xref: /third_party/libwebsockets/plugins/ssh-base/crypto/fe25519.c

/* $OpenBSD: fe25519.c,v 1.3 2013/12/09 11:03:45 markus Exp $ */

/*
 * Public Domain, Authors: Daniel J. Bernstein, Niels Duif, Tanja Lange,
 * Peter Schwabe, Bo-Yin Yang.
 * Copied from supercop-20130419/crypto_sign/ed25519/ref/fe25519.c
 */

#include "libwebsockets.h"

#define WINDOWSIZE 1 /* Should be 1,2, or 4 */
#define WINDOWMASK ((1<<WINDOWSIZE)-1)

#include "fe25519.h"

static uint32_t fe_equal(uint32_t a,uint32_t b) /* 16-bit inputs */
{
  uint32_t x = a ^ b; /* 0: yes; 1..65535: no */
  x -= 1; /* 4294967295: yes; 0..65534: no */
  x >>= 31; /* 1: yes; 0: no */
  return x;
}

static uint32_t ge(uint32_t a,uint32_t b) /* 16-bit inputs */
{
  unsigned int x = a;
  x -= (unsigned int) b; /* 0..65535: yes; 4294901761..4294967295: no */
  x >>= 31; /* 0: yes; 1: no */
  x ^= 1; /* 1: yes; 0: no */
  return x;
}

static uint32_t times19(uint32_t a)
{
  return (a << 4) + (a << 1) + a;
}

static uint32_t times38(uint32_t a)
{
  return (a << 5) + (a << 2) + (a << 1);
}

static void fe_reduce_add_sub(fe25519 *r)
{
  uint32_t t;
  int i,rep;

  for(rep=0;rep<4;rep++)
  {
    t = r->v[31] >> 7;
    r->v[31] &= 127;
    t = times19(t);
    r->v[0] += t;
    for(i=0;i<31;i++)
    {
      t = r->v[i] >> 8;
      r->v[i+1] += t;
      r->v[i] &= 255;
    }
  }
}

static void reduce_mul(fe25519 *r)
{
  uint32_t t;
  int i,rep;

  for(rep=0;rep<2;rep++)
  {
    t = r->v[31] >> 7;
    r->v[31] &= 127;
    t = times19(t);
    r->v[0] += t;
    for(i=0;i<31;i++)
    {
      t = r->v[i] >> 8;
      r->v[i+1] += t;
      r->v[i] &= 255;
    }
  }
}

/* reduction modulo 2^255-19 */
void fe25519_freeze(fe25519 *r)
{
  int i;
  uint32_t m = fe_equal(r->v[31],127);

  for(i=30;i>0;i--)
    m &= fe_equal(r->v[i],255);
  m &= ge(r->v[0],237);

  m = (uint32_t)-(int32_t)m;

  r->v[31] -= m&127;
  for(i=30;i>0;i--)
    r->v[i] -= m&255;
  r->v[0] -= m&237;
}

void fe25519_unpack(fe25519 *r, const unsigned char x[32])
{
  int i;
  for(i=0;i<32;i++) r->v[i] = x[i];
  r->v[31] &= 127;
}

/* Assumes input x being reduced below 2^255 */
void fe25519_pack(unsigned char r[32], const fe25519 *x)
{
  int i;
  fe25519 y = *x;
  fe25519_freeze(&y);
  for(i=0;i<32;i++)
    r[i] = (unsigned char)y.v[i];
}

int fe25519_iszero(const fe25519 *x)
{
  int i;
  int r;
  fe25519 t = *x;
  fe25519_freeze(&t);
  r = (int)fe_equal(t.v[0],0);
  for(i=1;i<32;i++)
    r &= (int)fe_equal(t.v[i],0);
  return r;
}

int fe25519_iseq_vartime(const fe25519 *x, const fe25519 *y)
{
  int i;
  fe25519 t1 = *x;
  fe25519 t2 = *y;
  fe25519_freeze(&t1);
  fe25519_freeze(&t2);
  for(i=0;i<32;i++)
    if(t1.v[i] != t2.v[i]) return 0;
  return 1;
}

void fe25519_cmov(fe25519 *r, const fe25519 *x, unsigned char b)
{
  int i;
  uint32_t mask = b;
  mask = (uint32_t)-(int32_t)mask;
  for(i=0;i<32;i++) r->v[i] ^= mask & (x->v[i] ^ r->v[i]);
}

unsigned char fe25519_getparity(const fe25519 *x)
{
  fe25519 t = *x;
  fe25519_freeze(&t);
  return (unsigned char)(t.v[0] & 1);
}

void fe25519_setone(fe25519 *r)
{
  int i;
  r->v[0] = 1;
  for(i=1;i<32;i++) r->v[i]=0;
}

