crypto/mldsa/mldsa.cc

/* Copyright (c) 2024, Google LLC
 *
 * Permission to use, copy, modify, and/or distribute this software for any
 * purpose with or without fee is hereby granted, provided that the above
 * copyright notice and this permission notice appear in all copies.
 *
 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY
 * SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION
 * OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN
 * CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. */

#include <openssl/mldsa.h>

#include <assert.h>
#include <stdlib.h>

#include <openssl/bytestring.h>
#include <openssl/mem.h>
#include <openssl/rand.h>

#include "../internal.h"
#include "../keccak/internal.h"
#include "./internal.h"

#define DEGREE 256
#define K 6
#define L 5
#define ETA 4
#define TAU 49
#define BETA 196
#define OMEGA 55

#define RHO_BYTES 32
#define SIGMA_BYTES 64
#define K_BYTES 32
#define TR_BYTES 64
#define MU_BYTES 64
#define RHO_PRIME_BYTES 64
#define LAMBDA_BITS 192
#define LAMBDA_BYTES (LAMBDA_BITS / 8)

// 2^23 - 2^13 + 1
static const uint32_t kPrime = 8380417;
// Inverse of -kPrime modulo 2^32
static const uint32_t kPrimeNegInverse = 4236238847;
static const int kDroppedBits = 13;
static const uint32_t kHalfPrime = (8380417 - 1) / 2;
static const uint32_t kGamma1 = 1 << 19;
static const uint32_t kGamma2 = (8380417 - 1) / 32;
// 256^-1 mod kPrime, in Montgomery form.
static const uint32_t kInverseDegreeMontgomery = 41978;

typedef struct scalar {
  uint32_t c[DEGREE];
} scalar;

typedef struct vectork {
  scalar v[K];
} vectork;

typedef struct vectorl {
  scalar v[L];
} vectorl;

typedef struct matrix {
  scalar v[K][L];
} matrix;

/* Arithmetic */

// This bit of Python will be referenced in some of the following comments:
//
// q = 8380417
// # Inverse of -q modulo 2^32
// q_neg_inverse = 4236238847
// # 2^64 modulo q
// montgomery_square = 2365951
//
// def bitreverse(i):
//     ret = 0
//     for n in range(8):
//         bit = i & 1
//         ret <<= 1
//         ret |= bit
//         i >>= 1
//     return ret
//
// def montgomery_reduce(x):
//     a = (x * q_neg_inverse) % 2**32
//     b = x + a * q
//     assert b & 0xFFFF_FFFF == 0
//     c = b >> 32
//     assert c < q
//     return c
//
// def montgomery_transform(x):
//     return montgomery_reduce(x * montgomery_square)

// kNTTRootsMontgomery = [
//   montgomery_transform(pow(1753, bitreverse(i), q)) for i in range(256)
// ]
static const uint32_t kNTTRootsMontgomery[256] = {
    4193792, 25847,   5771523, 7861508, 237124,  7602457, 7504169, 466468,
    1826347, 2353451, 8021166, 6288512, 3119733, 5495562, 3111497, 2680103,
    2725464, 1024112, 7300517, 3585928, 7830929, 7260833, 2619752, 6271868,
    6262231, 4520680, 6980856, 5102745, 1757237, 8360995, 4010497, 280005,
    2706023, 95776,   3077325, 3530437, 6718724, 4788269, 5842901, 3915439,
    4519302, 5336701, 3574422, 5512770, 3539968, 8079950, 2348700, 7841118,
    6681150, 6736599, 3505694, 4558682, 3507263, 6239768, 6779997, 3699596,
    811944,  531354,  954230,  3881043, 3900724, 5823537, 2071892, 5582638,
    4450022, 6851714, 4702672, 5339162, 6927966, 3475950, 2176455, 6795196,
    7122806, 1939314, 4296819, 7380215, 5190273, 5223087, 4747489, 126922,
    3412210, 7396998, 2147896, 2715295, 5412772, 4686924, 7969390, 5903370,
    7709315, 7151892, 8357436, 7072248, 7998430, 1349076, 1852771, 6949987,
    5037034, 264944,  508951,  3097992, 44288,   7280319, 904516,  3958618,
    4656075, 8371839, 1653064, 5130689, 2389356, 8169440, 759969,  7063561,
    189548,  4827145, 3159746, 6529015, 5971092, 8202977, 1315589, 1341330,
    1285669, 6795489, 7567685, 6940675, 5361315, 4499357, 4751448, 3839961,
    2091667, 3407706, 2316500, 3817976, 5037939, 2244091, 5933984, 4817955,
    266997,  2434439, 7144689, 3513181, 4860065, 4621053, 7183191, 5187039,
    900702,  1859098, 909542,  819034,  495491,  6767243, 8337157, 7857917,
    7725090, 5257975, 2031748, 3207046, 4823422, 7855319, 7611795, 4784579,
    342297,  286988,  5942594, 4108315, 3437287, 5038140, 1735879, 203044,
    2842341, 2691481, 5790267, 1265009, 4055324, 1247620, 2486353, 1595974,
    4613401, 1250494, 2635921, 4832145, 5386378, 1869119, 1903435, 7329447,
    7047359, 1237275, 5062207, 6950192, 7929317, 1312455, 3306115, 6417775,
    7100756, 1917081, 5834105, 7005614, 1500165, 777191,  2235880, 3406031,
    7838005, 5548557, 6709241, 6533464, 5796124, 4656147, 594136,  4603424,
    6366809, 2432395, 2454455, 8215696, 1957272, 3369112, 185531,  7173032,
    5196991, 162844,  1616392, 3014001, 810149,  1652634, 4686184, 6581310,
    5341501, 3523897, 3866901, 269760,  2213111, 7404533, 1717735, 472078,
    7953734, 1723600, 6577327, 1910376, 6712985, 7276084, 8119771, 4546524,
    5441381, 6144432, 7959518, 6094090, 183443,  7403526, 1612842, 4834730,
    7826001, 3919660, 8332111, 7018208, 3937738, 1400424, 7534263, 1976782};

// Reduces x mod kPrime in constant time, where 0 <= x < 2*kPrime.
static uint32_t reduce_once(uint32_t x) {
  declassify_assert(x < 2 * kPrime);
  // return x < kPrime ? x : x - kPrime;
  return constant_time_select_int(constant_time_lt_w(x, kPrime), x, x - kPrime);
}

// Returns the absolute value in constant time.
static uint32_t abs_signed(uint32_t x) {
  // return is_positive(x) ? x : -x;
  // Note: MSVC doesn't like applying the unary minus operator to unsigned types
  // (warning C4146), so we write the negation as a bitwise not plus one
  // (assuming two's complement representation).
  return constant_time_select_int(constant_time_lt_w(x, 0x80000000), x, 0u - x);
}

// Returns the absolute value modulo kPrime.
static uint32_t abs_mod_prime(uint32_t x) {
  declassify_assert(x < kPrime);
  // return x > kHalfPrime ? kPrime - x : x;
  return constant_time_select_int(constant_time_lt_w(kHalfPrime, x), kPrime - x,
                                  x);
}

// Returns the maximum of two values in constant time.
static uint32_t maximum(uint32_t x, uint32_t y) {
  // return x < y ? y : x;
  return constant_time_select_int(constant_time_lt_w(x, y), y, x);
}

static uint32_t mod_sub(uint32_t a, uint32_t b) {
  declassify_assert(a < kPrime);
  declassify_assert(b < kPrime);
  return reduce_once(kPrime + a - b);
}

static void scalar_add(scalar *out, const scalar *lhs, const scalar *rhs) {
  for (int i = 0; i < DEGREE; i++) {
    out->c[i] = reduce_once(lhs->c[i] + rhs->c[i]);
  }
}

static void scalar_sub(scalar *out, const scalar *lhs, const scalar *rhs) {
  for (int i = 0; i < DEGREE; i++) {
    out->c[i] = mod_sub(lhs->c[i], rhs->c[i]);
  }
}

static uint32_t reduce_montgomery(uint64_t x) {
  declassify_assert(x <= ((uint64_t)kPrime << 32));
  uint64_t a = (uint32_t)x * kPrimeNegInverse;
  uint64_t b = x + a * kPrime;
  declassify_assert((b & 0xffffffff) == 0);
  uint32_t c = b >> 32;
  return reduce_once(c);
}

// Multiply two scalars in the number theoretically transformed state.
static void scalar_mult(scalar *out, const scalar *lhs, const scalar *rhs) {
  for (int i = 0; i < DEGREE; i++) {
    out->c[i] = reduce_montgomery((uint64_t)lhs->c[i] * (uint64_t)rhs->c[i]);
  }
}

