/*

 * Copyright 2024-2026 The OpenSSL Project Authors. All Rights Reserved.

 *

 * Licensed under the Apache License 2.0 (the "License").  You may not use

 * this file except in compliance with the License.  You can obtain a copy

 * in the file LICENSE in the source distribution or at

 * https://www.openssl.org/source/license.html

 */

#include <openssl/byteorder.h>

#include <openssl/crypto.h>

#include "ml_dsa_local.h"

#include "ml_dsa_vector.h"

#include "ml_dsa_matrix.h"

#include "ml_dsa_hash.h"

#include "internal/constant_time.h"

#include "internal/sha3.h"

#include "internal/packet.h"

#define SHAKE128_BLOCKSIZE SHA3_BLOCKSIZE(128)

#define SHAKE256_BLOCKSIZE SHA3_BLOCKSIZE(256)

/*

 * This is a constant time version of n % 5

 * Note that 0xFFFF / 5 = 0x3333, 2 is added to make an over-estimate of 1/5

 * and then we divide by (0xFFFF + 1)

 */

#define MOD5(n) ((n) - 5 * (0x3335 * (n) >> 16))

#if SHAKE128_BLOCKSIZE % 3 != 0

#error "rej_ntt_poly() requires SHAKE128_BLOCKSIZE to be a multiple of 3"

#endif

typedef int(COEFF_FROM_NIBBLE_FUNC)(uint32_t nibble, uint32_t *out);

static COEFF_FROM_NIBBLE_FUNC coeff_from_nibble_4;

static COEFF_FROM_NIBBLE_FUNC coeff_from_nibble_2;

static ML_DSA_MATRIX_EXPAND_A_FN matrix_expand_A_scalar;

static ML_DSA_VECTOR_EXPAND_S_FN vector_expand_S_scalar;

static ML_DSA_VECTOR_EXPAND_MASK_FN vector_expand_mask_scalar;

/**

 * @brief Combine 3 bytes to form an coefficient.

 * See FIPS 204, Algorithm 14, CoeffFromThreeBytes()

 *

 * This is not constant time as it is used to generate the matrix A which is public.

 *

 * @param s A byte array of 3 uniformly distributed bytes.

 * @param out The returned coefficient in the range 0..q-1.

 * @returns 1 if the value is less than q or 0 otherwise.

 *          This is used for rejection sampling.

 */

static ossl_inline int coeff_from_three_bytes(const uint8_t *s, uint32_t *out)

{

    /* Zero out the top bit of the 3rd byte to get a value in the range 0..2^23-1) */

    *out = (uint32_t)s[0] | ((uint32_t)s[1] << 8) | (((uint32_t)s[2] & 0x7f) << 16);

    return *out < ML_DSA_Q;

}

/**

 * @brief Generate a value in the range (q-4..0..4)

 * See FIPS 204, Algorithm 15, CoeffFromHalfByte() where eta = 4

 * Note the FIPS 204 code uses the range -4..4 (whereas this code adds q to the

 * negative numbers).

 *

 * @param nibble A value in the range 0..15

 * @param out The returned value if the range (q-4)..0..4 if nibble is < 9

 * @returns 1 nibble was in range, or 0 if the nibble was rejected.

 */

static ossl_inline int coeff_from_nibble_4(uint32_t nibble, uint32_t *out)

{

    /*

     * This is not constant time but will not leak any important info since

     * the value is either chosen or thrown away.

     */

    if (value_barrier_32(nibble < 9)) {

        *out = mod_sub(4, nibble);

        return 1;

    }

    return 0;

}

/**

 * @brief Generate a value in the range (q-2..0..2)

 * See FIPS 204, Algorithm 15, CoeffFromHalfByte() where eta = 2

 * Note the FIPS 204 code uses the range -2..2 (whereas this code adds q to the

 * negative numbers).

 *

 * @param nibble A value in the range 0..15

 * @param out The returned value if the range (q-2)..0..2 if nibble is < 15

 * @returns 1 nibble was in range, or 0 if the nibble was rejected.

 */

static ossl_inline int coeff_from_nibble_2(uint32_t nibble, uint32_t *out)

{

    if (value_barrier_32(nibble < 15)) {

        *out = mod_sub(2, MOD5(nibble));

        return 1;

    }

    return 0;

}

/**

 * @brief Use a seed value to generate a polynomial with coefficients in the

 * range of 0..q-1 using rejection sampling.

 * SHAKE128 is used to absorb the seed, and then sequences of 3 sample bytes are

 * squeezed to try to produce coefficients.

 * The SHAKE128 stream is used to get uniformly distributed elements.

 * This algorithm is used for matrix expansion and only operates on public inputs.

 *

 * See FIPS 204, Algorithm 30, RejNTTPoly()

 *

 * @param g_ctx A EVP_MD_CTX object used for sampling the seed.