void fe25519_setzero(fe25519 *r)
{
  int i;
  for(i=0;i<32;i++) r->v[i]=0;
}

void fe25519_neg(fe25519 *r, const fe25519 *x)
{
  fe25519 t;
  int i;
  for(i=0;i<32;i++) t.v[i]=x->v[i];
  fe25519_setzero(r);
  fe25519_sub(r, r, &t);
}

void fe25519_add(fe25519 *r, const fe25519 *x, const fe25519 *y)
{
  int i;
  for(i=0;i<32;i++) r->v[i] = x->v[i] + y->v[i];
  fe_reduce_add_sub(r);
}

void fe25519_sub(fe25519 *r, const fe25519 *x, const fe25519 *y)
{
  int i;
  uint32_t t[32];
  t[0] = x->v[0] + 0x1da;
  t[31] = x->v[31] + 0xfe;
  for(i=1;i<31;i++) t[i] = x->v[i] + 0x1fe;
  for(i=0;i<32;i++) r->v[i] = t[i] - y->v[i];
  fe_reduce_add_sub(r);
}

void fe25519_mul(fe25519 *r, const fe25519 *x, const fe25519 *y)
{
  int i,j;
  uint32_t t[63];
  for(i=0;i<63;i++)t[i] = 0;

  for(i=0;i<32;i++)
    for(j=0;j<32;j++)
      t[i+j] += x->v[i] * y->v[j];

  for(i=32;i<63;i++)
    r->v[i-32] = t[i-32] + times38(t[i]);
  r->v[31] = t[31]; /* result now in r[0]...r[31] */

  reduce_mul(r);
}

void fe25519_square(fe25519 *r, const fe25519 *x)
{
  fe25519_mul(r, x, x);
}

void fe25519_invert(fe25519 *r, const fe25519 *x)
{
	fe25519 z2;
	fe25519 z9;
	fe25519 z11;
	fe25519 z2_5_0;
	fe25519 z2_10_0;
	fe25519 z2_20_0;
	fe25519 z2_50_0;
	fe25519 z2_100_0;
	fe25519 t0;
	fe25519 t1;
	int i;

	/* 2 */ fe25519_square(&z2,x);
	/* 4 */ fe25519_square(&t1,&z2);
	/* 8 */ fe25519_square(&t0,&t1);
	/* 9 */ fe25519_mul(&z9,&t0,x);
	/* 11 */ fe25519_mul(&z11,&z9,&z2);
	/* 22 */ fe25519_square(&t0,&z11);
	/* 2^5 - 2^0 = 31 */ fe25519_mul(&z2_5_0,&t0,&z9);

	/* 2^6 - 2^1 */ fe25519_square(&t0,&z2_5_0);
	/* 2^7 - 2^2 */ fe25519_square(&t1,&t0);
	/* 2^8 - 2^3 */ fe25519_square(&t0,&t1);
	/* 2^9 - 2^4 */ fe25519_square(&t1,&t0);
	/* 2^10 - 2^5 */ fe25519_square(&t0,&t1);
	/* 2^10 - 2^0 */ fe25519_mul(&z2_10_0,&t0,&z2_5_0);

	/* 2^11 - 2^1 */ fe25519_square(&t0,&z2_10_0);
	/* 2^12 - 2^2 */ fe25519_square(&t1,&t0);
	/* 2^20 - 2^10 */ for (i = 2;i < 10;i += 2) { fe25519_square(&t0,&t1); fe25519_square(&t1,&t0); }
	/* 2^20 - 2^0 */ fe25519_mul(&z2_20_0,&t1,&z2_10_0);

	/* 2^21 - 2^1 */ fe25519_square(&t0,&z2_20_0);
	/* 2^22 - 2^2 */ fe25519_square(&t1,&t0);
	/* 2^40 - 2^20 */ for (i = 2;i < 20;i += 2) { fe25519_square(&t0,&t1); fe25519_square(&t1,&t0); }
	/* 2^40 - 2^0 */ fe25519_mul(&t0,&t1,&z2_20_0);

	/* 2^41 - 2^1 */ fe25519_square(&t1,&t0);
	/* 2^42 - 2^2 */ fe25519_square(&t0,&t1);
	/* 2^50 - 2^10 */ for (i = 2;i < 10;i += 2) { fe25519_square(&t1,&t0); fe25519_square(&t0,&t1); }
	/* 2^50 - 2^0 */ fe25519_mul(&z2_50_0,&t0,&z2_10_0);

	/* 2^51 - 2^1 */ fe25519_square(&t0,&z2_50_0);
	/* 2^52 - 2^2 */ fe25519_square(&t1,&t0);
	/* 2^100 - 2^50 */ for (i = 2;i < 50;i += 2) { fe25519_square(&t0,&t1); fe25519_square(&t1,&t0); }
	/* 2^100 - 2^0 */ fe25519_mul(&z2_100_0,&t1,&z2_50_0);

