mirror of
https://github.com/henrydcase/pqc.git
synced 2024-11-30 03:11:43 +00:00
147 lines
5.0 KiB
C
147 lines
5.0 KiB
C
|
/*
|
||
|
This file is for functions related to 256-bit vectors
|
||
|
including functions for bitsliced field operations
|
||
|
*/
|
||
|
|
||
|
#include "vec256.h"
|
||
|
|
||
|
extern void PQCLEAN_MCELIECE6688128F_AVX_vec256_mul_asm(vec256 *, vec256 *, const vec256 *);
|
||
|
|
||
|
vec256 PQCLEAN_MCELIECE6688128F_AVX_vec256_set1_16b(uint16_t a) {
|
||
|
return _mm256_set1_epi16(a);
|
||
|
}
|
||
|
|
||
|
vec256 PQCLEAN_MCELIECE6688128F_AVX_vec256_setzero(void) {
|
||
|
return _mm256_setzero_si256();
|
||
|
}
|
||
|
|
||
|
vec256 PQCLEAN_MCELIECE6688128F_AVX_vec256_set4x(uint64_t a0, uint64_t a1, uint64_t a2, uint64_t a3) {
|
||
|
return _mm256_set_epi64x(a3, a2, a1, a0);
|
||
|
}
|
||
|
|
||
|
int PQCLEAN_MCELIECE6688128F_AVX_vec256_testz(vec256 a) {
|
||
|
return _mm256_testz_si256(a, a);
|
||
|
}
|
||
|
|
||
|
vec256 PQCLEAN_MCELIECE6688128F_AVX_vec256_and(vec256 a, vec256 b) {
|
||
|
return _mm256_and_si256(a, b);
|
||
|
}
|
||
|
|
||
|
vec256 PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(vec256 a, vec256 b) {
|
||
|
return _mm256_xor_si256(a, b);
|
||
|
}
|
||
|
|
||
|
vec256 PQCLEAN_MCELIECE6688128F_AVX_vec256_or(vec256 a, vec256 b) {
|
||
|
return _mm256_or_si256(a, b);
|
||
|
}
|
||
|
|
||
|
vec256 PQCLEAN_MCELIECE6688128F_AVX_vec256_sll_4x(vec256 a, int s) {
|
||
|
return _mm256_slli_epi64(a, s);
|
||
|
}
|
||
|
|
||
|
vec256 PQCLEAN_MCELIECE6688128F_AVX_vec256_srl_4x(vec256 a, int s) {
|
||
|
return _mm256_srli_epi64(a, s);
|
||
|
}
|
||
|
|
||
|
vec256 PQCLEAN_MCELIECE6688128F_AVX_vec256_unpack_low(vec256 a, vec256 b) {
|
||
|
return _mm256_permute2x128_si256 (a, b, 0x20);
|
||
|
}
|
||
|
|
||
|
vec256 PQCLEAN_MCELIECE6688128F_AVX_vec256_unpack_high(vec256 a, vec256 b) {
|
||
|
return _mm256_permute2x128_si256 (a, b, 0x31);
|
||
|
}
|
||
|
|
||
|
vec256 PQCLEAN_MCELIECE6688128F_AVX_vec256_unpack_low_2x(vec256 a, vec256 b) {
|
||
|
return _mm256_unpacklo_epi64 (a, b);
|
||
|
}
|
||
|
|
||
|
vec256 PQCLEAN_MCELIECE6688128F_AVX_vec256_unpack_high_2x(vec256 a, vec256 b) {
|
||
|
return _mm256_unpackhi_epi64 (a, b);
|
||
|
}
|
||
|
|
||
|
vec256 PQCLEAN_MCELIECE6688128F_AVX_vec256_or_reduce(const vec256 *a) {
|
||
|
int i;
|
||
|
vec256 ret;
|
||
|
|
||
|
ret = a[0];
|
||
|
for (i = 1; i < GFBITS; i++) {
|
||
|
ret = PQCLEAN_MCELIECE6688128F_AVX_vec256_or(ret, a[i]);
|
||
|
}
|
||
|
|
||
|
return ret;
|
||
|
}
|
||
|
|
||
|
void PQCLEAN_MCELIECE6688128F_AVX_vec256_copy(vec256 *dest, const vec256 *src) {
|
||
|
int i;
|
||
|
|
||
|
for (i = 0; i < GFBITS; i++) {
|
||
|
dest[i] = src[i];
|
||
|
}
|
||
|
}
|
||
|
|
||
|
|
||
|
void PQCLEAN_MCELIECE6688128F_AVX_vec256_mul(vec256 *h, vec256 *f, const vec256 *g) {
|
||
|
PQCLEAN_MCELIECE6688128F_AVX_vec256_mul_asm(h, f, g);
|
||
|
}
|
||
|
|
||
|
/* bitsliced field squarings */
|
||
|
void PQCLEAN_MCELIECE6688128F_AVX_vec256_sq(vec256 *out, const vec256 *in) {
|
||
|
int i;
|
||
|
vec256 result[GFBITS], t;
|
||
|
|
||
|
t = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(in[11], in[12]);
