pqc/crypto_kem/mceliece460896f/avx/pk_gen.c
Thom Wiggers ac2c20045c Classic McEliece (#259)
* Add McEliece reference implementations

* Add Vec implementations of McEliece

* Add sse implementations

* Add AVX2 implementations

* Get rid of stuff not supported by Mac ABI

* restrict to two cores

* Ditch .data files

* Remove .hidden from all .S files

* speed up duplicate consistency tests by batching

* make cpuinfo more robust

* Hope to stabilize macos cpuinfo without ccache

* Revert "Hope to stabilize macos cpuinfo without ccache"

This reverts commit 6129c3cabe1abbc8b956bc87e902a698e32bf322.

* Just hardcode what's available at travis

* Fixed-size types in api.h

* namespace all header files in mceliece

* Ditch operations.h

* Get rid of static inline functions

* fixup! Ditch operations.h
2021-03-24 21:02:45 +00:00

359 lines
9.8 KiB
C

/*
This file is for public-key generation
*/
#include "pk_gen.h"
#include "benes.h"
#include "controlbits.h"
#include "fft.h"
#include "params.h"
#include "util.h"
#include <stdint.h>
#define min(a, b) (((a) < (b)) ? (a) : (b))
static void de_bitslicing(uint64_t *out, vec256 in[][GFBITS]) {
int i, j, r;
uint64_t u = 0;
for (i = 0; i < (1 << GFBITS); i++) {
out[i] = 0 ;
}
for (i = 0; i < 32; i++) {
for (j = GFBITS - 1; j >= 0; j--) {
u = PQCLEAN_MCELIECE460896F_AVX_vec256_extract(in[i][j], 0);
for (r = 0; r < 64; r++) {
out[i * 256 + 0 * 64 + r] <<= 1;
out[i * 256 + 0 * 64 + r] |= (u >> r) & 1;
}
u = PQCLEAN_MCELIECE460896F_AVX_vec256_extract(in[i][j], 1);
for (r = 0; r < 64; r++) {
out[i * 256 + 1 * 64 + r] <<= 1;
out[i * 256 + 1 * 64 + r] |= (u >> r) & 1;
}
u = PQCLEAN_MCELIECE460896F_AVX_vec256_extract(in[i][j], 2);
for (r = 0; r < 64; r++) {
out[i * 256 + 2 * 64 + r] <<= 1;
out[i * 256 + 2 * 64 + r] |= (u >> r) & 1;
}
u = PQCLEAN_MCELIECE460896F_AVX_vec256_extract(in[i][j], 3);
for (r = 0; r < 64; r++) {
out[i * 256 + 3 * 64 + r] <<= 1;
out[i * 256 + 3 * 64 + r] |= (u >> r) & 1;
}
}
}
}
static void to_bitslicing_2x(vec256 out0[][GFBITS], vec256 out1[][GFBITS], const uint64_t *in) {
int i, j, k, r;
uint64_t u[4] = {0};
for (i = 0; i < 32; i++) {
for (j = GFBITS - 1; j >= 0; j--) {
for (k = 0; k < 4; k++) {
for (r = 63; r >= 0; r--) {
u[k] <<= 1;
u[k] |= (in[i * 256 + k * 64 + r] >> (j + GFBITS)) & 1;
}
}
out1[i][j] = PQCLEAN_MCELIECE460896F_AVX_vec256_set4x(u[0], u[1], u[2], u[3]);
}
for (j = GFBITS - 1; j >= 0; j--) {
for (k = 0; k < 4; k++) {
for (r = 63; r >= 0; r--) {
u[k] <<= 1;
u[k] |= (in[i * 256 + k * 64 + r] >> j) & 1;
}
}
out0[i][GFBITS - 1 - j] = PQCLEAN_MCELIECE460896F_AVX_vec256_set4x(u[0], u[1], u[2], u[3]);
}
}
}
static void transpose_64x64(uint64_t *out, const uint64_t *in) {
int i, j, s, d;
uint64_t x, y;
uint64_t masks[6][2] = {
{0x5555555555555555, 0xAAAAAAAAAAAAAAAA},
{0x3333333333333333, 0xCCCCCCCCCCCCCCCC},
{0x0F0F0F0F0F0F0F0F, 0xF0F0F0F0F0F0F0F0},
