fe25519.c revision 262566
1/* $OpenBSD: fe25519.c,v 1.3 2013/12/09 11:03:45 markus Exp $ */
2
3/*
4 * Public Domain, Authors: Daniel J. Bernstein, Niels Duif, Tanja Lange,
5 * Peter Schwabe, Bo-Yin Yang.
6 * Copied from supercop-20130419/crypto_sign/ed25519/ref/fe25519.c
7 */
8
9#include "includes.h"
10
11#define WINDOWSIZE 1 /* Should be 1,2, or 4 */
12#define WINDOWMASK ((1<<WINDOWSIZE)-1)
13
14#include "fe25519.h"
15
16static crypto_uint32 equal(crypto_uint32 a,crypto_uint32 b) /* 16-bit inputs */
17{
18  crypto_uint32 x = a ^ b; /* 0: yes; 1..65535: no */
19  x -= 1; /* 4294967295: yes; 0..65534: no */
20  x >>= 31; /* 1: yes; 0: no */
21  return x;
22}
23
24static crypto_uint32 ge(crypto_uint32 a,crypto_uint32 b) /* 16-bit inputs */
25{
26  unsigned int x = a;
27  x -= (unsigned int) b; /* 0..65535: yes; 4294901761..4294967295: no */
28  x >>= 31; /* 0: yes; 1: no */
29  x ^= 1; /* 1: yes; 0: no */
30  return x;
31}
32
33static crypto_uint32 times19(crypto_uint32 a)
34{
35  return (a << 4) + (a << 1) + a;
36}
37
38static crypto_uint32 times38(crypto_uint32 a)
39{
40  return (a << 5) + (a << 2) + (a << 1);
41}
42
43static void reduce_add_sub(fe25519 *r)
44{
45  crypto_uint32 t;
46  int i,rep;
47
48  for(rep=0;rep<4;rep++)
49  {
50    t = r->v[31] >> 7;
51    r->v[31] &= 127;
52    t = times19(t);
53    r->v[0] += t;
54    for(i=0;i<31;i++)
55    {
56      t = r->v[i] >> 8;
57      r->v[i+1] += t;
58      r->v[i] &= 255;
59    }
60  }
61}
62
63static void reduce_mul(fe25519 *r)
64{
65  crypto_uint32 t;
66  int i,rep;
67
68  for(rep=0;rep<2;rep++)
69  {
70    t = r->v[31] >> 7;
71    r->v[31] &= 127;
72    t = times19(t);
73    r->v[0] += t;
74    for(i=0;i<31;i++)
75    {
76      t = r->v[i] >> 8;
77      r->v[i+1] += t;
78      r->v[i] &= 255;
79    }
80  }
81}
82
83/* reduction modulo 2^255-19 */
84void fe25519_freeze(fe25519 *r)
85{
86  int i;
87  crypto_uint32 m = equal(r->v[31],127);
88  for(i=30;i>0;i--)
89    m &= equal(r->v[i],255);
90  m &= ge(r->v[0],237);
91
92  m = -m;
93
94  r->v[31] -= m&127;
95  for(i=30;i>0;i--)
96    r->v[i] -= m&255;
97  r->v[0] -= m&237;
98}
99
100void fe25519_unpack(fe25519 *r, const unsigned char x[32])
101{
102  int i;
103  for(i=0;i<32;i++) r->v[i] = x[i];
104  r->v[31] &= 127;
105}
106
107/* Assumes input x being reduced below 2^255 */
108void fe25519_pack(unsigned char r[32], const fe25519 *x)
109{
110  int i;
111  fe25519 y = *x;
112  fe25519_freeze(&y);
113  for(i=0;i<32;i++)
114    r[i] = y.v[i];
115}
116
117int fe25519_iszero(const fe25519 *x)
118{
119  int i;
120  int r;
121  fe25519 t = *x;
122  fe25519_freeze(&t);
123  r = equal(t.v[0],0);
124  for(i=1;i<32;i++)
125    r &= equal(t.v[i],0);
126  return r;
127}
128
129int fe25519_iseq_vartime(const fe25519 *x, const fe25519 *y)
130{
131  int i;
132  fe25519 t1 = *x;
133  fe25519 t2 = *y;
134  fe25519_freeze(&t1);
135  fe25519_freeze(&t2);
136  for(i=0;i<32;i++)
137    if(t1.v[i] != t2.v[i]) return 0;
138  return 1;
139}
140
141void fe25519_cmov(fe25519 *r, const fe25519 *x, unsigned char b)
142{
143  int i;
144  crypto_uint32 mask = b;