 * @param md A pre-fetched SHAKE128 object.

 * @param seed The seed to use for sampling.

 * @param seed_len The size of |seed|

 * @param out The returned polynomial with coefficients in the range of

 *            0..q-1. This range is required for NTT.

 * @returns 1 if the polynomial was successfully generated, or 0 if any of the

 *            digest operations failed.

 */

static int rej_ntt_poly(EVP_MD_CTX *g_ctx, const EVP_MD *md,

    const uint8_t *seed, size_t seed_len, POLY *out)

{

    int j = 0;

    uint8_t blocks[SHAKE128_BLOCKSIZE], *b, *end = blocks + sizeof(blocks);

    /*

     * Instead of just squeezing 3 bytes at a time, we grab a whole block

     * Note that the shake128 blocksize of 168 is divisible by 3.

     */

    if (!shake_xof(g_ctx, md, seed, seed_len, blocks, sizeof(blocks)))

        return 0;

    while (1) {

        for (b = blocks; b < end; b += 3) {

            if (coeff_from_three_bytes(b, &(out->coeff[j]))) {

                if (++j >= ML_DSA_NUM_POLY_COEFFICIENTS)

                    return 1; /* finished */

            }

        }

        if (!EVP_DigestSqueeze(g_ctx, blocks, sizeof(blocks)))

            return 0;

    }

}

/**

 * @brief Use a seed value to generate a polynomial with coefficients in the

 * range of ((q-eta)..0..eta) using rejection sampling. eta is either 2 or 4.

 * SHAKE256 is used to absorb the seed, and then samples are squeezed.

 * See FIPS 204, Algorithm 31, RejBoundedPoly()

 *

 * @param h_ctx A EVP_MD_CTX object context used to sample the seed.

 * @param md A pre-fetched SHAKE256 object.

 * @param coef_from_nibble A function that is dependent on eta, which takes a

 *                         nibble and tries to see if it is in the correct range.

 * @param seed The seed to use for sampling.

 * @param seed_len The size of |seed|

 * @param out The returned polynomial with coefficients in the range of

 *            ((q-eta)..0..eta)

 * @returns 1 if the polynomial was successfully generated, or 0 if any of the

 *            digest operations failed.

 */

static int rej_bounded_poly(EVP_MD_CTX *h_ctx, const EVP_MD *md,

    COEFF_FROM_NIBBLE_FUNC *coef_from_nibble,

    const uint8_t *seed, size_t seed_len, POLY *out)

{

    int ret = 0;

    int j = 0;

    uint32_t z0, z1;

    uint8_t blocks[SHAKE256_BLOCKSIZE], *b, *end = blocks + sizeof(blocks);

    /* Instead of just squeezing 1 byte at a time, we grab a whole block */

    if (!shake_xof(h_ctx, md, seed, seed_len, blocks, sizeof(blocks)))

        goto err;

    while (1) {

        for (b = blocks; b < end; b++) {

            z0 = *b & 0x0F; /* lower nibble of byte */

            z1 = *b >> 4; /* high nibble of byte */

            if (coef_from_nibble(z0, &out->coeff[j])

                && ++j >= ML_DSA_NUM_POLY_COEFFICIENTS) {

                ret = 1;

                goto err;

            }

            if (coef_from_nibble(z1, &out->coeff[j])

                && ++j >= ML_DSA_NUM_POLY_COEFFICIENTS) {

                ret = 1;

                goto err;

            }

        }

        if (!EVP_DigestSqueeze(h_ctx, blocks, sizeof(blocks)))

            goto err;

    }

err:

    OPENSSL_cleanse(blocks, sizeof(blocks));

    return ret;

}

/**

 * @brief Generate a k * l matrix that has uniformly distributed polynomial

 *        elements using rejection sampling.

 * See FIPS 204, Algorithm 32, ExpandA()

 *

 * @param g_ctx A EVP_MD_CTX context used for rejection sampling

 *              seed values generated from the seed rho.

 * @param md A pre-fetched SHAKE128 object

 * @param rho A 32 byte seed to generated the matrix from.

 * @param out The generated k * l matrix of polynomials with coefficients

 *            in the range of 0..q-1.

 * @returns 1 if the matrix was generated, or 0 on error.

 */

static int matrix_expand_A_scalar(EVP_MD_CTX *g_ctx, const EVP_MD *md,

    const uint8_t *rho, MATRIX *out)

{

    int ret = 0;

    size_t i, j;

    uint8_t derived_seed[ML_DSA_RHO_BYTES + 2];

    POLY *poly = out->m_poly;

    /*

     * The seeds derived below and the sampling buffers in rej_ntt_poly() are

     * not cleansed: per FIPS 204 section 3.6.3 the matrix A is easily

     * computed from the public key and does not require any special

     * protections.

