aboutsummaryrefslogtreecommitdiffstats
path: root/src/lib/pubkey/ed25519/ed25519_fe.cpp
blob: 5aa515e9ba8171111cacc8b94fbc9c4a1e6057b7 (plain)
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
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
/*
* Ed25519 field element
* (C) 2017 Ribose Inc
*
* Based on the public domain code from SUPERCOP ref10 by
* Peter Schwabe, Daniel J. Bernstein, Niels Duif, Tanja Lange, Bo-Yin Yang
*
* Botan is released under the Simplified BSD License (see license.txt)
*/

#include <botan/internal/ed25519_fe.h>
#include <botan/internal/ed25519_internal.h>

namespace Botan {

//static
FE_25519 FE_25519::invert(const FE_25519& z)
   {
   fe t0;
   fe t1;
   fe t2;
   fe t3;

   fe_sq(t0, z);
   fe_sq_iter(t1, t0, 2);
   fe_mul(t1, z, t1);
   fe_mul(t0, t0, t1);
   fe_sq(t2, t0);
   fe_mul(t1, t1, t2);
   fe_sq_iter(t2, t1, 5);
   fe_mul(t1, t2, t1);
   fe_sq_iter(t2, t1, 10);
   fe_mul(t2, t2, t1);
   fe_sq_iter(t3, t2, 20);
   fe_mul(t2, t3, t2);
   fe_sq_iter(t2, t2, 10);
   fe_mul(t1, t2, t1);
   fe_sq_iter(t2, t1, 50);
   fe_mul(t2, t2, t1);
   fe_sq_iter(t3, t2, 100);
   fe_mul(t2, t3, t2);
   fe_sq_iter(t2, t2, 50);
   fe_mul(t1, t2, t1);
   fe_sq_iter(t1, t1, 5);

   fe_mul(t0, t1, t0);
   return t0;
   }

FE_25519 FE_25519::pow_22523(const fe& z)
   {
   fe t0;
   fe t1;
   fe t2;

   fe_sq(t0, z);
   fe_sq_iter(t1, t0, 2);
   fe_mul(t1, z, t1);
   fe_mul(t0, t0, t1);
   fe_sq(t0, t0);
   fe_mul(t0, t1, t0);
   fe_sq_iter(t1, t0, 5);
   fe_mul(t0, t1, t0);
   fe_sq_iter(t1, t0, 10);
   fe_mul(t1, t1, t0);
   fe_sq_iter(t2, t1, 20);
   fe_mul(t1, t2, t1);
   fe_sq_iter(t1, t1, 10);
   fe_mul(t0, t1, t0);
   fe_sq_iter(t1, t0, 50);
   fe_mul(t1, t1, t0);
   fe_sq_iter(t2, t1, 100);
   fe_mul(t1, t2, t1);
   fe_sq_iter(t1, t1, 50);
   fe_mul(t0, t1, t0);
   fe_sq_iter(t0, t0, 2);

   fe_mul(t0, t0, z);
   return t0;
   }

/*
h = f * g
Can overlap h with f or g.

Preconditions:
|f| bounded by 1.65*2^26,1.65*2^25,1.65*2^26,1.65*2^25,etc.
|g| bounded by 1.65*2^26,1.65*2^25,1.65*2^26,1.65*2^25,etc.

Postconditions:
|h| bounded by 1.01*2^25,1.01*2^24,1.01*2^25,1.01*2^24,etc.
*/

/*
Notes on implementation strategy:

Using schoolbook multiplication.
Karatsuba would save a little in some cost models.

Most multiplications by 2 and 19 are 32-bit precomputations;
cheaper than 64-bit postcomputations.

There is one remaining multiplication by 19 in the carry chain;
one *19 precomputation can be merged into this,
but the resulting data flow is considerably less clean.

There are 12 carries below.
10 of them are 2-way parallelizable and vectorizable.
Can get away with 11 carries, but then data flow is much deeper.

