1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
|
/*
* Karatsuba Multiplication/Squaring
* (C) 1999-2010 Jack Lloyd
*
* Distributed under the terms of the Botan license
*/
#include <botan/internal/mp_core.h>
#include <botan/internal/mp_asmi.h>
#include <botan/mem_ops.h>
namespace Botan {
namespace {
/*
* Karatsuba Multiplication Operation
*/
void karatsuba_mul(word z[], const word x[], const word y[], u32bit N,
word workspace[])
{
if(N < BOTAN_KARAT_MUL_THRESHOLD || N % 2)
{
if(N == 6)
return bigint_comba_mul6(z, x, y);
else if(N == 8)
return bigint_comba_mul8(z, x, y);
else if(N == 16)
return bigint_comba_mul16(z, x, y);
else
return bigint_simple_mul(z, x, N, y, N);
}
const u32bit N2 = N / 2;
const word* x0 = x;
const word* x1 = x + N2;
const word* y0 = y;
const word* y1 = y + N2;
word* z0 = z;
word* z1 = z + N;
const s32bit cmp0 = bigint_cmp(x0, N2, x1, N2);
const s32bit cmp1 = bigint_cmp(y1, N2, y0, N2);
clear_mem(workspace, 2*N);
if(cmp0 && cmp1)
{
if(cmp0 > 0)
bigint_sub3(z0, x0, N2, x1, N2);
else
bigint_sub3(z0, x1, N2, x0, N2);
if(cmp1 > 0)
bigint_sub3(z1, y1, N2, y0, N2);
else
bigint_sub3(z1, y0, N2, y1, N2);
karatsuba_mul(workspace, z0, z1, N2, workspace+N);
}
karatsuba_mul(z0, x0, y0, N2, workspace+N);
karatsuba_mul(z1, x1, y1, N2, workspace+N);
const u32bit blocks_of_8 = N - (N % 8);
word ws_carry = 0;
for(u32bit j = 0; j != blocks_of_8; j += 8)
ws_carry = word8_add3(workspace + N + j, z0 + j, z1 + j, ws_carry);
for(u32bit j = blocks_of_8; j != N; ++j)
workspace[N + j] = word_add(z0[j], z1[j], &ws_carry);
word z_carry = 0;
for(u32bit j = 0; j != blocks_of_8; j += 8)
z_carry = word8_add2(z + N2 + j, workspace + N + j, z_carry);
for(u32bit j = blocks_of_8; j != N; ++j)
z[N2 + j] = word_add(z[N2 + j], workspace[N + j], &z_carry);
z[N + N2] = word_add(z[N + N2], ws_carry, &z_carry);
if(z_carry)
for(u32bit j = 1; j != N2; ++j)
if(++z[N + N2 + j])
break;
if((cmp0 == cmp1) || (cmp0 == 0) || (cmp1 == 0))
bigint_add2(z + N2, 2*N-N2, workspace, N);
else
bigint_sub2(z + N2, 2*N-N2, workspace, N);
}
/*
* Karatsuba Squaring Operation
*/
void karatsuba_sqr(word z[], const word x[], u32bit N, word workspace[])
{
if(N < BOTAN_KARAT_SQR_THRESHOLD || N % 2)
{
if(N == 6)
return bigint_comba_sqr6(z, x);
else if(N == 8)
return bigint_comba_sqr8(z, x);
else if(N == 16)
return bigint_comba_sqr16(z, x);
else
return bigint_simple_sqr(z, x, N);
}
const u32bit N2 = N / 2;
const word* x0 = x;
const word* x1 = x + N2;
word* z0 = z;
word* z1 = z + N;
const s32bit cmp = bigint_cmp(x0, N2, x1, N2);
clear_mem(workspace, 2*N);
if(cmp)
{
if(cmp > 0)
bigint_sub3(z0, x0, N2, x1, N2);
else
bigint_sub3(z0, x1, N2, x0, N2);
karatsuba_sqr(workspace, z0, N2, workspace+N);
}
karatsuba_sqr(z0, x0, N2, workspace+N);
karatsuba_sqr(z1, x1, N2, workspace+N);
const u32bit blocks_of_8 = N - (N % 8);
word ws_carry = 0;
for(u32bit j = 0; j != blocks_of_8; j += 8)
ws_carry = word8_add3(workspace + N + j, z0 + j, z1 + j, ws_carry);
for(u32bit j = blocks_of_8; j != N; ++j)
workspace[N + j] = word_add(z0[j], z1[j], &ws_carry);
word z_carry = 0;
for(u32bit j = 0; j != blocks_of_8; j += 8)
z_carry = word8_add2(z + N2 + j, workspace + N + j, z_carry);
for(u32bit j = blocks_of_8; j != N; ++j)
z[N2 + j] = word_add(z[N2 + j], workspace[N + j], &z_carry);
z[N + N2] = word_add(z[N + N2], ws_carry, &z_carry);
if(z_carry)
for(u32bit j = 1; j != N2; ++j)
if(++z[N + N2 + j])
break;
/*
* This is only actually required if cmp is != 0, however
* if cmp==0 then workspace[0:N] == 0 and avoiding the jump
* hides a timing channel.