	/* 2^101 - 2^1 */ fe25519_square(&t1,&z2_100_0);
	/* 2^102 - 2^2 */ fe25519_square(&t0,&t1);
	/* 2^200 - 2^100 */ for (i = 2;i < 100;i += 2) { fe25519_square(&t1,&t0); fe25519_square(&t0,&t1); }
	/* 2^200 - 2^0 */ fe25519_mul(&t1,&t0,&z2_100_0);

	/* 2^201 - 2^1 */ fe25519_square(&t0,&t1);
	/* 2^202 - 2^2 */ fe25519_square(&t1,&t0);
	/* 2^250 - 2^50 */ for (i = 2;i < 50;i += 2) { fe25519_square(&t0,&t1); fe25519_square(&t1,&t0); }
	/* 2^250 - 2^0 */ fe25519_mul(&t0,&t1,&z2_50_0);

	/* 2^251 - 2^1 */ fe25519_square(&t1,&t0);
	/* 2^252 - 2^2 */ fe25519_square(&t0,&t1);
	/* 2^253 - 2^3 */ fe25519_square(&t1,&t0);
	/* 2^254 - 2^4 */ fe25519_square(&t0,&t1);
	/* 2^255 - 2^5 */ fe25519_square(&t1,&t0);
	/* 2^255 - 21 */ fe25519_mul(r,&t1,&z11);
}

void fe25519_pow2523(fe25519 *r, const fe25519 *x)
{
	fe25519 z2;
	fe25519 z9;
	fe25519 z11;
	fe25519 z2_5_0;
	fe25519 z2_10_0;
	fe25519 z2_20_0;
	fe25519 z2_50_0;
	fe25519 z2_100_0;
	fe25519 t;
	int i;

	/* 2 */ fe25519_square(&z2,x);
	/* 4 */ fe25519_square(&t,&z2);
	/* 8 */ fe25519_square(&t,&t);
	/* 9 */ fe25519_mul(&z9,&t,x);
	/* 11 */ fe25519_mul(&z11,&z9,&z2);
	/* 22 */ fe25519_square(&t,&z11);
	/* 2^5 - 2^0 = 31 */ fe25519_mul(&z2_5_0,&t,&z9);

	/* 2^6 - 2^1 */ fe25519_square(&t,&z2_5_0);
	/* 2^10 - 2^5 */ for (i = 1;i < 5;i++) { fe25519_square(&t,&t); }
	/* 2^10 - 2^0 */ fe25519_mul(&z2_10_0,&t,&z2_5_0);

	/* 2^11 - 2^1 */ fe25519_square(&t,&z2_10_0);
	/* 2^20 - 2^10 */ for (i = 1;i < 10;i++) { fe25519_square(&t,&t); }
	/* 2^20 - 2^0 */ fe25519_mul(&z2_20_0,&t,&z2_10_0);

	/* 2^21 - 2^1 */ fe25519_square(&t,&z2_20_0);
	/* 2^40 - 2^20 */ for (i = 1;i < 20;i++) { fe25519_square(&t,&t); }
	/* 2^40 - 2^0 */ fe25519_mul(&t,&t,&z2_20_0);

	/* 2^41 - 2^1 */ fe25519_square(&t,&t);
	/* 2^50 - 2^10 */ for (i = 1;i < 10;i++) { fe25519_square(&t,&t); }
	/* 2^50 - 2^0 */ fe25519_mul(&z2_50_0,&t,&z2_10_0);

	/* 2^51 - 2^1 */ fe25519_square(&t,&z2_50_0);
	/* 2^100 - 2^50 */ for (i = 1;i < 50;i++) { fe25519_square(&t,&t); }
	/* 2^100 - 2^0 */ fe25519_mul(&z2_100_0,&t,&z2_50_0);

	/* 2^101 - 2^1 */ fe25519_square(&t,&z2_100_0);
	/* 2^200 - 2^100 */ for (i = 1;i < 100;i++) { fe25519_square(&t,&t); }
	/* 2^200 - 2^0 */ fe25519_mul(&t,&t,&z2_100_0);

	/* 2^201 - 2^1 */ fe25519_square(&t,&t);
	/* 2^250 - 2^50 */ for (i = 1;i < 50;i++) { fe25519_square(&t,&t); }
	/* 2^250 - 2^0 */ fe25519_mul(&t,&t,&z2_50_0);

	/* 2^251 - 2^1 */ fe25519_square(&t,&t);
	/* 2^252 - 2^2 */ fe25519_square(&t,&t);
	/* 2^252 - 3 */ fe25519_mul(r,&t,x);
}