With tighter constraints on inputs can squeeze carries into int32.
*/

//static
FE_25519 FE_25519::mul(const FE_25519& f, const FE_25519& g)
   {
   int32_t f0 = f[0];
   int32_t f1 = f[1];
   int32_t f2 = f[2];
   int32_t f3 = f[3];
   int32_t f4 = f[4];
   int32_t f5 = f[5];
   int32_t f6 = f[6];
   int32_t f7 = f[7];
   int32_t f8 = f[8];
   int32_t f9 = f[9];

   int32_t g0 = g[0];
   int32_t g1 = g[1];
   int32_t g2 = g[2];
   int32_t g3 = g[3];
   int32_t g4 = g[4];
   int32_t g5 = g[5];
   int32_t g6 = g[6];
   int32_t g7 = g[7];
   int32_t g8 = g[8];
   int32_t g9 = g[9];

   int32_t g1_19 = 19 * g1; /* 1.959375*2^29 */
   int32_t g2_19 = 19 * g2; /* 1.959375*2^30; still ok */
   int32_t g3_19 = 19 * g3;
   int32_t g4_19 = 19 * g4;
   int32_t g5_19 = 19 * g5;
   int32_t g6_19 = 19 * g6;
   int32_t g7_19 = 19 * g7;
   int32_t g8_19 = 19 * g8;
   int32_t g9_19 = 19 * g9;
   int32_t f1_2 = 2 * f1;
   int32_t f3_2 = 2 * f3;
   int32_t f5_2 = 2 * f5;
   int32_t f7_2 = 2 * f7;
   int32_t f9_2 = 2 * f9;
   int64_t f0g0    = f0   * static_cast<int64_t>(g0);
   int64_t f0g1    = f0   * static_cast<int64_t>(g1);
   int64_t f0g2    = f0   * static_cast<int64_t>(g2);
   int64_t f0g3    = f0   * static_cast<int64_t>(g3);
   int64_t f0g4    = f0   * static_cast<int64_t>(g4);
   int64_t f0g5    = f0   * static_cast<int64_t>(g5);
   int64_t f0g6    = f0   * static_cast<int64_t>(g6);
   int64_t f0g7    = f0   * static_cast<int64_t>(g7);
   int64_t f0g8    = f0   * static_cast<int64_t>(g8);
   int64_t f0g9    = f0   * static_cast<int64_t>(g9);
   int64_t f1g0    = f1   * static_cast<int64_t>(g0);
   int64_t f1g1_2  = f1_2 * static_cast<int64_t>(g1);
   int64_t f1g2    = f1   * static_cast<int64_t>(g2);
   int64_t f1g3_2  = f1_2 * static_cast<int64_t>(g3);
   int64_t f1g4    = f1   * static_cast<int64_t>(g4);
   int64_t f1g5_2  = f1_2 * static_cast<int64_t>(g5);
   int64_t f1g6    = f1   * static_cast<int64_t>(g6);
   int64_t f1g7_2  = f1_2 * static_cast<int64_t>(g7);
   int64_t f1g8    = f1   * static_cast<int64_t>(g8);
   int64_t f1g9_38 = f1_2 * static_cast<int64_t>(g9_19);
   int64_t f2g0    = f2   * static_cast<int64_t>(g0);
   int64_t f2g1    = f2   * static_cast<int64_t>(g1);
   int64_t f2g2    = f2   * static_cast<int64_t>(g2);
   int64_t f2g3    = f2   * static_cast<int64_t>(g3);
   int64_t f2g4    = f2   * static_cast<int64_t>(g4);
   int64_t f2g5    = f2   * static_cast<int64_t>(g5);
   int64_t f2g6    = f2   * static_cast<int64_t>(g6);
   int64_t f2g7    = f2   * static_cast<int64_t>(g7);
   int64_t f2g8_19 = f2   * static_cast<int64_t>(g8_19);
   int64_t f2g9_19 = f2   * static_cast<int64_t>(g9_19);