// In place number theoretic transform of a given scalar.
//
// FIPS 204, Algorithm 41 (`NTT`).
static void scalar_ntt(scalar *s) {
  // Step: 1, 2, 4, 8, ..., 128
  // Offset: 128, 64, 32, 16, ..., 1
  int offset = DEGREE;
  for (int step = 1; step < DEGREE; step <<= 1) {
    offset >>= 1;
    int k = 0;
    for (int i = 0; i < step; i++) {
      assert(k == 2 * offset * i);
      const uint32_t step_root = kNTTRootsMontgomery[step + i];
      for (int j = k; j < k + offset; j++) {
        uint32_t even = s->c[j];
        // |reduce_montgomery| works on values up to kPrime*R and R > 2*kPrime.
        // |step_root| < kPrime because it's static data. |s->c[...]| is <
        // kPrime by the invariants of that struct.
        uint32_t odd =
            reduce_montgomery((uint64_t)step_root * (uint64_t)s->c[j + offset]);
        s->c[j] = reduce_once(odd + even);
        s->c[j + offset] = mod_sub(even, odd);
      }
      k += 2 * offset;
    }
  }
}

// In place inverse number theoretic transform of a given scalar.
//
// FIPS 204, Algorithm 42 (`NTT^-1`).
static void scalar_inverse_ntt(scalar *s) {
  // Step: 128, 64, 32, 16, ..., 1
  // Offset: 1, 2, 4, 8, ..., 128
  int step = DEGREE;
  for (int offset = 1; offset < DEGREE; offset <<= 1) {
    step >>= 1;
    int k = 0;
    for (int i = 0; i < step; i++) {
      assert(k == 2 * offset * i);
      const uint32_t step_root =
          kPrime - kNTTRootsMontgomery[step + (step - 1 - i)];
      for (int j = k; j < k + offset; j++) {
        uint32_t even = s->c[j];
        uint32_t odd = s->c[j + offset];
        s->c[j] = reduce_once(odd + even);

        // |reduce_montgomery| works on values up to kPrime*R and R > 2*kPrime.
        // kPrime + even < 2*kPrime because |even| < kPrime, by the invariants
        // of that structure. Thus kPrime + even - odd < 2*kPrime because odd >=
        // 0, because it's unsigned and less than kPrime. Lastly step_root <
        // kPrime, because |kNTTRootsMontgomery| is static data.
        s->c[j + offset] = reduce_montgomery((uint64_t)step_root *
                                             (uint64_t)(kPrime + even - odd));
      }
      k += 2 * offset;
    }
  }
  for (int i = 0; i < DEGREE; i++) {
    s->c[i] = reduce_montgomery((uint64_t)s->c[i] *
                                (uint64_t)kInverseDegreeMontgomery);
  }
}

static void vectork_zero(vectork *out) { OPENSSL_memset(out, 0, sizeof(*out)); }

static void vectork_add(vectork *out, const vectork *lhs, const vectork *rhs) {
  for (int i = 0; i < K; i++) {
    scalar_add(&out->v[i], &lhs->v[i], &rhs->v[i]);
  }
}

static void vectork_sub(vectork *out, const vectork *lhs, const vectork *rhs) {
  for (int i = 0; i < K; i++) {
    scalar_sub(&out->v[i], &lhs->v[i], &rhs->v[i]);
  }
}

static void vectork_mult_scalar(vectork *out, const vectork *lhs,
                                const scalar *rhs) {
  for (int i = 0; i < K; i++) {
    scalar_mult(&out->v[i], &lhs->v[i], rhs);
  }
}

static void vectork_ntt(vectork *a) {
  for (int i = 0; i < K; i++) {
    scalar_ntt(&a->v[i]);
  }
}

static void vectork_inverse_ntt(vectork *a) {
  for (int i = 0; i < K; i++) {
    scalar_inverse_ntt(&a->v[i]);
  }
}

static void vectorl_add(vectorl *out, const vectorl *lhs, const vectorl *rhs) {
  for (int i = 0; i < L; i++) {
    scalar_add(&out->v[i], &lhs->v[i], &rhs->v[i]);
  }
}

static void vectorl_mult_scalar(vectorl *out, const vectorl *lhs,
                                const scalar *rhs) {
  for (int i = 0; i < L; i++) {
    scalar_mult(&out->v[i], &lhs->v[i], rhs);
  }
}

static void vectorl_ntt(vectorl *a) {
  for (int i = 0; i < L; i++) {
    scalar_ntt(&a->v[i]);
  }
}

static void vectorl_inverse_ntt(vectorl *a) {
  for (int i = 0; i < L; i++) {
    scalar_inverse_ntt(&a->v[i]);
  }
}

static void matrix_mult(vectork *out, const matrix *m, const vectorl *a) {
  vectork_zero(out);
  for (int i = 0; i < K; i++) {
    for (int j = 0; j < L; j++) {
      scalar product;
      scalar_mult(&product, &m->v[i][j], &a->v[j]);
      scalar_add(&out->v[i], &out->v[i], &product);
    }
  }
}

/* Rounding & hints */

// FIPS 204, Algorithm 35 (`Power2Round`).
static void power2_round(uint32_t *r1, uint32_t *r0, uint32_t r) {
  *r1 = r >> kDroppedBits;
  *r0 = r - (*r1 << kDroppedBits);

  uint32_t r0_adjusted = mod_sub(*r0, 1 << kDroppedBits);
  uint32_t r1_adjusted = *r1 + 1;

  // Mask is set iff r0 > 2^(dropped_bits - 1).
  crypto_word_t mask =
      constant_time_lt_w((uint32_t)(1 << (kDroppedBits - 1)), *r0);
  // r0 = mask ? r0_adjusted : r0
  *r0 = constant_time_select_int(mask, r0_adjusted, *r0);
  // r1 = mask ? r1_adjusted : r1
  *r1 = constant_time_select_int(mask, r1_adjusted, *r1);
}

// Scale back previously rounded value.
static void scale_power2_round(uint32_t *out, uint32_t r1) {
  // Pre-condition: 0 <= r1 <= 2^10 - 1
  assert(r1 < (1u << 10));

  *out = r1 << kDroppedBits;

  // Post-condition: 0 <= out <= 2^23 - 2^13 = kPrime - 1
  assert(*out < kPrime);
}

// FIPS 204, Algorithm 37 (`HighBits`).
static uint32_t high_bits(uint32_t x) {
  // Reference description (given 0 <= x < q):
  //
  // ```
  // int32_t r0 = x mod+- (2 * kGamma2);
  // if (x - r0 == q - 1) {
  //   return 0;
  // } else {
  //   return (x - r0) / (2 * kGamma2);
  // }
  // ```
  //
  // Below is the formula taken from the reference implementation.
  //
  // Here, kGamma2 == 2^18 - 2^8
  // This returns ((ceil(x / 2^7) * (2^10 + 1) + 2^21) / 2^22) mod 2^4
  uint32_t r1 = (x + 127) >> 7;
  r1 = (r1 * 1025 + (1 << 21)) >> 22;
  r1 &= 15;
  return r1;
}

// FIPS 204, Algorithm 36 (`Decompose`).
static void decompose(uint32_t *r1, int32_t *r0, uint32_t r) {
  *r1 = high_bits(r);

  *r0 = r;
  *r0 -= *r1 * 2 * (int32_t)kGamma2;
  *r0 -= (((int32_t)kHalfPrime - *r0) >> 31) & (int32_t)kPrime;
}

// FIPS 204, Algorithm 38 (`LowBits`).
static int32_t low_bits(uint32_t x) {
  uint32_t r1;
  int32_t r0;
  decompose(&r1, &r0, x);
  return r0;
}

// FIPS 204, Algorithm 39 (`MakeHint`).
//
// In the spec this takes two arguments, z and r, and is called with
//   z = -ct0
//   r = w - cs2 + ct0
//
// It then computes HighBits (algorithm 37) of z and z+r. But z+r is just w -
// cs2, so this takes three arguments and saves an addition.
static int32_t make_hint(uint32_t ct0, uint32_t cs2, uint32_t w) {
  uint32_t r_plus_z = mod_sub(w, cs2);
  uint32_t r = reduce_once(r_plus_z + ct0);
  return high_bits(r) != high_bits(r_plus_z);
}

