#include "align.h" #include "cbd.h" #include "consts.h" #include "ntt.h" #include "params.h" #include "poly.h" #include "reduce.h" #include "symmetric.h" #include #include /************************************************* * Name: PQCLEAN_KYBER76890S_AVX2_poly_compress * * Description: Compression and subsequent serialization of a polynomial. * The coefficients of the input polynomial are assumed to * lie in the invertal [0,q], i.e. the polynomial must be reduced * by PQCLEAN_KYBER76890S_AVX2_poly_reduce(). * * Arguments: - uint8_t *r: pointer to output byte array * (of length KYBER_POLYCOMPRESSEDBYTES) * - const poly *a: pointer to input polynomial **************************************************/ void PQCLEAN_KYBER76890S_AVX2_poly_compress(uint8_t r[128], const poly *restrict a) { unsigned int i; __m256i f0, f1, f2, f3; const __m256i v = _mm256_load_si256(&PQCLEAN_KYBER76890S_AVX2_qdata.vec[_16XV / 16]); const __m256i shift1 = _mm256_set1_epi16(1 << 9); const __m256i mask = _mm256_set1_epi16(15); const __m256i shift2 = _mm256_set1_epi16((16 << 8) + 1); const __m256i permdidx = _mm256_set_epi32(7, 3, 6, 2, 5, 1, 4, 0); for (i = 0; i < KYBER_N / 64; i++) { f0 = _mm256_load_si256(&a->vec[4 * i + 0]); f1 = _mm256_load_si256(&a->vec[4 * i + 1]); f2 = _mm256_load_si256(&a->vec[4 * i + 2]); f3 = _mm256_load_si256(&a->vec[4 * i + 3]); f0 = _mm256_mulhi_epi16(f0, v); f1 = _mm256_mulhi_epi16(f1, v); f2 = _mm256_mulhi_epi16(f2, v); f3 = _mm256_mulhi_epi16(f3, v); f0 = _mm256_mulhrs_epi16(f0, shift1); f1 = _mm256_mulhrs_epi16(f1, shift1); f2 = _mm256_mulhrs_epi16(f2, shift1); f3 = _mm256_mulhrs_epi16(f3, shift1); f0 = _mm256_and_si256(f0, mask); f1 = _mm256_and_si256(f1, mask); f2 = _mm256_and_si256(f2, mask); f3 = _mm256_and_si256(f3, mask); f0 = _mm256_packus_epi16(f0, f1); f2 = _mm256_packus_epi16(f2, f3); f0 = _mm256_maddubs_epi16(f0, shift2); f2 = _mm256_maddubs_epi16(f2, shift2); f0 = _mm256_packus_epi16(f0, f2); f0 = _mm256_permutevar8x32_epi32(f0, permdidx); _mm256_storeu_si256((__m256i *)&r[32 * i], f0); } } void PQCLEAN_KYBER76890S_AVX2_poly_decompress(poly *restrict r, const uint8_t a[128]) { unsigned int i; __m128i t; __m256i f; const __m256i q = _mm256_load_si256(&PQCLEAN_KYBER76890S_AVX2_qdata.vec[_16XQ / 16]); const __m256i shufbidx = _mm256_set_epi8(7, 7, 7, 7, 6, 6, 6, 6, 5, 5, 5, 5, 4, 4, 4, 4, 3, 3, 3, 3, 2, 2, 2, 2, 1, 1, 1, 1, 0, 0, 0, 0); const __m256i mask = _mm256_set1_epi32(0x00F0000F); const __m256i shift = _mm256_set1_epi32((128 << 16) + 2048); for (i = 0; i < KYBER_N / 16; i++) { t = _mm_loadl_epi64((__m128i *)&a[8 * i]); f = _mm256_broadcastsi128_si256(t); f = _mm256_shuffle_epi8(f, shufbidx); f = _mm256_and_si256(f, mask); f = _mm256_mullo_epi16(f, shift); f = _mm256_mulhrs_epi16(f, q); _mm256_store_si256(&r->vec[i], f); } } /************************************************* * Name: PQCLEAN_KYBER76890S_AVX2_poly_tobytes * * Description: Serialization of a polynomial in NTT representation. * The coefficients of the input polynomial are assumed to * lie in the invertal [0,q], i.e. the polynomial must be reduced * by PQCLEAN_KYBER76890S_AVX2_poly_reduce(). The coefficients are orderd as output by * PQCLEAN_KYBER76890S_AVX2_poly_ntt(); the serialized output coefficients are in bitreversed * order. * * Arguments: - uint8_t *r: pointer to output byte array * (needs space for KYBER_POLYBYTES bytes) * - poly *a: pointer to input