145  mask = -mask;
146  for(i=0;i<32;i++) r->v[i] ^= mask & (x->v[i] ^ r->v[i]);
147}
148
149unsigned char fe25519_getparity(const fe25519 *x)
150{
151  fe25519 t = *x;
152  fe25519_freeze(&t);
153  return t.v[0] & 1;
154}
155
156void fe25519_setone(fe25519 *r)
157{
158  int i;
159  r->v[0] = 1;
160  for(i=1;i<32;i++) r->v[i]=0;
161}
162
163void fe25519_setzero(fe25519 *r)
164{
165  int i;
166  for(i=0;i<32;i++) r->v[i]=0;
167}
168
169void fe25519_neg(fe25519 *r, const fe25519 *x)
170{
171  fe25519 t;
172  int i;
173  for(i=0;i<32;i++) t.v[i]=x->v[i];
174  fe25519_setzero(r);
175  fe25519_sub(r, r, &t);
176}
177
178void fe25519_add(fe25519 *r, const fe25519 *x, const fe25519 *y)
179{
180  int i;
181  for(i=0;i<32;i++) r->v[i] = x->v[i] + y->v[i];
182  reduce_add_sub(r);
183}
184
185void fe25519_sub(fe25519 *r, const fe25519 *x, const fe25519 *y)
186{
187  int i;
188  crypto_uint32 t[32];
189  t[0] = x->v[0] + 0x1da;
190  t[31] = x->v[31] + 0xfe;
191  for(i=1;i<31;i++) t[i] = x->v[i] + 0x1fe;
192  for(i=0;i<32;i++) r->v[i] = t[i] - y->v[i];
193  reduce_add_sub(r);
194}
195
196void fe25519_mul(fe25519 *r, const fe25519 *x, const fe25519 *y)
197{
198  int i,j;
199  crypto_uint32 t[63];
200  for(i=0;i<63;i++)t[i] = 0;
201
202  for(i=0;i<32;i++)
203    for(j=0;j<32;j++)
204      t[i+j] += x->v[i] * y->v[j];
205
206  for(i=32;i<63;i++)
207    r->v[i-32] = t[i-32] + times38(t[i]);
208  r->v[31] = t[31]; /* result now in r[0]...r[31] */
209
210  reduce_mul(r);
211}
212
213void fe25519_square(fe25519 *r, const fe25519 *x)
214{
215  fe25519_mul(r, x, x);
216}
217
218void fe25519_invert(fe25519 *r, const fe25519 *x)
219{
220	fe25519 z2;
221	fe25519 z9;
222	fe25519 z11;
223	fe25519 z2_5_0;
224	fe25519 z2_10_0;
225	fe25519 z2_20_0;
226	fe25519 z2_50_0;
227	fe25519 z2_100_0;
228	fe25519 t0;
229	fe25519 t1;
230	int i;
231
232	/* 2 */ fe25519_square(&z2,x);
233	/* 4 */ fe25519_square(&t1,&z2);
234	/* 8 */ fe25519_square(&t0,&t1);
235	/* 9 */ fe25519_mul(&z9,&t0,x);
236	/* 11 */ fe25519_mul(&z11,&z9,&z2);
237	/* 22 */ fe25519_square(&t0,&z11);
238	/* 2^5 - 2^0 = 31 */ fe25519_mul(&z2_5_0,&t0,&z9);
239
240	/* 2^6 - 2^1 */ fe25519_square(&t0,&z2_5_0);
241	/* 2^7 - 2^2 */ fe25519_square(&t1,&t0);
242	/* 2^8 - 2^3 */ fe25519_square(&t0,&t1);
243	/* 2^9 - 2^4 */ fe25519_square(&t1,&t0);
244	/* 2^10 - 2^5 */ fe25519_square(&t0,&t1);
245	/* 2^10 - 2^0 */ fe25519_mul(&z2_10_0,&t0,&z2_5_0);
246
247	/* 2^11 - 2^1 */ fe25519_square(&t0,&z2_10_0);
248	/* 2^12 - 2^2 */ fe25519_square(&t1,&t0);
249	/* 2^20 - 2^10 */ for (i = 2;i < 10;i += 2) { fe25519_square(&t0,&t1); fe25519_square(&t1,&t0); }
250	/* 2^20 - 2^0 */ fe25519_mul(&z2_20_0,&t1,&z2_10_0);
251
252	/* 2^21 - 2^1 */ fe25519_square(&t0,&z2_20_0);
253	/* 2^22 - 2^2 */ fe25519_square(&t1,&t0);
254	/* 2^40 - 2^20 */ for (i = 2;i < 20;i += 2) { fe25519_square(&t0,&t1); fe25519_square(&t1,&t0); }
255	/* 2^40 - 2^0 */ fe25519_mul(&t0,&t1,&z2_20_0);
256
257	/* 2^41 - 2^1 */ fe25519_square(&t1,&t0);
258	/* 2^42 - 2^2 */ fe25519_square(&t0,&t1);
259	/* 2^50 - 2^10 */ for (i = 2;i < 10;i += 2) { fe25519_square(&t1,&t0); fe25519_square(&t0,&t1); }