// FIPS 204, Algorithm 40 (`UseHint`).
static uint32_t use_hint_vartime(uint32_t h, uint32_t r) {
  uint32_t r1;
  int32_t r0;
  decompose(&r1, &r0, r);

  if (h) {
    if (r0 > 0) {
      // m = 16, thus |mod m| in the spec turns into |& 15|.
      return (r1 + 1) & 15;
    } else {
      return (r1 - 1) & 15;
    }
  }
  return r1;
}

static void scalar_power2_round(scalar *s1, scalar *s0, const scalar *s) {
  for (int i = 0; i < DEGREE; i++) {
    power2_round(&s1->c[i], &s0->c[i], s->c[i]);
  }
}

static void scalar_scale_power2_round(scalar *out, const scalar *in) {
  for (int i = 0; i < DEGREE; i++) {
    scale_power2_round(&out->c[i], in->c[i]);
  }
}

static void scalar_high_bits(scalar *out, const scalar *in) {
  for (int i = 0; i < DEGREE; i++) {
    out->c[i] = high_bits(in->c[i]);
  }
}

static void scalar_low_bits(scalar *out, const scalar *in) {
  for (int i = 0; i < DEGREE; i++) {
    out->c[i] = low_bits(in->c[i]);
  }
}

static void scalar_max(uint32_t *max, const scalar *s) {
  for (int i = 0; i < DEGREE; i++) {
    uint32_t abs = abs_mod_prime(s->c[i]);
    *max = maximum(*max, abs);
  }
}

static void scalar_max_signed(uint32_t *max, const scalar *s) {
  for (int i = 0; i < DEGREE; i++) {
    uint32_t abs = abs_signed(s->c[i]);
    *max = maximum(*max, abs);
  }
}

static void scalar_make_hint(scalar *out, const scalar *ct0, const scalar *cs2,
                             const scalar *w) {
  for (int i = 0; i < DEGREE; i++) {
    out->c[i] = make_hint(ct0->c[i], cs2->c[i], w->c[i]);
  }
}

static void scalar_use_hint_vartime(scalar *out, const scalar *h,
                                    const scalar *r) {
  for (int i = 0; i < DEGREE; i++) {
    out->c[i] = use_hint_vartime(h->c[i], r->c[i]);
  }
}

static void vectork_power2_round(vectork *t1, vectork *t0, const vectork *t) {
  for (int i = 0; i < K; i++) {
    scalar_power2_round(&t1->v[i], &t0->v[i], &t->v[i]);
  }
}

static void vectork_scale_power2_round(vectork *out, const vectork *in) {
  for (int i = 0; i < K; i++) {
    scalar_scale_power2_round(&out->v[i], &in->v[i]);
  }
}

static void vectork_high_bits(vectork *out, const vectork *in) {
  for (int i = 0; i < K; i++) {
    scalar_high_bits(&out->v[i], &in->v[i]);
  }
}

static void vectork_low_bits(vectork *out, const vectork *in) {
  for (int i = 0; i < K; i++) {
    scalar_low_bits(&out->v[i], &in->v[i]);
  }
}

static uint32_t vectork_max(const vectork *a) {
  uint32_t max = 0;
  for (int i = 0; i < K; i++) {
    scalar_max(&max, &a->v[i]);
  }
  return max;
}

static uint32_t vectork_max_signed(const vectork *a) {
  uint32_t max = 0;
  for (int i = 0; i < K; i++) {
    scalar_max_signed(&max, &a->v[i]);
  }
  return max;
}

// The input vector contains only zeroes and ones.
static size_t vectork_count_ones(const vectork *a) {
  size_t count = 0;
  for (int i = 0; i < K; i++) {
    for (int j = 0; j < DEGREE; j++) {
      count += a->v[i].c[j];
    }
  }
  return count;
}

static void vectork_make_hint(vectork *out, const vectork *ct0,
                              const vectork *cs2, const vectork *w) {
  for (int i = 0; i < K; i++) {
    scalar_make_hint(&out->v[i], &ct0->v[i], &cs2->v[i], &w->v[i]);
  }
}

static void vectork_use_hint_vartime(vectork *out, const vectork *h,
                                     const vectork *r) {
  for (int i = 0; i < K; i++) {
    scalar_use_hint_vartime(&out->v[i], &h->v[i], &r->v[i]);
  }
}

static uint32_t vectorl_max(const vectorl *a) {
  uint32_t max = 0;
  for (int i = 0; i < L; i++) {
    scalar_max(&max, &a->v[i]);
  }
  return max;
}

/* Bit packing */

// FIPS 204, Algorithm 16 (`SimpleBitPack`). Specialized to bitlen(b) = 4.
static void scalar_encode_4(uint8_t out[128], const scalar *s) {
  // Every two elements lands on a byte boundary.
  static_assert(DEGREE % 2 == 0, "DEGREE must be a multiple of 2");
  for (int i = 0; i < DEGREE / 2; i++) {
    uint32_t a = s->c[2 * i];
    uint32_t b = s->c[2 * i + 1];
    declassify_assert(a < 16);
    declassify_assert(b < 16);
    out[i] = a | (b << 4);
  }
}

// FIPS 204, Algorithm 16 (`SimpleBitPack`). Specialized to bitlen(b) = 10.
static void scalar_encode_10(uint8_t out[320], const scalar *s) {
  // Every four elements lands on a byte boundary.
  static_assert(DEGREE % 4 == 0, "DEGREE must be a multiple of 4");
  for (int i = 0; i < DEGREE / 4; i++) {
    uint32_t a = s->c[4 * i];
    uint32_t b = s->c[4 * i + 1];
    uint32_t c = s->c[4 * i + 2];
    uint32_t d = s->c[4 * i + 3];
    declassify_assert(a < 1024);
    declassify_assert(b < 1024);
    declassify_assert(c < 1024);
    declassify_assert(d < 1024);
    out[5 * i] = (uint8_t)a;
    out[5 * i + 1] = (uint8_t)((a >> 8) | (b << 2));
    out[5 * i + 2] = (uint8_t)((b >> 6) | (c << 4));
    out[5 * i + 3] = (uint8_t)((c >> 4) | (d << 6));
    out[5 * i + 4] = (uint8_t)(d >> 2);
  }
}

// FIPS 204, Algorithm 17 (`BitPack`). Specialized to bitlen(b) = 4 and b =
// 2^19.
static void scalar_encode_signed_4_eta(uint8_t out[128], const scalar *s) {
  // Every two elements lands on a byte boundary.
  static_assert(DEGREE % 2 == 0, "DEGREE must be a multiple of 2");
  for (int i = 0; i < DEGREE / 2; i++) {
    uint32_t a = mod_sub(ETA, s->c[2 * i]);
    uint32_t b = mod_sub(ETA, s->c[2 * i + 1]);
    declassify_assert(a < 16);
    declassify_assert(b < 16);
    out[i] = a | (b << 4);
  }
}

// FIPS 204, Algorithm 17 (`BitPack`). Specialized to bitlen(b) = 13 and b =
// 2^12.
static void scalar_encode_signed_13_12(uint8_t out[416], const scalar *s) {
  static const uint32_t kMax = 1u << 12;
  // Every two elements lands on a byte boundary.
  static_assert(DEGREE % 8 == 0, "DEGREE must be a multiple of 8");
  for (int i = 0; i < DEGREE / 8; i++) {
    uint32_t a = mod_sub(kMax, s->c[8 * i]);
    uint32_t b = mod_sub(kMax, s->c[8 * i + 1]);
    uint32_t c = mod_sub(kMax, s->c[8 * i + 2]);
    uint32_t d = mod_sub(kMax, s->c[8 * i + 3]);
    uint32_t e = mod_sub(kMax, s->c[8 * i + 4]);
    uint32_t f = mod_sub(kMax, s->c[8 * i + 5]);
    uint32_t g = mod_sub(kMax, s->c[8 * i + 6]);
    uint32_t h = mod_sub(kMax, s->c[8 * i + 7]);
    declassify_assert(a < (1u << 13));
    declassify_assert(b < (1u << 13));
    declassify_assert(c < (1u << 13));
    declassify_assert(d < (1u << 13));
    declassify_assert(e < (1u << 13));
    declassify_assert(f < (1u << 13));
    declassify_assert(g < (1u << 13));
    declassify_assert(h < (1u << 13));
    a |= b << 13;
    a |= c << 26;
    c >>= 6;
    c |= d << 7;
    c |= e << 20;
    e >>= 12;
    e |= f << 1;
    e |= g << 14;
    e |= h << 27;
    h >>= 5;
    OPENSSL_memcpy(&out[13 * i], &a, sizeof(a));
    OPENSSL_memcpy(&out[13 * i + 4], &c, sizeof(c));
    OPENSSL_memcpy(&out[13 * i + 8], &e, sizeof(e));
    OPENSSL_memcpy(&out[13 * i + 12], &h, 1);
  }
}

// FIPS 204, Algorithm 17 (`BitPack`). Specialized to bitlen(b) = 20 and b =
// 2^19.
static void scalar_encode_signed_20_19(uint8_t out[640], const scalar *s) {
  static const uint32_t kMax = 1u << 19;
  // Every two elements lands on a byte boundary.
  static_assert(DEGREE % 4 == 0, "DEGREE must be a multiple of 4");
  for (int i = 0; i < DEGREE / 4; i++) {
    uint32_t a = mod_sub(kMax, s->c[4 * i]);
    uint32_t b = mod_sub(kMax, s->c[4 * i + 1]);
    uint32_t c = mod_sub(kMax, s->c[4 * i + 2]);
    uint32_t d = mod_sub(kMax, s->c[4 * i + 3]);
    declassify_assert(a < (1u << 20));
    declassify_assert(b < (1u << 20));
    declassify_assert(c < (1u << 20));
    declassify_assert(d < (1u << 20));
    a |= b << 20;
    b >>= 12;
    b |= c << 8;
    b |= d << 28;
    d >>= 4;
    OPENSSL_memcpy(&out[10 * i], &a, sizeof(a));
    OPENSSL_memcpy(&out[10 * i + 4], &b, sizeof(b));
    OPENSSL_memcpy(&out[10 * i + 8], &d, 2);
  }
}