     */

    /* The seed used for each matrix element is rho + column_index + row_index */

    memcpy(derived_seed, rho, ML_DSA_RHO_BYTES);

    for (i = 0; i < out->k; i++) {

        for (j = 0; j < out->l; j++) {

            derived_seed[ML_DSA_RHO_BYTES + 1] = (uint8_t)i;

            derived_seed[ML_DSA_RHO_BYTES] = (uint8_t)j;

            /* Generate the polynomial for each matrix element using a unique seed */

            if (!rej_ntt_poly(g_ctx, md, derived_seed, sizeof(derived_seed), poly++))

                goto err;

        }

    }

    ret = 1;

err:

    return ret;

}

/**

 * @brief Generates 2 vectors using rejection sampling whose polynomial

 * coefficients are in the interval [q-eta..0..eta]

 *

 * See FIPS 204, Algorithm 33, ExpandS().

 * Note that in FIPS 204 the range -eta..eta is used.

 *

 * @param h_ctx A EVP_MD_CTX context to use to sample the seed.

 * @param md A pre-fetched SHAKE256 object.

 * @param eta Is either 2 or 4, and determines the range of the coefficients for

 *            s1 and s2.

 * @param seed A 64 byte seed to use for sampling.

 * @param s1 A 1 * l column vector containing polynomials with coefficients in

 *           the range (q-eta)..0..eta

 * @param s2 A 1 * k column vector containing polynomials with coefficients in

 *           the range (q-eta)..0..eta

 * @returns 1 if s1 and s2 were successfully generated, or 0 otherwise.

 */

static int vector_expand_S_scalar(EVP_MD_CTX *h_ctx, const EVP_MD *md, int eta,

    const uint8_t *seed, VECTOR *s1, VECTOR *s2)

{

    int ret = 0;

    size_t i;

    size_t l = s1->num_poly;

    size_t k = s2->num_poly;

    uint8_t derived_seed[ML_DSA_PRIV_SEED_BYTES + 2];

    COEFF_FROM_NIBBLE_FUNC *coef_from_nibble_fn;

    coef_from_nibble_fn = (eta == ML_DSA_ETA_4) ? coeff_from_nibble_4 : coeff_from_nibble_2;

    /*

     * Each polynomial generated uses a unique seed that consists of

     * seed + counter (where the counter is 2 bytes starting at 0)

     */

    memcpy(derived_seed, seed, ML_DSA_PRIV_SEED_BYTES);

    derived_seed[ML_DSA_PRIV_SEED_BYTES] = 0;

    derived_seed[ML_DSA_PRIV_SEED_BYTES + 1] = 0;

    for (i = 0; i < l; i++) {

        if (!rej_bounded_poly(h_ctx, md, coef_from_nibble_fn,

                derived_seed, sizeof(derived_seed), &s1->poly[i]))

            goto err;

        ++derived_seed[ML_DSA_PRIV_SEED_BYTES];

    }

    for (i = 0; i < k; i++) {

        if (!rej_bounded_poly(h_ctx, md, coef_from_nibble_fn,

                derived_seed, sizeof(derived_seed), &s2->poly[i]))

            goto err;

        ++derived_seed[ML_DSA_PRIV_SEED_BYTES];

    }

    ret = 1;

err:

    OPENSSL_cleanse(derived_seed, sizeof(derived_seed));

    return ret;

}

/* See FIPS 204, Algorithm 34, ExpandMask(), Step 4 & 5 */

int ossl_ml_dsa_poly_expand_mask(POLY *out, const uint8_t *seed, size_t seed_len,

    uint32_t gamma1,

    EVP_MD_CTX *h_ctx, const EVP_MD *md)

{

    uint8_t buf[32 * 20];

    size_t buf_len = 32 * (gamma1 == ML_DSA_GAMMA1_TWO_POWER_19 ? 20 : 18);

    int ret = shake_xof(h_ctx, md, seed, seed_len, buf, buf_len)

        && ossl_ml_dsa_poly_decode_expand_mask(out, buf, buf_len, gamma1);

    OPENSSL_cleanse(buf, sizeof(buf));

    return ret;

}

/*

 * @brief Sample a polynomial with coefficients in the range {-1..1}.

 * The number of non zero values (hamming weight) is given by tau

 *

 * See FIPS 204, Algorithm 29, SampleInBall()

 * This function is assumed to not be constant time.

 * The algorithm is based on Durstenfeld's version of the Fisher-Yates shuffle.

 *

 * Note that the coefficients returned by this implementation are positive

 * i.e one of q-1, 0, or 1.