polynomial **************************************************/ void PQCLEAN_KYBER76890S_AVX2_poly_tobytes(uint8_t r[KYBER_POLYBYTES], poly *a) { PQCLEAN_KYBER76890S_AVX2_ntttobytes_avx(r, a->vec, PQCLEAN_KYBER76890S_AVX2_qdata.vec); } /************************************************* * Name: PQCLEAN_KYBER76890S_AVX2_poly_frombytes * * Description: De-serialization of a polynomial; * inverse of PQCLEAN_KYBER76890S_AVX2_poly_tobytes * * Arguments: - poly *r: pointer to output polynomial * - const uint8_t *a: pointer to input byte array * (of KYBER_POLYBYTES bytes) **************************************************/ void PQCLEAN_KYBER76890S_AVX2_poly_frombytes(poly *r, const uint8_t a[KYBER_POLYBYTES]) { PQCLEAN_KYBER76890S_AVX2_nttfrombytes_avx(r->vec, a, PQCLEAN_KYBER76890S_AVX2_qdata.vec); } /************************************************* * Name: PQCLEAN_KYBER76890S_AVX2_poly_frommsg * * Description: Convert 32-byte message to polynomial * * Arguments: - poly *r: pointer to output polynomial * - const uint8_t *msg: pointer to input message **************************************************/ void PQCLEAN_KYBER76890S_AVX2_poly_frommsg(poly *restrict r, const uint8_t msg[KYBER_INDCPA_MSGBYTES]) { __m256i f, g0, g1, g2, g3, h0, h1, h2, h3; const __m256i shift = _mm256_broadcastsi128_si256(_mm_set_epi32(0, 1, 2, 3)); const __m256i idx = _mm256_broadcastsi128_si256(_mm_set_epi8(15, 14, 11, 10, 7, 6, 3, 2, 13, 12, 9, 8, 5, 4, 1, 0)); const __m256i hqs = _mm256_set1_epi16((KYBER_Q + 1) / 2); #define FROMMSG64(i) \ g3 = _mm256_shuffle_epi32(f,0x55*(i)); \ g3 = _mm256_sllv_epi32(g3,shift); \ g3 = _mm256_shuffle_epi8(g3,idx); \ g0 = _mm256_slli_epi16(g3,12); \ g1 = _mm256_slli_epi16(g3,8); \ g2 = _mm256_slli_epi16(g3,4); \ g0 = _mm256_srai_epi16(g0,15); \ g1 = _mm256_srai_epi16(g1,15); \ g2 = _mm256_srai_epi16(g2,15); \ g3 = _mm256_srai_epi16(g3,15); \ g0 = _mm256_and_si256(g0,hqs); /* 19 18 17 16 3 2 1 0 */ \ g1 = _mm256_and_si256(g1,hqs); /* 23 22 21 20 7 6 5 4 */ \ g2 = _mm256_and_si256(g2,hqs); /* 27 26 25 24 11 10 9 8 */ \ g3 = _mm256_and_si256(g3,hqs); /* 31 30 29 28 15 14 13 12 */ \ h0 = _mm256_unpacklo_epi64(g0,g1); \ h2 = _mm256_unpackhi_epi64(g0,g1); \ h1 = _mm256_unpacklo_epi64(g2,g3); \ h3 = _mm256_unpackhi_epi64(g2,g3); \ g0 = _mm256_permute2x128_si256(h0,h1,0x20); \ g2 = _mm256_permute2x128_si256(h0,h1,0x31); \ g1 = _mm256_permute2x128_si256(h2,h3,0x20); \ g3 = _mm256_permute2x128_si256(h2,h3,0x31); \ _mm256_store_si256(&r->vec[0+2*(i)+0],g0); \ _mm256_store_si256(&r->vec[0+2*(i)+1],g1); \ _mm256_store_si256(&r->vec[8+2*(i)+0],g2); \ _mm256_store_si256(&r->vec[8+2*(i)+1],g3) f = _mm256_loadu_si256((__m256i *)msg); FROMMSG64(0); FROMMSG64(1); FROMMSG64(2); FROMMSG64(3); } /************************************************* * Name: PQCLEAN_KYBER76890S_AVX2_poly_tomsg * * Description: Convert polynomial to 32-byte message. * The coefficients of the input polynomial are assumed to * lie in the invertal [0,q], i.e. the polynomial must be reduced * by PQCLEAN_KYBER76890S_AVX2_poly_reduce(). * * Arguments: - uint8_t *msg: pointer to output message * - poly *a: pointer to input polynomial **************************************************/ void PQCLEAN_KYBER76890S_AVX2_poly_tomsg(uint8_t msg[KYBER_INDCPA_MSGBYTES], poly *restrict a) { unsigned int i; uint32_t small; __m256i f0, f1, g0, g1; const __m256i hq = _mm256_set1_epi16((KYBER_Q - 1) / 2); const __m256i hhq = _mm256_set1_epi16((KYBER_Q - 1) / 4); for (i = 0; i < KYBER_N / 32; i++) { f0 = _mm256_load_si256(&a->vec[2 * i + 0]); f1 = _mm256_load_si256(&a->vec[2 * i + 1]); f0 = _mm256_sub_epi16(hq, f0); f1 = _mm256_sub_epi16(hq, f1); g0 = _mm256_srai_epi16(f0, 15); g1 = _mm256_srai_epi16(f1, 15); f0 = _mm256_xor_si256(f0, g0); f1 = _mm256_xor_si256(f1, g1); f0 = _mm256_sub_epi16(f0, hhq); f1 = _mm256_sub_epi16(f1, hhq); f0 = _mm256_packs_epi16(f0, f1); small = _mm256_movemask_epi8(f0); msg[4 * i + 0] = small; msg[4 * i + 1] = small >> 16; msg[4 * i + 2] = small >> 8; msg[4 * i + 3] = small >> 24; } } /************************************************* * Name: PQCLEAN_KYBER76890S_AVX2_poly_getnoise_eta1 * * Description: Sample a polynomial deterministically from a seed and a nonce, * with output polynomial close to centered binomial distribution * with parameter KYBER_ETA1 * * Arguments: - poly *r: pointer to output polynomial * - const uint8_t *seed: pointer to input seed * (of length KYBER_SYMBYTES bytes) * - uint8_t nonce: one-byte input nonce **************************************************/ void PQCLEAN_KYBER76890S_AVX2_poly_getnoise_eta1(poly *r, const uint8_t seed[KYBER_SYMBYTES], uint8_t nonce) { ALIGNED_UINT8(KYBER_ETA1 * KYBER_N / 4 + 32) buf; // +32 bytes as required by PQCLEAN_KYBER76890S_AVX2_poly_cbd_eta1 prf(buf.coeffs, KYBER_ETA1 * KYBER_N / 4, seed, nonce); PQCLEAN_KYBER76890S_AVX2_poly_cbd_eta1(r, buf.vec); } /************************************************* * Name: PQCLEAN_KYBER76890S_AVX2_poly_getnoise_eta2 * * Description: Sample a polynomial deterministically from a seed and a nonce, * with output polynomial close to centered binomial distribution * with parameter KYBER_ETA2 * * Arguments: - poly *r: pointer to output polynomial * - const uint8_t *seed: pointer to input seed * (of length KYBER_SYMBYTES bytes) * - uint8_t nonce: one-byte input nonce **************************************************/ void PQCLEAN_KYBER76890S_AVX2_poly_getnoise_eta2(poly *r, const uint8_t seed[KYBER_SYMBYTES], uint8_t nonce) { ALIGNED_UINT8(KYBER_ETA2 * KYBER_N / 4) buf; prf(buf.coeffs, KYBER_ETA2 * KYBER_N / 4, seed, nonce); PQCLEAN_KYBER76890S_AVX2_poly_cbd_eta2(r, buf.vec); } /************************************************* * Name: PQCLEAN_KYBER76890S_AVX2_poly_ntt * * Description: Computes negacyclic number-theoretic transform (NTT) of * a polynomial in place. * Input coefficients assumed to be in normal order, * output coefficients are in special order that is natural * for the vectorization. Input coefficients are assumed to be * bounded by q in absolute value, output coefficients are bounded * by 16118 in absolute value. * * Arguments: - poly *r: pointer to in/output polynomial **************************************************/ void PQCLEAN_KYBER76890S_AVX2_poly_ntt(poly *r) { PQCLEAN_KYBER76890S_AVX2_ntt_avx(r->vec, PQCLEAN_KYBER76890S_AVX2_qdata.vec); } /************************************************* * Name: PQCLEAN_KYBER76890S_AVX2_poly_invntt_tomont * * Description: Computes inverse of negacyclic number-theoretic transform (NTT) * of a polynomial in place; * Input coefficients assumed to be in special order from vectorized * forward ntt, output in normal order. Input coefficients can be * arbitrary 16-bit integers, output coefficients are bounded by 14870 * in absolute value. * * Arguments: - poly *a: pointer to