// FIPS 204, Algorithm 17 (`BitPack`).
static void scalar_encode_signed(uint8_t *out, const scalar *s, int bits,
                                 uint32_t max) {
  if (bits == 4) {
    assert(max == ETA);
    scalar_encode_signed_4_eta(out, s);
  } else if (bits == 20) {
    assert(max == 1u << 19);
    scalar_encode_signed_20_19(out, s);
  } else {
    assert(bits == 13);
    assert(max == 1u << 12);
    scalar_encode_signed_13_12(out, s);
  }
}

// FIPS 204, Algorithm 18 (`SimpleBitUnpack`). Specialized for bitlen(b) == 10.
static void scalar_decode_10(scalar *out, const uint8_t in[320]) {
  uint32_t v;
  static_assert(DEGREE % 4 == 0, "DEGREE must be a multiple of 4");
  for (int i = 0; i < DEGREE / 4; i++) {
    OPENSSL_memcpy(&v, &in[5 * i], sizeof(v));
    out->c[4 * i] = v & 0x3ff;
    out->c[4 * i + 1] = (v >> 10) & 0x3ff;
    out->c[4 * i + 2] = (v >> 20) & 0x3ff;
    out->c[4 * i + 3] = (v >> 30) | (((uint32_t)in[5 * i + 4]) << 2);
  }
}

// FIPS 204, Algorithm 19 (`BitUnpack`). Specialized to bitlen(a+b) = 4 and b =
// eta.
static int scalar_decode_signed_4_eta(scalar *out, const uint8_t in[128]) {
  uint32_t v;
  static_assert(DEGREE % 8 == 0, "DEGREE must be a multiple of 8");
  for (int i = 0; i < DEGREE / 8; i++) {
    OPENSSL_memcpy(&v, &in[4 * i], sizeof(v));
    static_assert(ETA == 4, "ETA must be 4");
    // None of the nibbles may be >= 9. So if the MSB of any nibble is set, none
    // of the other bits may be set. First, select all the MSBs.
    const uint32_t msbs = v & 0x88888888u;
    // For each nibble where the MSB is set, form a mask of all the other bits.
    const uint32_t mask = (msbs >> 1) | (msbs >> 2) | (msbs >> 3);
    // A nibble is only out of range in the case of invalid input, in which case
    // it is okay to leak the value.
    if (constant_time_declassify_int((mask & v) != 0)) {
      return 0;
    }

    out->c[i * 8] = mod_sub(ETA, v & 15);
    out->c[i * 8 + 1] = mod_sub(ETA, (v >> 4) & 15);
    out->c[i * 8 + 2] = mod_sub(ETA, (v >> 8) & 15);
    out->c[i * 8 + 3] = mod_sub(ETA, (v >> 12) & 15);
    out->c[i * 8 + 4] = mod_sub(ETA, (v >> 16) & 15);
    out->c[i * 8 + 5] = mod_sub(ETA, (v >> 20) & 15);
    out->c[i * 8 + 6] = mod_sub(ETA, (v >> 24) & 15);
    out->c[i * 8 + 7] = mod_sub(ETA, v >> 28);
  }
  return 1;
}

// FIPS 204, Algorithm 19 (`BitUnpack`). Specialized to bitlen(a+b) = 13 and b =
// 2^12.
static void scalar_decode_signed_13_12(scalar *out, const uint8_t in[416]) {
  static const uint32_t kMax = 1u << 12;
  static const uint32_t k13Bits = (1u << 13) - 1;
  static const uint32_t k7Bits = (1u << 7) - 1;

  uint32_t a, b, c;
  uint8_t d;
  static_assert(DEGREE % 8 == 0, "DEGREE must be a multiple of 8");
  for (int i = 0; i < DEGREE / 8; i++) {
    OPENSSL_memcpy(&a, &in[13 * i], sizeof(a));
    OPENSSL_memcpy(&b, &in[13 * i + 4], sizeof(b));
    OPENSSL_memcpy(&c, &in[13 * i + 8], sizeof(c));
    d = in[13 * i + 12];

    // It's not possible for a 13-bit number to be out of range when the max is
    // 2^12.
    out->c[i * 8] = mod_sub(kMax, a & k13Bits);
    out->c[i * 8 + 1] = mod_sub(kMax, (a >> 13) & k13Bits);
    out->c[i * 8 + 2] = mod_sub(kMax, (a >> 26) | ((b & k7Bits) << 6));
    out->c[i * 8 + 3] = mod_sub(kMax, (b >> 7) & k13Bits);
    out->c[i * 8 + 4] = mod_sub(kMax, (b >> 20) | ((c & 1) << 12));
    out->c[i * 8 + 5] = mod_sub(kMax, (c >> 1) & k13Bits);
    out->c[i * 8 + 6] = mod_sub(kMax, (c >> 14) & k13Bits);
    out->c[i * 8 + 7] = mod_sub(kMax, (c >> 27) | ((uint32_t)d) << 5);
  }
}

// FIPS 204, Algorithm 19 (`BitUnpack`). Specialized to bitlen(a+b) = 20 and b =
// 2^19.
static void scalar_decode_signed_20_19(scalar *out, const uint8_t in[640]) {
  static const uint32_t kMax = 1u << 19;
  static const uint32_t k20Bits = (1u << 20) - 1;

  uint32_t a, b;
  uint16_t c;
  static_assert(DEGREE % 4 == 0, "DEGREE must be a multiple of 4");
  for (int i = 0; i < DEGREE / 4; i++) {
    OPENSSL_memcpy(&a, &in[10 * i], sizeof(a));
    OPENSSL_memcpy(&b, &in[10 * i + 4], sizeof(b));
    OPENSSL_memcpy(&c, &in[10 * i + 8], sizeof(c));

    // It's not possible for a 20-bit number to be out of range when the max is
    // 2^19.
    out->c[i * 4] = mod_sub(kMax, a & k20Bits);
    out->c[i * 4 + 1] = mod_sub(kMax, (a >> 20) | ((b & 0xff) << 12));
    out->c[i * 4 + 2] = mod_sub(kMax, (b >> 8) & k20Bits);
    out->c[i * 4 + 3] = mod_sub(kMax, (b >> 28) | ((uint32_t)c) << 4);
  }
}

// FIPS 204, Algorithm 19 (`BitUnpack`).
static int scalar_decode_signed(scalar *out, const uint8_t *in, int bits,
                                uint32_t max) {
  if (bits == 4) {
    assert(max == ETA);
    return scalar_decode_signed_4_eta(out, in);
  } else if (bits == 13) {
    assert(max == (1u << 12));
    scalar_decode_signed_13_12(out, in);
    return 1;
  } else if (bits == 20) {
    assert(max == (1u << 19));
    scalar_decode_signed_20_19(out, in);
    return 1;
  } else {
    abort();
  }
}

/* Expansion functions */

// FIPS 204, Algorithm 30 (`RejNTTPoly`).
//
// Rejection samples a Keccak stream to get uniformly distributed elements. This
// is used for matrix expansion and only operates on public inputs.
static void scalar_from_keccak_vartime(
    scalar *out, const uint8_t derived_seed[RHO_BYTES + 2]) {
  struct BORINGSSL_keccak_st keccak_ctx;
  BORINGSSL_keccak_init(&keccak_ctx, boringssl_shake128);
  BORINGSSL_keccak_absorb(&keccak_ctx, derived_seed, RHO_BYTES + 2);
  assert(keccak_ctx.squeeze_offset == 0);
  assert(keccak_ctx.rate_bytes == 168);
  static_assert(168 % 3 == 0, "block and coefficient boundaries do not align");

  int done = 0;
  while (done < DEGREE) {
    uint8_t block[168];
    BORINGSSL_keccak_squeeze(&keccak_ctx, block, sizeof(block));
    for (size_t i = 0; i < sizeof(block) && done < DEGREE; i += 3) {
      // FIPS 204, Algorithm 14 (`CoeffFromThreeBytes`).
      uint32_t value = (uint32_t)block[i] | ((uint32_t)block[i + 1] << 8) |
                       (((uint32_t)block[i + 2] & 0x7f) << 16);
      if (value < kPrime) {
        out->c[done++] = value;
      }
    }
  }
}