{0x00FF00FF00FF00FF, 0xFF00FF00FF00FF00},
{0x0000FFFF0000FFFF, 0xFFFF0000FFFF0000},
{0x00000000FFFFFFFF, 0xFFFFFFFF00000000}
};
for (i = 0; i < 64; i++) {
out[i] = in[i];
}
for (d = 5; d >= 0; d--) {
s = 1 << d;
for (i = 0; i < 64; i += s * 2) {
for (j = i; j < i + s; j++) {
x = (out[j] & masks[d][0]) | ((out[j + s] & masks[d][0]) << s);
y = ((out[j] & masks[d][1]) >> s) | (out[j + s] & masks[d][1]);
out[j + 0] = x;
out[j + s] = y;
}
}
}
}
/* return number of trailing zeros of the non-zero input in */
static inline int ctz(uint64_t in) {
return (int)_tzcnt_u64(in);
}
static inline uint64_t same_mask(uint16_t x, uint16_t y) {
uint64_t mask;
mask = x ^ y;
mask -= 1;
mask >>= 63;
mask = -mask;
return mask;
}
static int mov_columns(uint64_t mat[][ ((SYS_N + 255) / 256) * 4 ], uint32_t *perm) {
int i, j, k, s, block_idx, row;
uint64_t buf[64], ctz_list[32], t, d, mask;
row = GFBITS * SYS_T - 32;
block_idx = row / 64;
// extract the 32x64 matrix
for (i = 0; i < 32; i++) {
buf[i] = mat[ row + i ][ block_idx ];
}
// compute the column indices of pivots by Gaussian elimination.
// the indices are stored in ctz_list
for (i = 0; i < 32; i++) {
t = buf[i];
for (j = i + 1; j < 32; j++) {
t |= buf[j];
}
if (t == 0) {
return -1; // return if buf is not full rank
}
ctz_list[i] = s = ctz(t);
for (j = i + 1; j < 32; j++) {
mask = (buf[i] >> s) & 1;
mask -= 1;
buf[i] ^= buf[j] & mask;
}
for (j = 0; j < i; j++) {
mask = (buf[j] >> s) & 1;
mask = -mask;
buf[j] ^= buf[i] & mask;
}
for (j = i + 1; j < 32; j++) {
mask = (buf[j] >> s) & 1;
mask = -mask;
buf[j] ^= buf[i] & mask;
}
}
// updating permutation
for (j = 0; j < 32; j++) {
for (k = j + 1; k < 64; k++) {
d = perm[ row + j ] ^ perm[ row + k ];
d &= same_mask(k, ctz_list[j]);
perm[ row + j ] ^= d;
perm[ row + k ] ^= d;
}
}
// moving columns of mat according to the column indices of pivots
for (i = 0; i < GFBITS * SYS_T; i += 64) {
for (j = 0; j < min(64, GFBITS * SYS_T - i); j++) {
buf[j] = mat[ i + j ][ block_idx ];
}
transpose_64x64(buf, buf);
for (j = 0; j < 32; j++) {
for (k = j + 1; k < 64; k++) {
d = buf[ j ] ^ buf[ k ];
d &= same_mask(k, ctz_list[j]);
buf[ j ] ^= d;
buf[ k ] ^= d;
}
}
transpose_64x64(buf, buf);
for (j = 0; j < min(64, GFBITS * SYS_T - i); j++) {
mat[ i + j ][ block_idx ] = buf[j];
}
}
return 0;
}
#define NBLOCKS1_H ((SYS_N + 63) / 64)
#define NBLOCKS2_H ((SYS_N + 255) / 256)
#define NBLOCKS1_I ((GFBITS * SYS_T + 63) / 64)
#define NBLOCKS2_I ((GFBITS * SYS_T + 255) / 256)
int PQCLEAN_MCELIECE460896F_AVX_pk_gen(unsigned char *pk, uint32_t *perm, const unsigned char *sk) {
int tail = (GFBITS * SYS_T) % 64;
int i, j, k;
int row, c;
uint64_t mat[ GFBITS * SYS_T ][ NBLOCKS2_H * 4 ];
uint64_t mask;
vec128 sk_int[ GFBITS ];
vec256 consts[ 32 ][ GFBITS ];
vec256 eval[ 32 ][ GFBITS ];
vec256 prod[ 32 ][ GFBITS ];
vec256 tmp[ GFBITS ];
uint64_t list[1 << GFBITS];
// compute the inverses