in/output polynomial **************************************************/ void PQCLEAN_KYBER76890S_AVX2_poly_invntt_tomont(poly *r) { PQCLEAN_KYBER76890S_AVX2_invntt_avx(r->vec, PQCLEAN_KYBER76890S_AVX2_qdata.vec); } void PQCLEAN_KYBER76890S_AVX2_poly_nttunpack(poly *r) { PQCLEAN_KYBER76890S_AVX2_nttunpack_avx(r->vec, PQCLEAN_KYBER76890S_AVX2_qdata.vec); } /************************************************* * Name: PQCLEAN_KYBER76890S_AVX2_poly_basemul_montgomery * * Description: Multiplication of two polynomials in NTT domain. * One of the input polynomials needs to have coefficients * bounded by q, the other polynomial can have arbitrary * coefficients. Output coefficients are bounded by 6656. * * Arguments: - poly *r: pointer to output polynomial * - const poly *a: pointer to first input polynomial * - const poly *b: pointer to second input polynomial **************************************************/ void PQCLEAN_KYBER76890S_AVX2_poly_basemul_montgomery(poly *r, const poly *a, const poly *b) { PQCLEAN_KYBER76890S_AVX2_basemul_avx(r->vec, a->vec, b->vec, PQCLEAN_KYBER76890S_AVX2_qdata.vec); } /************************************************* * Name: PQCLEAN_KYBER76890S_AVX2_poly_tomont * * Description: Inplace conversion of all coefficients of a polynomial * from normal domain to Montgomery domain * * Arguments: - poly *r: pointer to input/output polynomial **************************************************/ void PQCLEAN_KYBER76890S_AVX2_poly_tomont(poly *r) { PQCLEAN_KYBER76890S_AVX2_tomont_avx(r->vec, PQCLEAN_KYBER76890S_AVX2_qdata.vec); } /************************************************* * Name: PQCLEAN_KYBER76890S_AVX2_poly_reduce * * Description: Applies Barrett reduction to all coefficients of a polynomial * for details of the Barrett reduction see comments in reduce.c * * Arguments: - poly *r: pointer to input/output polynomial **************************************************/ void PQCLEAN_KYBER76890S_AVX2_poly_reduce(poly *r) { PQCLEAN_KYBER76890S_AVX2_reduce_avx(r->vec, PQCLEAN_KYBER76890S_AVX2_qdata.vec); } /************************************************* * Name: PQCLEAN_KYBER76890S_AVX2_poly_add * * Description: Add two polynomials. No modular reduction * is performed. * * Arguments: - poly *r: pointer to output polynomial * - const poly *a: pointer to first input polynomial * - const poly *b: pointer to second input polynomial **************************************************/ void PQCLEAN_KYBER76890S_AVX2_poly_add(poly *r, const poly *a, const poly *b) { unsigned int i; __m256i f0, f1; for (i = 0; i < KYBER_N / 16; i++) { f0 = _mm256_load_si256(&a->vec[i]); f1 = _mm256_load_si256(&b->vec[i]); f0 = _mm256_add_epi16(f0, f1); _mm256_store_si256(&r->vec[i], f0); } } /************************************************* * Name: PQCLEAN_KYBER76890S_AVX2_poly_sub * * Description: Subtract two polynomials. No modular reduction * is performed. * * Arguments: - poly *r: pointer to output polynomial * - const poly *a: pointer to first input polynomial * - const poly *b: pointer to second input polynomial **************************************************/ void PQCLEAN_KYBER76890S_AVX2_poly_sub(poly *r, const poly *a, const poly *b) { unsigned int i; __m256i f0, f1; for (i = 0; i < KYBER_N / 16; i++) { f0 = _mm256_load_si256(&a->vec[i]); f1 = _mm256_load_si256(&b->vec[i]); f0 = _mm256_sub_epi16(f0, f1); _mm256_store_si256(&r->vec[i], f0); } }