// FIPS 204, Algorithm 31 (`RejBoundedPoly`).
static void scalar_uniform_eta_4(scalar *out,
                                 const uint8_t derived_seed[SIGMA_BYTES + 2]) {
  static_assert(ETA == 4, "This implementation is specialized for ETA == 4");

  struct BORINGSSL_keccak_st keccak_ctx;
  BORINGSSL_keccak_init(&keccak_ctx, boringssl_shake256);
  BORINGSSL_keccak_absorb(&keccak_ctx, derived_seed, SIGMA_BYTES + 2);
  assert(keccak_ctx.squeeze_offset == 0);
  assert(keccak_ctx.rate_bytes == 136);

  int done = 0;
  while (done < DEGREE) {
    uint8_t block[136];
    BORINGSSL_keccak_squeeze(&keccak_ctx, block, sizeof(block));
    for (size_t i = 0; i < sizeof(block) && done < DEGREE; ++i) {
      uint32_t t0 = block[i] & 0x0F;
      uint32_t t1 = block[i] >> 4;
      // FIPS 204, Algorithm 15 (`CoefFromHalfByte`). Although both the input
      // and output here are secret, it is OK to leak when we rejected a byte.
      // Individual bytes of the SHAKE-256 stream are (indistiguishable from)
      // independent of each other and the original seed, so leaking information
      // about the rejected bytes does not reveal the input or output.
      if (constant_time_declassify_int(t0 < 9)) {
        out->c[done++] = mod_sub(ETA, t0);
      }
      if (done < DEGREE && constant_time_declassify_int(t1 < 9)) {
        out->c[done++] = mod_sub(ETA, t1);
      }
    }
  }
}

// FIPS 204, Algorithm 34 (`ExpandMask`), but just a single step.
static void scalar_sample_mask(
    scalar *out, const uint8_t derived_seed[RHO_PRIME_BYTES + 2]) {
  uint8_t buf[640];
  BORINGSSL_keccak(buf, sizeof(buf), derived_seed, RHO_PRIME_BYTES + 2,
                   boringssl_shake256);

  scalar_decode_signed_20_19(out, buf);
}

// FIPS 204, Algorithm 29 (`SampleInBall`).
static void scalar_sample_in_ball_vartime(scalar *out, const uint8_t *seed,
                                          int len) {
  assert(len == 2 * LAMBDA_BYTES);

  struct BORINGSSL_keccak_st keccak_ctx;
  BORINGSSL_keccak_init(&keccak_ctx, boringssl_shake256);
  BORINGSSL_keccak_absorb(&keccak_ctx, seed, len);
  assert(keccak_ctx.squeeze_offset == 0);
  assert(keccak_ctx.rate_bytes == 136);

  uint8_t block[136];
  BORINGSSL_keccak_squeeze(&keccak_ctx, block, sizeof(block));

  uint64_t signs = CRYPTO_load_u64_le(block);
  int offset = 8;
  // SampleInBall implements a Fisher–Yates shuffle, which unavoidably leaks
  // where the zeros are by memory access pattern. Although this leak happens
  // before bad signatures are rejected, this is safe. See
  // https://boringssl-review.googlesource.com/c/boringssl/+/67747/comment/8d8f01ac_70af3f21/
  CONSTTIME_DECLASSIFY(block + offset, sizeof(block) - offset);

  OPENSSL_memset(out, 0, sizeof(*out));
  for (size_t i = DEGREE - TAU; i < DEGREE; i++) {
    size_t byte;
    for (;;) {
      if (offset == 136) {
        BORINGSSL_keccak_squeeze(&keccak_ctx, block, sizeof(block));
        // See above.
        CONSTTIME_DECLASSIFY(block, sizeof(block));
        offset = 0;
      }

      byte = block[offset++];
      if (byte <= i) {
        break;
      }
    }

    out->c[i] = out->c[byte];
    out->c[byte] = mod_sub(1, 2 * (signs & 1));
    signs >>= 1;
  }
}

// FIPS 204, Algorithm 32 (`ExpandA`).
static void matrix_expand(matrix *out, const uint8_t rho[RHO_BYTES]) {
  static_assert(K <= 0x100, "K must fit in 8 bits");
  static_assert(L <= 0x100, "L must fit in 8 bits");

  uint8_t derived_seed[RHO_BYTES + 2];
  OPENSSL_memcpy(derived_seed, rho, RHO_BYTES);
  for (int i = 0; i < K; i++) {
    for (int j = 0; j < L; j++) {
      derived_seed[RHO_BYTES + 1] = (uint8_t)i;
      derived_seed[RHO_BYTES] = (uint8_t)j;
      scalar_from_keccak_vartime(&out->v[i][j], derived_seed);
    }
  }
}

// FIPS 204, Algorithm 33 (`ExpandS`).
static void vector_expand_short(vectorl *s1, vectork *s2,
                                const uint8_t sigma[SIGMA_BYTES]) {
  static_assert(K <= 0x100, "K must fit in 8 bits");
  static_assert(L <= 0x100, "L must fit in 8 bits");
  static_assert(K + L <= 0x100, "K+L must fit in 8 bits");

  uint8_t derived_seed[SIGMA_BYTES + 2];
  OPENSSL_memcpy(derived_seed, sigma, SIGMA_BYTES);
  derived_seed[SIGMA_BYTES] = 0;
  derived_seed[SIGMA_BYTES + 1] = 0;
  for (int i = 0; i < L; i++) {
    scalar_uniform_eta_4(&s1->v[i], derived_seed);
    ++derived_seed[SIGMA_BYTES];
  }
  for (int i = 0; i < K; i++) {
    scalar_uniform_eta_4(&s2->v[i], derived_seed);
    ++derived_seed[SIGMA_BYTES];
  }
}

// FIPS 204, Algorithm 34 (`ExpandMask`).
static void vectorl_expand_mask(vectorl *out,
                                const uint8_t seed[RHO_PRIME_BYTES],
                                size_t kappa) {
  assert(kappa + L <= 0x10000);

  uint8_t derived_seed[RHO_PRIME_BYTES + 2];
  OPENSSL_memcpy(derived_seed, seed, RHO_PRIME_BYTES);
  for (int i = 0; i < L; i++) {
    size_t index = kappa + i;
    derived_seed[RHO_PRIME_BYTES] = index & 0xFF;
    derived_seed[RHO_PRIME_BYTES + 1] = (index >> 8) & 0xFF;
    scalar_sample_mask(&out->v[i], derived_seed);
  }
}

/* Encoding */

// FIPS 204, Algorithm 16 (`SimpleBitPack`).
//
// Encodes an entire vector into 32*K*|bits| bytes. Note that since 256 (DEGREE)
// is divisible by 8, the individual vector entries will always fill a whole
// number of bytes, so we do not need to worry about bit packing here.
static void vectork_encode(uint8_t *out, const vectork *a, int bits) {
  if (bits == 4) {
    for (int i = 0; i < K; i++) {
      scalar_encode_4(out + i * bits * DEGREE / 8, &a->v[i]);
    }
  } else {
    assert(bits == 10);
    for (int i = 0; i < K; i++) {
      scalar_encode_10(out + i * bits * DEGREE / 8, &a->v[i]);
    }
  }
}

// FIPS 204, Algorithm 18 (`SimpleBitUnpack`).
static void vectork_decode_10(vectork *out, const uint8_t *in) {
  for (int i = 0; i < K; i++) {
    scalar_decode_10(&out->v[i], in + i * 10 * DEGREE / 8);
  }
}

static void vectork_encode_signed(uint8_t *out, const vectork *a, int bits,
                                  uint32_t max) {
  for (int i = 0; i < K; i++) {
    scalar_encode_signed(out + i * bits * DEGREE / 8, &a->v[i], bits, max);
  }
}

static int vectork_decode_signed(vectork *out, const uint8_t *in, int bits,
                                 uint32_t max) {
  for (int i = 0; i < K; i++) {
    if (!scalar_decode_signed(&out->v[i], in + i * bits * DEGREE / 8, bits,
                              max)) {
      return 0;
    }
  }
  return 1;
}

// FIPS 204, Algorithm 17 (`BitPack`).
//
// Encodes an entire vector into 32*L*|bits| bytes. Note that since 256 (DEGREE)
// is divisible by 8, the individual vector entries will always fill a whole
// number of bytes, so we do not need to worry about bit packing here.
static void vectorl_encode_signed(uint8_t *out, const vectorl *a, int bits,
                                  uint32_t max) {
  for (int i = 0; i < L; i++) {
    scalar_encode_signed(out + i * bits * DEGREE / 8, &a->v[i], bits, max);
  }
}

static int vectorl_decode_signed(vectorl *out, const uint8_t *in, int bits,
                                 uint32_t max) {
  for (int i = 0; i < L; i++) {
    if (!scalar_decode_signed(&out->v[i], in + i * bits * DEGREE / 8, bits,
                              max)) {
      return 0;
    }
  }
  return 1;
}

// FIPS 204, Algorithm 28 (`w1Encode`).
static void w1_encode(uint8_t out[128 * K], const vectork *w1) {
  vectork_encode(out, w1, 4);
}