   int64_t f3g0    = f3   * static_cast<int64_t>(g0);
   int64_t f3g1_2  = f3_2 * static_cast<int64_t>(g1);
   int64_t f3g2    = f3   * static_cast<int64_t>(g2);
   int64_t f3g3_2  = f3_2 * static_cast<int64_t>(g3);
   int64_t f3g4    = f3   * static_cast<int64_t>(g4);
   int64_t f3g5_2  = f3_2 * static_cast<int64_t>(g5);
   int64_t f3g6    = f3   * static_cast<int64_t>(g6);
   int64_t f3g7_38 = f3_2 * static_cast<int64_t>(g7_19);
   int64_t f3g8_19 = f3   * static_cast<int64_t>(g8_19);
   int64_t f3g9_38 = f3_2 * static_cast<int64_t>(g9_19);
   int64_t f4g0    = f4   * static_cast<int64_t>(g0);
   int64_t f4g1    = f4   * static_cast<int64_t>(g1);
   int64_t f4g2    = f4   * static_cast<int64_t>(g2);
   int64_t f4g3    = f4   * static_cast<int64_t>(g3);
   int64_t f4g4    = f4   * static_cast<int64_t>(g4);
   int64_t f4g5    = f4   * static_cast<int64_t>(g5);
   int64_t f4g6_19 = f4   * static_cast<int64_t>(g6_19);
   int64_t f4g7_19 = f4   * static_cast<int64_t>(g7_19);
   int64_t f4g8_19 = f4   * static_cast<int64_t>(g8_19);
   int64_t f4g9_19 = f4   * static_cast<int64_t>(g9_19);
   int64_t f5g0    = f5   * static_cast<int64_t>(g0);
   int64_t f5g1_2  = f5_2 * static_cast<int64_t>(g1);
   int64_t f5g2    = f5   * static_cast<int64_t>(g2);
   int64_t f5g3_2  = f5_2 * static_cast<int64_t>(g3);
   int64_t f5g4    = f5   * static_cast<int64_t>(g4);
   int64_t f5g5_38 = f5_2 * static_cast<int64_t>(g5_19);
   int64_t f5g6_19 = f5   * static_cast<int64_t>(g6_19);
   int64_t f5g7_38 = f5_2 * static_cast<int64_t>(g7_19);
   int64_t f5g8_19 = f5   * static_cast<int64_t>(g8_19);
   int64_t f5g9_38 = f5_2 * static_cast<int64_t>(g9_19);
   int64_t f6g0    = f6   * static_cast<int64_t>(g0);
   int64_t f6g1    = f6   * static_cast<int64_t>(g1);
   int64_t f6g2    = f6   * static_cast<int64_t>(g2);
   int64_t f6g3    = f6   * static_cast<int64_t>(g3);
   int64_t f6g4_19 = f6   * static_cast<int64_t>(g4_19);
   int64_t f6g5_19 = f6   * static_cast<int64_t>(g5_19);
   int64_t f6g6_19 = f6   * static_cast<int64_t>(g6_19);
   int64_t f6g7_19 = f6   * static_cast<int64_t>(g7_19);
   int64_t f6g8_19 = f6   * static_cast<int64_t>(g8_19);
   int64_t f6g9_19 = f6   * static_cast<int64_t>(g9_19);
   int64_t f7g0    = f7   * static_cast<int64_t>(g0);
   int64_t f7g1_2  = f7_2 * static_cast<int64_t>(g1);
   int64_t f7g2    = f7   * static_cast<int64_t>(g2);
   int64_t f7g3_38 = f7_2 * static_cast<int64_t>(g3_19);
   int64_t f7g4_19 = f7   * static_cast<int64_t>(g4_19);
   int64_t f7g5_38 = f7_2 * static_cast<int64_t>(g5_19);
   int64_t f7g6_19 = f7   * static_cast<int64_t>(g6_19);
   int64_t f7g7_38 = f7_2 * static_cast<int64_t>(g7_19);