*/
bigint_sub2(z + N2, 2*N-N2, workspace, N);
}
/*
* Pick a good size for the Karatsuba multiply
*/
u32bit karatsuba_size(u32bit z_size,
u32bit x_size, u32bit x_sw,
u32bit y_size, u32bit y_sw)
{
if(x_sw > x_size || x_sw > y_size || y_sw > x_size || y_sw > y_size)
return 0;
if(((x_size == x_sw) && (x_size % 2)) ||
((y_size == y_sw) && (y_size % 2)))
return 0;
const u32bit start = (x_sw > y_sw) ? x_sw : y_sw;
const u32bit end = (x_size < y_size) ? x_size : y_size;
if(start == end)
{
if(start % 2)
return 0;
return start;
}
for(u32bit j = start; j <= end; ++j)
{
if(j % 2)
continue;
if(2*j > z_size)
return 0;
if(x_sw <= j && j <= x_size && y_sw <= j && j <= y_size)
{
if(j % 4 == 2 &&
(j+2) <= x_size && (j+2) <= y_size && 2*(j+2) <= z_size)
return j+2;
return j;
}
}
return 0;
}
/*
* Pick a good size for the Karatsuba squaring
*/
u32bit karatsuba_size(u32bit z_size, u32bit x_size, u32bit x_sw)
{
if(x_sw == x_size)
{
if(x_sw % 2)
return 0;
return x_sw;
}
for(u32bit j = x_sw; j <= x_size; ++j)
{
if(j % 2)
continue;
if(2*j > z_size)
return 0;
if(j % 4 == 2 && (j+2) <= x_size && 2*(j+2) <= z_size)
return j+2;
return j;
}
return 0;
}
}
/*
* Multiplication Algorithm Dispatcher
*/
void bigint_mul(word z[], u32bit z_size, word workspace[],
const word x[], u32bit x_size, u32bit x_sw,
const word y[], u32bit y_size, u32bit y_sw)
{
if(x_sw == 1)
{
bigint_linmul3(z, y, y_sw, x[0]);
}
else if(y_sw == 1)
{
bigint_linmul3(z, x, x_sw, y[0]);
}
else if(x_sw <= 4 && x_size >= 4 &&
y_sw <= 4 && y_size >= 4 && z_size >= 8)
{
bigint_comba_mul4(z, x, y);
}
else if(x_sw <= 6 && x_size >= 6 &&
y_sw <= 6 && y_size >= 6 && z_size >= 12)
{
bigint_comba_mul6(z, x, y);
}
else if(x_sw <= 8 && x_size >= 8 &&
y_sw <= 8 && y_size >= 8 && z_size >= 16)
{
bigint_comba_mul8(z, x, y);
}
else if(x_sw <= 16 && x_size >= 16 &&
y_sw <= 16 && y_size >= 16 && z_size >= 32)
{
bigint_comba_mul16(z, x, y);
}
else if(x_sw < BOTAN_KARAT_MUL_THRESHOLD ||
y_sw < BOTAN_KARAT_MUL_THRESHOLD ||
!workspace)
{
bigint_simple_mul(z, x, x_sw, y, y_sw);
}
else
{
const u32bit N = karatsuba_size(z_size, x_size, x_sw, y_size, y_sw);
if(N)
{
clear_mem(workspace, 2*N);
karatsuba_mul(z, x, y, N, workspace);
}
else
bigint_simple_mul(z, x, x_sw, y, y_sw);
}
}
/*
* Squaring Algorithm Dispatcher
*/
void bigint_sqr(word z[], u32bit z_size, word workspace[],
const word x[], u32bit x_size, u32bit x_sw)
{
if(x_sw == 1)
{
bigint_linmul3(z, x, x_sw, x[0]);
}
else if(x_sw <= 4 && x_size >= 4 && z_size >= 8)
{
bigint_comba_sqr4(z, x);
}
else if(x_sw <= 6 && x_size >= 6 && z_size >= 12)
{
bigint_comba_sqr6(z, x);
}
else if(x_sw <= 8 && x_size >= 8 && z_size >= 16)
{
bigint_comba_sqr8(z, x);
}
else if(x_sw <= 16 && x_size >= 16 && z_size >= 32)
{
bigint_comba_sqr16(z, x);
}
else if(x_size < BOTAN_KARAT_SQR_THRESHOLD || !workspace)
{
bigint_simple_sqr(z, x, x_sw);
}
else
{
const u32bit N = karatsuba_size(z_size, x_size, x_sw);
if(N)
{
clear_mem(workspace, 2*N);
karatsuba_sqr(z, x, N, workspace);
}
else
bigint_simple_sqr(z, x, x_sw);
}
}
}
|