// FIPS 204, Algorithm 20 (`HintBitPack`).
static void hint_bit_pack(uint8_t out[OMEGA + K], const vectork *h) {
  OPENSSL_memset(out, 0, OMEGA + K);
  int index = 0;
  for (int i = 0; i < K; i++) {
    for (int j = 0; j < DEGREE; j++) {
      if (h->v[i].c[j]) {
        // h must have at most OMEGA non-zero coefficients.
        BSSL_CHECK(index < OMEGA);
        out[index++] = j;
      }
    }
    out[OMEGA + i] = index;
  }
}

// FIPS 204, Algorithm 21 (`HintBitUnpack`).
static int hint_bit_unpack(vectork *h, const uint8_t in[OMEGA + K]) {
  vectork_zero(h);
  int index = 0;
  for (int i = 0; i < K; i++) {
    const int limit = in[OMEGA + i];
    if (limit < index || limit > OMEGA) {
      return 0;
    }

    int last = -1;
    while (index < limit) {
      int byte = in[index++];
      if (last >= 0 && byte <= last) {
        return 0;
      }
      last = byte;
      static_assert(DEGREE == 256,
                    "DEGREE must be 256 for this write to be in bounds");
      h->v[i].c[byte] = 1;
    }
  }
  for (; index < OMEGA; index++) {
    if (in[index] != 0) {
      return 0;
    }
  }
  return 1;
}

struct public_key {
  uint8_t rho[RHO_BYTES];
  vectork t1;
  // Pre-cached value(s).
  uint8_t public_key_hash[TR_BYTES];
};

struct private_key {
  uint8_t rho[RHO_BYTES];
  uint8_t k[K_BYTES];
  uint8_t public_key_hash[TR_BYTES];
  vectorl s1;
  vectork s2;
  vectork t0;
};

struct signature {
  uint8_t c_tilde[2 * LAMBDA_BYTES];
  vectorl z;
  vectork h;
};

// FIPS 204, Algorithm 22 (`pkEncode`).
static int mldsa_marshal_public_key(CBB *out, const struct public_key *pub) {
  if (!CBB_add_bytes(out, pub->rho, sizeof(pub->rho))) {
    return 0;
  }

  uint8_t *vectork_output;
  if (!CBB_add_space(out, &vectork_output, 320 * K)) {
    return 0;
  }
  vectork_encode(vectork_output, &pub->t1, 10);

  return 1;
}

// FIPS 204, Algorithm 23 (`pkDecode`).
static int mldsa_parse_public_key(struct public_key *pub, CBS *in) {
  if (!CBS_copy_bytes(in, pub->rho, sizeof(pub->rho))) {
    return 0;
  }

  CBS t1_bytes;
  if (!CBS_get_bytes(in, &t1_bytes, 320 * K)) {
    return 0;
  }
  vectork_decode_10(&pub->t1, CBS_data(&t1_bytes));

  return 1;
}

// FIPS 204, Algorithm 24 (`skEncode`).
static int mldsa_marshal_private_key(CBB *out, const struct private_key *priv) {
  if (!CBB_add_bytes(out, priv->rho, sizeof(priv->rho)) ||
      !CBB_add_bytes(out, priv->k, sizeof(priv->k)) ||
      !CBB_add_bytes(out, priv->public_key_hash,
                     sizeof(priv->public_key_hash))) {
    return 0;
  }

  uint8_t *vectorl_output;
  if (!CBB_add_space(out, &vectorl_output, 128 * L)) {
    return 0;
  }
  vectorl_encode_signed(vectorl_output, &priv->s1, 4, ETA);

  uint8_t *vectork_output;
  if (!CBB_add_space(out, &vectork_output, 128 * K)) {
    return 0;
  }
  vectork_encode_signed(vectork_output, &priv->s2, 4, ETA);

  if (!CBB_add_space(out, &vectork_output, 416 * K)) {
    return 0;
  }
  vectork_encode_signed(vectork_output, &priv->t0, 13, 1 << 12);

  return 1;
}

// FIPS 204, Algorithm 25 (`skDecode`).
static int mldsa_parse_private_key(struct private_key *priv, CBS *in) {
  CBS s1_bytes;
  CBS s2_bytes;
  CBS t0_bytes;
  if (!CBS_copy_bytes(in, priv->rho, sizeof(priv->rho)) ||
      !CBS_copy_bytes(in, priv->k, sizeof(priv->k)) ||
      !CBS_copy_bytes(in, priv->public_key_hash,
                      sizeof(priv->public_key_hash)) ||
      !CBS_get_bytes(in, &s1_bytes, 128 * L) ||
      !vectorl_decode_signed(&priv->s1, CBS_data(&s1_bytes), 4, ETA) ||
      !CBS_get_bytes(in, &s2_bytes, 128 * K) ||
      !vectork_decode_signed(&priv->s2, CBS_data(&s2_bytes), 4, ETA) ||
      !CBS_get_bytes(in, &t0_bytes, 416 * K) ||
      // Note: Decoding 13 bits into (-2^12, 2^12] cannot fail.
      !vectork_decode_signed(&priv->t0, CBS_data(&t0_bytes), 13, 1 << 12)) {
    return 0;
  }

  return 1;
}

// FIPS 204, Algorithm 26 (`sigEncode`).
static int mldsa_marshal_signature(CBB *out, const struct signature *sign) {
  if (!CBB_add_bytes(out, sign->c_tilde, sizeof(sign->c_tilde))) {
    return 0;
  }

  uint8_t *vectorl_output;
  if (!CBB_add_space(out, &vectorl_output, 640 * L)) {
    return 0;
  }
  vectorl_encode_signed(vectorl_output, &sign->z, 20, 1 << 19);

  uint8_t *hint_output;
  if (!CBB_add_space(out, &hint_output, OMEGA + K)) {
    return 0;
  }
  hint_bit_pack(hint_output, &sign->h);

  return 1;
}

// FIPS 204, Algorithm 27 (`sigDecode`).
static int mldsa_parse_signature(struct signature *sign, CBS *in) {
  CBS z_bytes;
  CBS hint_bytes;
  if (!CBS_copy_bytes(in, sign->c_tilde, sizeof(sign->c_tilde)) ||
      !CBS_get_bytes(in, &z_bytes, 640 * L) ||
      // Note: Decoding 20 bits into (-2^19, 2^19] cannot fail.
      !vectorl_decode_signed(&sign->z, CBS_data(&z_bytes), 20, 1 << 19) ||
      !CBS_get_bytes(in, &hint_bytes, OMEGA + K) ||
      !hint_bit_unpack(&sign->h, CBS_data(&hint_bytes))) {
    return 0;
  };

  return 1;
}

static struct private_key *private_key_from_external(
    const struct MLDSA65_private_key *external) {
  static_assert(
      sizeof(struct MLDSA65_private_key) == sizeof(struct private_key),
      "Kyber private key size incorrect");
  static_assert(
      alignof(struct MLDSA65_private_key) == alignof(struct private_key),
      "Kyber private key align incorrect");
  return (struct private_key *)external;
}

static struct public_key *public_key_from_external(
    const struct MLDSA65_public_key *external) {
  static_assert(sizeof(struct MLDSA65_public_key) == sizeof(struct public_key),
                "mldsa public key size incorrect");
  static_assert(
      alignof(struct MLDSA65_public_key) == alignof(struct public_key),
      "mldsa public key align incorrect");
  return (struct public_key *)external;
}

/* API */

// Calls |MLDSA_generate_key_external_entropy| with random bytes from
// |RAND_bytes|. Returns 1 on success and 0 on failure.
int MLDSA65_generate_key(
    uint8_t out_encoded_public_key[MLDSA65_PUBLIC_KEY_BYTES],
    uint8_t out_seed[MLDSA_SEED_BYTES],
    struct MLDSA65_private_key *out_private_key) {
  RAND_bytes(out_seed, MLDSA_SEED_BYTES);
  return MLDSA65_generate_key_external_entropy(out_encoded_public_key,
                                               out_private_key, out_seed);
}

int MLDSA65_private_key_from_seed(struct MLDSA65_private_key *out_private_key,
                                  const uint8_t *seed, size_t seed_len) {
  if (seed_len != MLDSA_SEED_BYTES) {
    return 0;
  }
  uint8_t public_key[MLDSA65_PUBLIC_KEY_BYTES];
  return MLDSA65_generate_key_external_entropy(public_key, out_private_key,
                                               seed);
}

template <typename T>
struct DeleterFree {
  void operator()(T *ptr) { OPENSSL_free(ptr); }
};