260	/* 2^50 - 2^0 */ fe25519_mul(&z2_50_0,&t0,&z2_10_0);
261
262	/* 2^51 - 2^1 */ fe25519_square(&t0,&z2_50_0);
263	/* 2^52 - 2^2 */ fe25519_square(&t1,&t0);
264	/* 2^100 - 2^50 */ for (i = 2;i < 50;i += 2) { fe25519_square(&t0,&t1); fe25519_square(&t1,&t0); }
265	/* 2^100 - 2^0 */ fe25519_mul(&z2_100_0,&t1,&z2_50_0);
266
267	/* 2^101 - 2^1 */ fe25519_square(&t1,&z2_100_0);
268	/* 2^102 - 2^2 */ fe25519_square(&t0,&t1);
269	/* 2^200 - 2^100 */ for (i = 2;i < 100;i += 2) { fe25519_square(&t1,&t0); fe25519_square(&t0,&t1); }
270	/* 2^200 - 2^0 */ fe25519_mul(&t1,&t0,&z2_100_0);
271
272	/* 2^201 - 2^1 */ fe25519_square(&t0,&t1);
273	/* 2^202 - 2^2 */ fe25519_square(&t1,&t0);
274	/* 2^250 - 2^50 */ for (i = 2;i < 50;i += 2) { fe25519_square(&t0,&t1); fe25519_square(&t1,&t0); }
275	/* 2^250 - 2^0 */ fe25519_mul(&t0,&t1,&z2_50_0);
276
277	/* 2^251 - 2^1 */ fe25519_square(&t1,&t0);
278	/* 2^252 - 2^2 */ fe25519_square(&t0,&t1);
279	/* 2^253 - 2^3 */ fe25519_square(&t1,&t0);
280	/* 2^254 - 2^4 */ fe25519_square(&t0,&t1);
281	/* 2^255 - 2^5 */ fe25519_square(&t1,&t0);
282	/* 2^255 - 21 */ fe25519_mul(r,&t1,&z11);
283}
284
285void fe25519_pow2523(fe25519 *r, const fe25519 *x)
286{
287	fe25519 z2;
288	fe25519 z9;
289	fe25519 z11;
290	fe25519 z2_5_0;
291	fe25519 z2_10_0;
292	fe25519 z2_20_0;
293	fe25519 z2_50_0;
294	fe25519 z2_100_0;
295	fe25519 t;
296	int i;
297
298	/* 2 */ fe25519_square(&z2,x);
299	/* 4 */ fe25519_square(&t,&z2);
300	/* 8 */ fe25519_square(&t,&t);
301	/* 9 */ fe25519_mul(&z9,&t,x);
302	/* 11 */ fe25519_mul(&z11,&z9,&z2);
303	/* 22 */ fe25519_square(&t,&z11);
304	/* 2^5 - 2^0 = 31 */ fe25519_mul(&z2_5_0,&t,&z9);
305
306	/* 2^6 - 2^1 */ fe25519_square(&t,&z2_5_0);
307	/* 2^10 - 2^5 */ for (i = 1;i < 5;i++) { fe25519_square(&t,&t); }
308	/* 2^10 - 2^0 */ fe25519_mul(&z2_10_0,&t,&z2_5_0);
309
310	/* 2^11 - 2^1 */ fe25519_square(&t,&z2_10_0);
311	/* 2^20 - 2^10 */ for (i = 1;i < 10;i++) { fe25519_square(&t,&t); }
312	/* 2^20 - 2^0 */ fe25519_mul(&z2_20_0,&t,&z2_10_0);
313
314	/* 2^21 - 2^1 */ fe25519_square(&t,&z2_20_0);
315	/* 2^40 - 2^20 */ for (i = 1;i < 20;i++) { fe25519_square(&t,&t); }
316	/* 2^40 - 2^0 */ fe25519_mul(&t,&t,&z2_20_0);
317
318	/* 2^41 - 2^1 */ fe25519_square(&t,&t);
319	/* 2^50 - 2^10 */ for (i = 1;i < 10;i++) { fe25519_square(&t,&t); }
320	/* 2^50 - 2^0 */ fe25519_mul(&z2_50_0,&t,&z2_10_0);
321
322	/* 2^51 - 2^1 */ fe25519_square(&t,&z2_50_0);
323	/* 2^100 - 2^50 */ for (i = 1;i < 50;i++) { fe25519_square(&t,&t); }
324	/* 2^100 - 2^0 */ fe25519_mul(&z2_100_0,&t,&z2_50_0);
325
326	/* 2^101 - 2^1 */ fe25519_square(&t,&z2_100_0);
327	/* 2^200 - 2^100 */ for (i = 1;i < 100;i++) { fe25519_square(&t,&t); }
328	/* 2^200 - 2^0 */ fe25519_mul(&t,&t,&z2_100_0);
329
330	/* 2^201 - 2^1 */ fe25519_square(&t,&t);
331	/* 2^250 - 2^50 */ for (i = 1;i < 50;i++) { fe25519_square(&t,&t); }
332	/* 2^250 - 2^0 */ fe25519_mul(&t,&t,&z2_50_0);
333
334	/* 2^251 - 2^1 */ fe25519_square(&t,&t);
335	/* 2^252 - 2^2 */ fe25519_square(&t,&t);
336	/* 2^252 - 3 */ fe25519_mul(r,&t,x);
337}
338