PQCLEAN_MCELIECE460896F_AVX_irr_load(sk_int, sk);
PQCLEAN_MCELIECE460896F_AVX_fft(eval, sk_int);
PQCLEAN_MCELIECE460896F_AVX_vec256_copy(prod[0], eval[0]);
for (i = 1; i < 32; i++) {
PQCLEAN_MCELIECE460896F_AVX_vec256_mul(prod[i], prod[i - 1], eval[i]);
}
PQCLEAN_MCELIECE460896F_AVX_vec256_inv(tmp, prod[31]);
for (i = 30; i >= 0; i--) {
PQCLEAN_MCELIECE460896F_AVX_vec256_mul(prod[i + 1], prod[i], tmp);
PQCLEAN_MCELIECE460896F_AVX_vec256_mul(tmp, tmp, eval[i + 1]);
}
PQCLEAN_MCELIECE460896F_AVX_vec256_copy(prod[0], tmp);
// fill matrix
de_bitslicing(list, prod);
for (i = 0; i < (1 << GFBITS); i++) {
list[i] <<= GFBITS;
list[i] |= i;
list[i] |= ((uint64_t) perm[i]) << 31;
}
PQCLEAN_MCELIECE460896F_AVX_sort_63b(1 << GFBITS, list);
to_bitslicing_2x(consts, prod, list);
for (i = 0; i < (1 << GFBITS); i++) {
perm[i] = list[i] & GFMASK;
}
for (j = 0; j < NBLOCKS2_H; j++) {
for (k = 0; k < GFBITS; k++) {
mat[ k ][ 4 * j + 0 ] = PQCLEAN_MCELIECE460896F_AVX_vec256_extract(prod[ j ][ k ], 0);
mat[ k ][ 4 * j + 1 ] = PQCLEAN_MCELIECE460896F_AVX_vec256_extract(prod[ j ][ k ], 1);
mat[ k ][ 4 * j + 2 ] = PQCLEAN_MCELIECE460896F_AVX_vec256_extract(prod[ j ][ k ], 2);
mat[ k ][ 4 * j + 3 ] = PQCLEAN_MCELIECE460896F_AVX_vec256_extract(prod[ j ][ k ], 3);
}
}
for (i = 1; i < SYS_T; i++) {
for (j = 0; j < NBLOCKS2_H; j++) {
PQCLEAN_MCELIECE460896F_AVX_vec256_mul(prod[j], prod[j], consts[j]);
for (k = 0; k < GFBITS; k++) {
mat[ i * GFBITS + k ][ 4 * j + 0 ] = PQCLEAN_MCELIECE460896F_AVX_vec256_extract(prod[ j ][ k ], 0);
mat[ i * GFBITS + k ][ 4 * j + 1 ] = PQCLEAN_MCELIECE460896F_AVX_vec256_extract(prod[ j ][ k ], 1);
mat[ i * GFBITS + k ][ 4 * j + 2 ] = PQCLEAN_MCELIECE460896F_AVX_vec256_extract(prod[ j ][ k ], 2);
mat[ i * GFBITS + k ][ 4 * j + 3 ] = PQCLEAN_MCELIECE460896F_AVX_vec256_extract(prod[ j ][ k ], 3);
}
}
}
// gaussian elimination
for (row = 0; row < PK_NROWS; row++) {
i = row >> 6;
j = row & 63;
if (row == GFBITS * SYS_T - 32) {
if (mov_columns(mat, perm)) {
return -1;
}
}
for (k = row + 1; k < PK_NROWS; k++) {
mask = mat[ row ][ i ] >> j;
mask &= 1;
mask -= 1;
for (c = 0; c < NBLOCKS1_H; c++) {
mat[ row ][ c ] ^= mat[ k ][ c ] & mask;
}
}
if ( ((mat[ row ][ i ] >> j) & 1) == 0 ) { // return if not systematic
return -1;
}
for (k = 0; k < row; k++) {
mask = mat[ k ][ i ] >> j;
mask &= 1;
mask = -mask;
for (c = 0; c < NBLOCKS1_H; c++) {
mat[ k ][ c ] ^= mat[ row ][ c ] & mask;
}
}
for (k = row + 1; k < PK_NROWS; k++) {
mask = mat[ k ][ i ] >> j;
mask &= 1;
mask = -mask;
for (c = 0; c < NBLOCKS1_H; c++) {
mat[ k ][ c ] ^= mat[ row ][ c ] & mask;
}
}
}
for (i = 0; i < GFBITS * SYS_T; i++) {
PQCLEAN_MCELIECE460896F_AVX_store_i(pk, mat[i][ NBLOCKS1_I - 1 ] >> tail, (64 - tail) / 8);
pk += (64 - tail) / 8;
for (j = NBLOCKS1_I; j < NBLOCKS1_H; j++) {
PQCLEAN_MCELIECE460896F_AVX_store8(pk, mat[i][j]);
pk += 8;
}
}
//
return 0;
}