// FIPS 204, Algorithm 6 (`ML-DSA.KeyGen_internal`). Returns 1 on success and 0
// on failure.
int MLDSA65_generate_key_external_entropy(
    uint8_t out_encoded_public_key[MLDSA65_PUBLIC_KEY_BYTES],
    struct MLDSA65_private_key *out_private_key,
    const uint8_t entropy[MLDSA_SEED_BYTES]) {
  // Intermediate values, allocated on the heap to allow use when there is a
  // limited amount of stack.
  struct values_st {
    struct public_key pub;
    matrix a_ntt;
    vectorl s1_ntt;
    vectork t;
  };
  std::unique_ptr<values_st, DeleterFree<values_st>> values(
      reinterpret_cast<struct values_st *>(OPENSSL_malloc(sizeof(values_st))));
  if (values == NULL) {
    return 0;
  }

  struct private_key *priv = private_key_from_external(out_private_key);

  uint8_t augmented_entropy[MLDSA_SEED_BYTES + 2];
  OPENSSL_memcpy(augmented_entropy, entropy, MLDSA_SEED_BYTES);
  // The k and l parameters are appended to the seed.
  augmented_entropy[MLDSA_SEED_BYTES] = K;
  augmented_entropy[MLDSA_SEED_BYTES + 1] = L;
  uint8_t expanded_seed[RHO_BYTES + SIGMA_BYTES + K_BYTES];
  BORINGSSL_keccak(expanded_seed, sizeof(expanded_seed), augmented_entropy,
                   sizeof(augmented_entropy), boringssl_shake256);
  const uint8_t *const rho = expanded_seed;
  const uint8_t *const sigma = expanded_seed + RHO_BYTES;
  const uint8_t *const k = expanded_seed + RHO_BYTES + SIGMA_BYTES;
  // rho is public.
  CONSTTIME_DECLASSIFY(rho, RHO_BYTES);
  OPENSSL_memcpy(values->pub.rho, rho, sizeof(values->pub.rho));
  OPENSSL_memcpy(priv->rho, rho, sizeof(priv->rho));
  OPENSSL_memcpy(priv->k, k, sizeof(priv->k));

  matrix_expand(&values->a_ntt, rho);
  vector_expand_short(&priv->s1, &priv->s2, sigma);

  OPENSSL_memcpy(&values->s1_ntt, &priv->s1, sizeof(values->s1_ntt));
  vectorl_ntt(&values->s1_ntt);

  matrix_mult(&values->t, &values->a_ntt, &values->s1_ntt);
  vectork_inverse_ntt(&values->t);
  vectork_add(&values->t, &values->t, &priv->s2);

  vectork_power2_round(&values->pub.t1, &priv->t0, &values->t);
  // t1 is public.
  CONSTTIME_DECLASSIFY(&values->pub.t1, sizeof(values->pub.t1));

  CBB cbb;
  CBB_init_fixed(&cbb, out_encoded_public_key, MLDSA65_PUBLIC_KEY_BYTES);
  if (!mldsa_marshal_public_key(&cbb, &values->pub)) {
    return 0;
  }
  assert(CBB_len(&cbb) == MLDSA65_PUBLIC_KEY_BYTES);

  BORINGSSL_keccak(priv->public_key_hash, sizeof(priv->public_key_hash),
                   out_encoded_public_key, MLDSA65_PUBLIC_KEY_BYTES,
                   boringssl_shake256);

  return 1;
}

int MLDSA65_public_from_private(struct MLDSA65_public_key *out_public_key,
                                const struct MLDSA65_private_key *private_key) {
  // Intermediate values, allocated on the heap to allow use when there is a
  // limited amount of stack.
  struct values_st {
    matrix a_ntt;
    vectorl s1_ntt;
    vectork t;
    vectork t0;
  };
  std::unique_ptr<values_st, DeleterFree<values_st>> values(
      reinterpret_cast<struct values_st *>(OPENSSL_malloc(sizeof(values_st))));
  if (values == NULL) {
    return 0;
  }

  const struct private_key *priv = private_key_from_external(private_key);
  struct public_key *pub = public_key_from_external(out_public_key);

  OPENSSL_memcpy(pub->rho, priv->rho, sizeof(pub->rho));
  OPENSSL_memcpy(pub->public_key_hash, priv->public_key_hash,
                 sizeof(pub->public_key_hash));

  matrix_expand(&values->a_ntt, priv->rho);

  OPENSSL_memcpy(&values->s1_ntt, &priv->s1, sizeof(values->s1_ntt));
  vectorl_ntt(&values->s1_ntt);

  matrix_mult(&values->t, &values->a_ntt, &values->s1_ntt);
  vectork_inverse_ntt(&values->t);
  vectork_add(&values->t, &values->t, &priv->s2);

  vectork_power2_round(&pub->t1, &values->t0, &values->t);
  return 1;
}

// FIPS 204, Algorithm 7 (`ML-DSA.Sign_internal`). Returns 1 on success and 0 on
// failure.
int MLDSA65_sign_internal(
    uint8_t out_encoded_signature[MLDSA65_SIGNATURE_BYTES],
    const struct MLDSA65_private_key *private_key, const uint8_t *msg,
    size_t msg_len, const uint8_t *context_prefix, size_t context_prefix_len,
    const uint8_t *context, size_t context_len,
    const uint8_t randomizer[MLDSA_SIGNATURE_RANDOMIZER_BYTES]) {
  const struct private_key *priv = private_key_from_external(private_key);

  uint8_t mu[MU_BYTES];
  struct BORINGSSL_keccak_st keccak_ctx;
  BORINGSSL_keccak_init(&keccak_ctx, boringssl_shake256);
  BORINGSSL_keccak_absorb(&keccak_ctx, priv->public_key_hash,
                          sizeof(priv->public_key_hash));
  BORINGSSL_keccak_absorb(&keccak_ctx, context_prefix, context_prefix_len);
  BORINGSSL_keccak_absorb(&keccak_ctx, context, context_len);
  BORINGSSL_keccak_absorb(&keccak_ctx, msg, msg_len);
  BORINGSSL_keccak_squeeze(&keccak_ctx, mu, MU_BYTES);

  uint8_t rho_prime[RHO_PRIME_BYTES];
  BORINGSSL_keccak_init(&keccak_ctx, boringssl_shake256);
  BORINGSSL_keccak_absorb(&keccak_ctx, priv->k, sizeof(priv->k));
  BORINGSSL_keccak_absorb(&keccak_ctx, randomizer,
                          MLDSA_SIGNATURE_RANDOMIZER_BYTES);
  BORINGSSL_keccak_absorb(&keccak_ctx, mu, MU_BYTES);
  BORINGSSL_keccak_squeeze(&keccak_ctx, rho_prime, RHO_PRIME_BYTES);

  // Intermediate values, allocated on the heap to allow use when there is a
  // limited amount of stack.
  struct values_st {
    struct signature sign;
    vectorl s1_ntt;
    vectork s2_ntt;
    vectork t0_ntt;
    matrix a_ntt;
    vectorl y;
    vectork w;
    vectork w1;
    vectorl cs1;
    vectork cs2;
  };
  std::unique_ptr<values_st, DeleterFree<values_st>> values(
      reinterpret_cast<struct values_st *>(OPENSSL_malloc(sizeof(values_st))));
  if (values == NULL) {
    return 0;
  }
  OPENSSL_memcpy(&values->s1_ntt, &priv->s1, sizeof(values->s1_ntt));
  vectorl_ntt(&values->s1_ntt);

  OPENSSL_memcpy(&values->s2_ntt, &priv->s2, sizeof(values->s2_ntt));
  vectork_ntt(&values->s2_ntt);

  OPENSSL_memcpy(&values->t0_ntt, &priv->t0, sizeof(values->t0_ntt));
  vectork_ntt(&values->t0_ntt);

  matrix_expand(&values->a_ntt, priv->rho);

  // kappa must not exceed 2**16/L = 13107. But the probability of it exceeding
  // even 1000 iterations is vanishingly small.
  for (size_t kappa = 0;; kappa += L) {
    vectorl_expand_mask(&values->y, rho_prime, kappa);

    vectorl *y_ntt = &values->cs1;
    OPENSSL_memcpy(y_ntt, &values->y, sizeof(*y_ntt));
    vectorl_ntt(y_ntt);

    matrix_mult(&values->w, &values->a_ntt, y_ntt);
    vectork_inverse_ntt(&values->w);

    vectork_high_bits(&values->w1, &values->w);
    uint8_t w1_encoded[128 * K];
    w1_encode(w1_encoded, &values->w1);

    BORINGSSL_keccak_init(&keccak_ctx, boringssl_shake256);
    BORINGSSL_keccak_absorb(&keccak_ctx, mu, MU_BYTES);
    BORINGSSL_keccak_absorb(&keccak_ctx, w1_encoded, 128 * K);
    BORINGSSL_keccak_squeeze(&keccak_ctx, values->sign.c_tilde,
                             2 * LAMBDA_BYTES);

    scalar c_ntt;
    scalar_sample_in_ball_vartime(&c_ntt, values->sign.c_tilde,
                                  sizeof(values->sign.c_tilde));
    scalar_ntt(&c_ntt);

    vectorl_mult_scalar(&values->cs1, &values->s1_ntt, &c_ntt);
    vectorl_inverse_ntt(&values->cs1);
    vectork_mult_scalar(&values->cs2, &values->s2_ntt, &c_ntt);
    vectork_inverse_ntt(&values->cs2);

    vectorl_add(&values->sign.z, &values->y, &values->cs1);

    vectork *r0 = &values->w1;
    vectork_sub(r0, &values->w, &values->cs2);
    vectork_low_bits(r0, r0);