|
||
|
|
||
|
result[0] = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(in[0], in[11]);
|
||
|
result[1] = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(in[7], t);
|
||
|
result[2] = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(in[1], in[7]);
|
||
|
result[3] = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(in[8], t);
|
||
|
result[4] = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(in[2], in[7]);
|
||
|
result[4] = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(result[4], in[8]);
|
||
|
result[4] = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(result[4], t);
|
||
|
result[5] = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(in[7], in[9]);
|
||
|
result[6] = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(in[3], in[8]);
|
||
|
result[6] = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(result[6], in[9]);
|
||
|
result[6] = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(result[6], in[12]);
|
||
|
result[7] = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(in[8], in[10]);
|
||
|
result[8] = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(in[4], in[9]);
|
||
|
result[8] = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(result[8], in[10]);
|
||
|
result[9] = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(in[9], in[11]);
|
||
|
result[10] = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(in[5], in[10]);
|
||
|
result[10] = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(result[10], in[11]);
|
||
|
result[11] = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(in[10], in[12]);
|
||
|
result[12] = PQCLEAN_MCELIECE6688128F_AVX_vec256_xor(in[6], t);
|
||
|
|
||
|
for (i = 0; i < GFBITS; i++) {
|
||
|
out[i] = result[i];
|
||
|
}
|
||
|
}
|
||
|
|
||
|
/* bitsliced field inverses */
|
||
|
void PQCLEAN_MCELIECE6688128F_AVX_vec256_inv(vec256 *out, const vec256 *in) {
|
||
|
vec256 tmp_11[ GFBITS ];
|
||
|
vec256 tmp_1111[ GFBITS ];
|
||
|
|
||
|
PQCLEAN_MCELIECE6688128F_AVX_vec256_copy(out, in);
|
||
|
|
||
|
PQCLEAN_MCELIECE6688128F_AVX_vec256_sq(out, out);
|
||
|
PQCLEAN_MCELIECE6688128F_AVX_vec256_mul(tmp_11, out, in); // ^11
|
||
|
|
||
|
PQCLEAN_MCELIECE6688128F_AVX_vec256_sq(out, tmp_11);
|
||
|
PQCLEAN_MCELIECE6688128F_AVX_vec256_sq(out, out);
|
||
|
PQCLEAN_MCELIECE6688128F_AVX_vec256_mul(tmp_1111, out, tmp_11); // ^1111
|
||
|
|
||
|
PQCLEAN_MCELIECE6688128F_AVX_vec256_sq(out, tmp_1111);
|
||
|
PQCLEAN_MCELIECE6688128F_AVX_vec256_sq(out, out);
|
||
|
PQCLEAN_MCELIECE6688128F_AVX_vec256_sq(out, out);
|
||
|
PQCLEAN_MCELIECE6688128F_AVX_vec256_sq(out, out);
|
||
|
PQCLEAN_MCELIECE6688128F_AVX_vec256_mul(out, out, tmp_1111); // ^11111111
|
||
|
|
||
|
PQCLEAN_MCELIECE6688128F_AVX_vec256_sq(out, out);
|
||
|
PQCLEAN_MCELIECE6688128F_AVX_vec256_sq(out, out);
|
||
|
PQCLEAN_MCELIECE6688128F_AVX_vec256_sq(out, out);
|
||
|
PQCLEAN_MCELIECE6688128F_AVX_vec256_sq(out, out);
|
||
|
PQCLEAN_MCELIECE6688128F_AVX_vec256_mul(out, out, tmp_1111); // ^111111111111
|
||
|
|
||
|
PQCLEAN_MCELIECE6688128F_AVX_vec256_sq(out, out); // ^1111111111110
|
||
|
}
|