 23, 22, 21, 20, 19, 18, 17, 16 const fesm2 idx3 = set_i64(REPL8(0x1f1e1d1c1b1a1918)); // 31, 30, 29, 28, 27, 26, 25, 24 const fesm2 idx4 = set_i64(REPL8(0x2726252423222120)); // 39, 38, 37, 36, 35, 34, 33, 32 const fesm2 idx5 = set_i64(REPL8(0x2f2e2d2c2b2a2928)); // 47, 46, 45, 44, 43, 42, 41, 40 /* shift right 64 bit line [R >> 64] */ const mask64 mask_sr64 = 0x00FFFFFFFFFFFFFF; const fesm2 idx_sr64 = set_i64(0x0, // 0, 0, 0, 0, 0, 0, 0, 0 0x3f3e3d3c3b3a3938, // 63, 62, 61, 60, 59, 58, 57, 56 0x3736353433323130, // 55, 54, 53, 52, 51, 50, 49, 48 0x2f2e2d2c2b2a2928, // 47, 46, 45, 44, 43, 42, 41, 40 0x2726252423222120, // 39, 38, 37, 36, 35, 34, 33, 32 0x1f1e1d1c1b1a1918, // 31, 30, 29, 28, 27, 26, 25, 24 0x1716151413121110, // 23, 22, 21, 20, 19, 18, 17, 16 0x0f0e0d0c0b0a0908); // 15, 14, 13, 12, 11, 10, 9, 8 fesm2 r = setzero_i64(); /* PSM2 * m' mod b = 1 * * Algorithm * a[] b[] - input data ((in radix 2^52)) m[] - module nsm2 * * when u = R[0]*m' mod b * 1) R = R + a[] * b[i] (lo) * 2) R = R + m[] * u (lo) * 3) R = R >> 64 * 4) R = R + a[] * b[i] (hi) * 5) R = R + m[] * u (hi) */ /* one round = O(32) */ MULT_ROUND(r, a, b, 0) MULT_ROUND(r, a, b, 1) MULT_ROUND(r, a, b, 2) MULT_ROUND(r, a, b, 3) MULT_ROUND(r, a, b, 4) MULT_ROUND(r, a, b, 5) return r; } IPP_OWN_DEFN(fesm2, fesm2_add_norder_norm, (const fesm2 a, const fesm2 b)) { const fesm2 zero = setzero_i64(); const fesm2 N = FESM2_LOADU(nsm2_x1); /* r = a + b */ fesm2 r = add_i64(a, b); r = ifma_lnorm52(r); /* t = r - N */ fesm2 t = sub_i64(r, N); t = ifma_norm52(t); /* lt = t < 0 */ const mask8 lt = cmp_i64_mask(zero, srli_i64(t, DIGIT_SIZE_52 - 1), _MM_CMPINT_LT); const mask8 mask = (mask8)((mask8)0 - ((lt >> 4) & 1)); /* mask != 0 ? a : r */ r = mask_mov_i64(t, mask, r); return r; } IPP_OWN_DEFN(fesm2, fesm2_sub_norder_norm, (const fesm2 a, const fesm2 b)) { const fesm2 zero = setzero_i64(); const fesm2 N = FESM2_LOADU(nsm2_x1); /* r = a - b */ fesm2 r = sub_i64(a, b); r = ifma_norm52(r); /* t = r + M */ fesm2 t = add_i64(r, N); t = ifma_lnorm52(t); /* lt = r < 0 */ const mask8 lt = cmp_i64_mask(zero, srli_i64(r, DIGIT_SIZE_52 - 1), _MM_CMPINT_LT); const mask8 mask = (mask8)((mask8)0 - ((lt >> 4) & 1)); /* mask != 0 ? t : r */ r = mask_mov_i64(r, mask, t); return r; } IPP_OWN_DEFN(fesm2, fesm2_fast_reduction_norder, (const fesm2 a)) { const fesm2 zero = setzero_i64(); const fesm2 N = FESM2_LOADU(nsm2_x1); fesm2 r = sub_i64(a, N); r = ifma_norm52(r); const mask8 lt = cmp_i64_mask(zero, srli_i64(r, DIGIT_SIZE_52 - 1), _MM_CMPINT_LT); const mask8 mask = (mask8)((mask8)0 - ((lt >> 4) & 1)); /* mask != 0 ? a : r */ r = mask_mov_i64(r, mask, a); return r; } IPP_OWN_DEFN(fesm2, fesm2_to_mont_norder, (const fesm2 a)) { const fesm2 RR = FESM2_LOADU(nsm2_rr); const fesm2 r = fesm2_mul_norder(a, RR); return ifma_lnorm52(r); } IPP_OWN_DEFN(fesm2, fesm2_from_mont_norder, (const fesm2 a)) { const fesm2 ONE = FESM2_LOADU(ones); fesm2 r = fesm2_mul_norder(a, ONE); r = ifma_lnorm52(r); r = fesm2_fast_reduction_norder(r); return r; } IPPCP_INLINE fesm2 mul_norder_norm(const fesm2 a, const fesm2 b) { const fesm2 r = fesm2_mul_norder(a, b); return ifma_lnorm52(r); } IPPCP_INLINE fesm2 sqr_norder_norm(const fesm2 a) { const fesm2 r = fesm2_mul_norder(a, a); return ifma_lnorm52(r); } #define SIZE_TBL (16) #define sqr(R, A) (R) = sqr_norder_norm((A)) #define mul(R, A, B) (R) = mul_norder_norm((A), (B)) IPP_OWN_DEFN(fesm2, fesm2_inv_norder_norm, (const fesm2 z)) { const fesm2 r_norder = FESM2_LOADU(nsm2_r); int i; // pwr_z_Tbl[i][] = z^i, i=0,..,15 __ALIGN64 fesm2 pwr_z_Tbl[SIZE_TBL]; pwr_z_Tbl[0] = r_norder; pwr_z_Tbl[1] = z; for (i = 2; i < SIZE_TBL; i += 2) { sqr(pwr_z_Tbl[i], pwr_z_Tbl[i / 2]); mul(pwr_z_Tbl[i + 1], pwr_z_Tbl[i], z); } // pwr = (nsm2-2) in big endian const Ipp8u pwr[] = "\xFF\xFF\xFF\xFE\xFF\xFF\xFF\xFF" "\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF" "\x72\x03\xDF\x6B\x21\xC6\x05\x2B" "\x53\xBB\xF4\x09\x39\xD5\x41\x21"; // init r = 1 fesm2 r = pwr_z_Tbl[0]; for (i = 0; i < 32; ++i) { const int v = pwr[i]; const int hi = (v >> 4) & 0xF; const int lo = v & 0xF; sqr(r, r); sqr(r, r); sqr(r, r); sqr(r, r); if (hi) mul(r, r, pwr_z_Tbl[hi]); sqr(r, r); sqr(r, r); sqr(r, r); sqr(r, r); if (lo) mul(r, r, pwr_z_Tbl[lo]); } return r; } #undef SIZE_TBL #undef sqr #undef mul #endif // (_IPP32E >= _IPP32E_K1)