 *

 * @param tau is the number of +1 or -1's in the polynomial 'out_c' (39, 49 or 60)

 *            that is less than or equal to 64

 */

int ossl_ml_dsa_poly_sample_in_ball(POLY *out_c, const uint8_t *seed, int seed_len,

    EVP_MD_CTX *h_ctx, const EVP_MD *md,

    uint32_t tau)

{

    uint8_t block[SHAKE256_BLOCKSIZE];

    uint64_t signs;

    int offset = 8;

    size_t end;

    int ret = 0;

    /*

     * Rather than squeeze 8 bytes followed by lots of 1 byte squeezes

     * the SHAKE blocksize is squeezed each time and buffered into 'block'.

     */

    if (!shake_xof(h_ctx, md, seed, seed_len, block, sizeof(block)))

        goto err;

    /*

     * grab the first 64 bits - since tau < 64

     * Each bit gives a +1 or -1 value.

     */

    OPENSSL_load_u64_le(&signs, block);

    /*

     * SampleInBall implements a Fisher-Yates shuffle whose rejection-sampling

     * inner loop and data-dependent array index unavoidably leak the structure

     * of the challenge polynomial via memory-access pattern and branch timing.

     * This is safe: c_tilde = H(mu ‖ w1) is the Fiat-Shamir commitment and is

     * published in the accepted signature, so the SHAKE bytes that build c are

     * effectively public.  See the BoringSSL design discussion at

     * https://boringssl-review.googlesource.com/c/boringssl/+/67747/comment/8d8f01ac_70af3f21/

     *

     * The first 8 bytes (the sign bits loaded into |signs| above) are left

     * tainted: they determine only the ±1 values written into c, which flow

     * into the CT arithmetic of cs1/cs2/ct0 alongside the already-tainted

     * secret polynomials and cause no spurious violations there.

     * Only the rejection-sampling bytes need to be declassified.

     */

    CONSTTIME_DECLASSIFY(block + offset, sizeof(block) - offset);

    poly_zero(out_c);

    /* Loop tau times */

    for (end = 256 - tau; end < 256; end++) {

        size_t index; /* index is a random offset to write +1 or -1 */

        /* rejection sample in {0..end} to choose an index to place -1 or 1 into */

        for (;;) {

            if (offset == sizeof(block)) {

                /* squeeze another block if the bytes from block have been used */

                if (!EVP_DigestSqueeze(h_ctx, block, sizeof(block)))

                    goto err;

                /* See comment above for why the block is declassified. */

                CONSTTIME_DECLASSIFY(block, sizeof(block));

                offset = 0;

            }

            index = block[offset++];

            if (index <= end)

                break;

        }

        /*

         * In-place swap the coefficient we are about to replace to the end so

         * we don't lose any values that have been already written.

         */

        out_c->coeff[end] = out_c->coeff[index];

        /* set the random coefficient value to either 1 or q-1 */

        out_c->coeff[index] = mod_sub(1, 2 * (signs & 1));

        signs >>= 1; /* grab the next random bit */

    }

    ret = 1;

err:

    OPENSSL_cleanse(block, sizeof(block));

    return ret;

}

static void vector_expand_mask_scalar(VECTOR *out,

    const uint8_t rho_prime[ML_DSA_RHO_PRIME_BYTES], uint32_t kappa, uint32_t gamma1,

    EVP_MD_CTX *h_ctx, const EVP_MD *md)

{

    size_t i;

    uint8_t derived_seed[ML_DSA_RHO_PRIME_BYTES + 2];

    memcpy(derived_seed, rho_prime, ML_DSA_RHO_PRIME_BYTES);

    for (i = 0; i < out->num_poly; i++) {

        size_t index = kappa + i;

        derived_seed[ML_DSA_RHO_PRIME_BYTES] = index & 0xFF;

        derived_seed[ML_DSA_RHO_PRIME_BYTES + 1] = (index >> 8) & 0xFF;

        poly_expand_mask(out->poly + i, derived_seed, sizeof(derived_seed),

            gamma1, h_ctx, md);

    }

    OPENSSL_cleanse(derived_seed, sizeof(derived_seed));

}

static const OSSL_ML_DSA_SAMPLE_OPS ml_dsa_sample_generic_meth = {

    matrix_expand_A_scalar,

    vector_expand_S_scalar,

    vector_expand_mask_scalar

};

#if defined(KECCAK1600_ASM)                                                               \

    && (defined(__x86_64) || defined(__x86_64__) || defined(_M_AMD64) || defined(_M_X64)) \

    && !defined(OPENSSL_NO_ASM)

#include "ml_dsa_sample_hw_x86_64.inc"

const OSSL_ML_DSA_SAMPLE_OPS *ossl_ml_dsa_sample_ops(void)

{

    if (SHA3_avx512vl_capable())

        return &ml_dsa_sample_x86_64;

    return &ml_dsa_sample_generic_meth;

}

#else

const OSSL_ML_DSA_SAMPLE_OPS *ossl_ml_dsa_sample_ops(void)

{

    return &ml_dsa_sample_generic_meth;

}

#endif