    // Leaking the fact that a signature was rejected is fine as the next
    // attempt at a signature will be (indistinguishable from) independent of
    // this one. Note, however, that we additionally leak which of the two
    // branches rejected the signature. Section 5.5 of
    // https://pq-crystals.org/dilithium/data/dilithium-specification-round3.pdf
    // describes this leak as OK. Note we leak less than what is described by
    // the paper; we do not reveal which coefficient violated the bound, and we
    // hide which of the |z_max| or |r0_max| bound failed. See also
    // https://boringssl-review.googlesource.com/c/boringssl/+/67747/comment/2bbab0fa_d241d35a/
    uint32_t z_max = vectorl_max(&values->sign.z);
    uint32_t r0_max = vectork_max_signed(r0);
    if (constant_time_declassify_w(
            constant_time_ge_w(z_max, kGamma1 - BETA) |
            constant_time_ge_w(r0_max, kGamma2 - BETA))) {
      continue;
    }

    vectork *ct0 = &values->w1;
    vectork_mult_scalar(ct0, &values->t0_ntt, &c_ntt);
    vectork_inverse_ntt(ct0);
    vectork_make_hint(&values->sign.h, ct0, &values->cs2, &values->w);

    // See above.
    uint32_t ct0_max = vectork_max(ct0);
    size_t h_ones = vectork_count_ones(&values->sign.h);
    if (constant_time_declassify_w(constant_time_ge_w(ct0_max, kGamma2) |
                                   constant_time_lt_w(OMEGA, h_ones))) {
      continue;
    }

    // Although computed with the private key, the signature is public.
    CONSTTIME_DECLASSIFY(values->sign.c_tilde, sizeof(values->sign.c_tilde));
    CONSTTIME_DECLASSIFY(&values->sign.z, sizeof(values->sign.z));
    CONSTTIME_DECLASSIFY(&values->sign.h, sizeof(values->sign.h));

    CBB cbb;
    CBB_init_fixed(&cbb, out_encoded_signature, MLDSA65_SIGNATURE_BYTES);
    if (!mldsa_marshal_signature(&cbb, &values->sign)) {
      return 0;
    }

    BSSL_CHECK(CBB_len(&cbb) == MLDSA65_SIGNATURE_BYTES);
    return 1;
  }
}

// mldsa signature in randomized mode, filling the random bytes with
// |RAND_bytes|. Returns 1 on success and 0 on failure.
int MLDSA65_sign(uint8_t out_encoded_signature[MLDSA65_SIGNATURE_BYTES],
                 const struct MLDSA65_private_key *private_key,
                 const uint8_t *msg, size_t msg_len, const uint8_t *context,
                 size_t context_len) {
  if (context_len > 255) {
    return 0;
  }

  uint8_t randomizer[MLDSA_SIGNATURE_RANDOMIZER_BYTES];
  RAND_bytes(randomizer, sizeof(randomizer));

  const uint8_t context_prefix[2] = {0, static_cast<uint8_t>(context_len)};
  return MLDSA65_sign_internal(out_encoded_signature, private_key, msg, msg_len,
                               context_prefix, sizeof(context_prefix), context,
                               context_len, randomizer);
}

// FIPS 204, Algorithm 3 (`ML-DSA.Verify`).
int MLDSA65_verify(const struct MLDSA65_public_key *public_key,
                   const uint8_t *signature, size_t signature_len,
                   const uint8_t *msg, size_t msg_len, const uint8_t *context,
                   size_t context_len) {
  if (context_len > 255 || signature_len != MLDSA65_SIGNATURE_BYTES) {
    return 0;
  }

  const uint8_t context_prefix[2] = {0, static_cast<uint8_t>(context_len)};
  return MLDSA65_verify_internal(public_key, signature, msg, msg_len,
                                 context_prefix, sizeof(context_prefix),
                                 context, context_len);
}

// FIPS 204, Algorithm 8 (`ML-DSA.Verify_internal`).
int MLDSA65_verify_internal(
    const struct MLDSA65_public_key *public_key,
    const uint8_t encoded_signature[MLDSA65_SIGNATURE_BYTES],
    const uint8_t *msg, size_t msg_len, const uint8_t *context_prefix,
    size_t context_prefix_len, const uint8_t *context, size_t context_len) {
  // Intermediate values, allocated on the heap to allow use when there is a
  // limited amount of stack.
  struct values_st {
    struct signature sign;
    matrix a_ntt;
    vectorl z_ntt;
    vectork az_ntt;
    vectork ct1_ntt;
  };
  std::unique_ptr<values_st, DeleterFree<values_st>> values(
      reinterpret_cast<struct values_st *>(OPENSSL_malloc(sizeof(values_st))));
  if (values == NULL) {
    return 0;
  }

  const struct public_key *pub = public_key_from_external(public_key);

  CBS cbs;
  CBS_init(&cbs, encoded_signature, MLDSA65_SIGNATURE_BYTES);
  if (!mldsa_parse_signature(&values->sign, &cbs)) {
    return 0;
  }

  matrix_expand(&values->a_ntt, pub->rho);

  uint8_t mu[MU_BYTES];
  struct BORINGSSL_keccak_st keccak_ctx;
  BORINGSSL_keccak_init(&keccak_ctx, boringssl_shake256);
  BORINGSSL_keccak_absorb(&keccak_ctx, pub->public_key_hash,
                          sizeof(pub->public_key_hash));
  BORINGSSL_keccak_absorb(&keccak_ctx, context_prefix, context_prefix_len);
  BORINGSSL_keccak_absorb(&keccak_ctx, context, context_len);
  BORINGSSL_keccak_absorb(&keccak_ctx, msg, msg_len);
  BORINGSSL_keccak_squeeze(&keccak_ctx, mu, MU_BYTES);

  scalar c_ntt;
  scalar_sample_in_ball_vartime(&c_ntt, values->sign.c_tilde,
                                sizeof(values->sign.c_tilde));
  scalar_ntt(&c_ntt);

  OPENSSL_memcpy(&values->z_ntt, &values->sign.z, sizeof(values->z_ntt));
  vectorl_ntt(&values->z_ntt);

  matrix_mult(&values->az_ntt, &values->a_ntt, &values->z_ntt);

  vectork_scale_power2_round(&values->ct1_ntt, &pub->t1);
  vectork_ntt(&values->ct1_ntt);

  vectork_mult_scalar(&values->ct1_ntt, &values->ct1_ntt, &c_ntt);

  vectork *const w1 = &values->az_ntt;
  vectork_sub(w1, &values->az_ntt, &values->ct1_ntt);
  vectork_inverse_ntt(w1);

  vectork_use_hint_vartime(w1, &values->sign.h, w1);
  uint8_t w1_encoded[128 * K];
  w1_encode(w1_encoded, w1);

  uint8_t c_tilde[2 * LAMBDA_BYTES];
  BORINGSSL_keccak_init(&keccak_ctx, boringssl_shake256);
  BORINGSSL_keccak_absorb(&keccak_ctx, mu, MU_BYTES);
  BORINGSSL_keccak_absorb(&keccak_ctx, w1_encoded, 128 * K);
  BORINGSSL_keccak_squeeze(&keccak_ctx, c_tilde, 2 * LAMBDA_BYTES);

  uint32_t z_max = vectorl_max(&values->sign.z);
  return z_max < kGamma1 - BETA &&
         OPENSSL_memcmp(c_tilde, values->sign.c_tilde, 2 * LAMBDA_BYTES) == 0;
}

/* Serialization of keys. */

int MLDSA65_marshal_public_key(CBB *out,
                               const struct MLDSA65_public_key *public_key) {
  return mldsa_marshal_public_key(out, public_key_from_external(public_key));
}

int MLDSA65_parse_public_key(struct MLDSA65_public_key *public_key, CBS *in) {
  struct public_key *pub = public_key_from_external(public_key);
  CBS orig_in = *in;
  if (!mldsa_parse_public_key(pub, in) || CBS_len(in) != 0) {
    return 0;
  }

  // Compute pre-cached values.
  BORINGSSL_keccak(pub->public_key_hash, sizeof(pub->public_key_hash),
                   CBS_data(&orig_in), CBS_len(&orig_in), boringssl_shake256);
  return 1;
}

int MLDSA65_marshal_private_key(CBB *out,
                                const struct MLDSA65_private_key *private_key) {
  return mldsa_marshal_private_key(out, private_key_from_external(private_key));
}

int MLDSA65_parse_private_key(struct MLDSA65_private_key *private_key,
                              CBS *in) {
  struct private_key *priv = private_key_from_external(private_key);
  return mldsa_parse_private_key(priv, in) && CBS_len(in) == 0;
}