   int64_t f7g8_19 = f7   * static_cast<int64_t>(g8_19);
   int64_t f7g9_38 = f7_2 * static_cast<int64_t>(g9_19);
   int64_t f8g0    = f8   * static_cast<int64_t>(g0);
   int64_t f8g1    = f8   * static_cast<int64_t>(g1);
   int64_t f8g2_19 = f8   * static_cast<int64_t>(g2_19);
   int64_t f8g3_19 = f8   * static_cast<int64_t>(g3_19);
   int64_t f8g4_19 = f8   * static_cast<int64_t>(g4_19);
   int64_t f8g5_19 = f8   * static_cast<int64_t>(g5_19);
   int64_t f8g6_19 = f8   * static_cast<int64_t>(g6_19);
   int64_t f8g7_19 = f8   * static_cast<int64_t>(g7_19);
   int64_t f8g8_19 = f8   * static_cast<int64_t>(g8_19);
   int64_t f8g9_19 = f8   * static_cast<int64_t>(g9_19);
   int64_t f9g0    = f9   * static_cast<int64_t>(g0);
   int64_t f9g1_38 = f9_2 * static_cast<int64_t>(g1_19);
   int64_t f9g2_19 = f9   * static_cast<int64_t>(g2_19);
   int64_t f9g3_38 = f9_2 * static_cast<int64_t>(g3_19);
   int64_t f9g4_19 = f9   * static_cast<int64_t>(g4_19);
   int64_t f9g5_38 = f9_2 * static_cast<int64_t>(g5_19);
   int64_t f9g6_19 = f9   * static_cast<int64_t>(g6_19);
   int64_t f9g7_38 = f9_2 * static_cast<int64_t>(g7_19);
   int64_t f9g8_19 = f9   * static_cast<int64_t>(g8_19);
   int64_t f9g9_38 = f9_2 * static_cast<int64_t>(g9_19);
   int64_t h0 = f0g0+f1g9_38+f2g8_19+f3g7_38+f4g6_19+f5g5_38+f6g4_19+f7g3_38+f8g2_19+f9g1_38;
   int64_t h1 = f0g1+f1g0   +f2g9_19+f3g8_19+f4g7_19+f5g6_19+f6g5_19+f7g4_19+f8g3_19+f9g2_19;
   int64_t h2 = f0g2+f1g1_2 +f2g0   +f3g9_38+f4g8_19+f5g7_38+f6g6_19+f7g5_38+f8g4_19+f9g3_38;
   int64_t h3 = f0g3+f1g2   +f2g1   +f3g0   +f4g9_19+f5g8_19+f6g7_19+f7g6_19+f8g5_19+f9g4_19;
   int64_t h4 = f0g4+f1g3_2 +f2g2   +f3g1_2 +f4g0   +f5g9_38+f6g8_19+f7g7_38+f8g6_19+f9g5_38;
   int64_t h5 = f0g5+f1g4   +f2g3   +f3g2   +f4g1   +f5g0   +f6g9_19+f7g8_19+f8g7_19+f9g6_19;
   int64_t h6 = f0g6+f1g5_2 +f2g4   +f3g3_2 +f4g2   +f5g1_2 +f6g0   +f7g9_38+f8g8_19+f9g7_38;
   int64_t h7 = f0g7+f1g6   +f2g5   +f3g4   +f4g3   +f5g2   +f6g1   +f7g0   +f8g9_19+f9g8_19;
   int64_t h8 = f0g8+f1g7_2 +f2g6   +f3g5_2 +f4g4   +f5g3_2 +f6g2   +f7g1_2 +f8g0   +f9g9_38;
   int64_t h9 = f0g9+f1g8   +f2g7   +f3g6   +f4g5   +f5g4   +f6g3   +f7g2   +f8g1   +f9g0   ;
   int64_t carry0;
   int64_t carry1;
   int64_t carry2;
   int64_t carry3;
   int64_t carry4;
   int64_t carry5;
   int64_t carry6;
   int64_t carry7;
   int64_t carry8;
   int64_t carry9;

   /*
   |h0| <= (1.65*1.65*2^52*(1+19+19+19+19)+1.65*1.65*2^50*(38+38+38+38+38))
   i.e. |h0| <= 1.4*2^60; narrower ranges for h2, h4, h6, h8
   |h1| <= (1.65*1.65*2^51*(1+1+19+19+19+19+19+19+19+19))
   i.e. |h1| <= 1.7*2^59; narrower ranges for h3, h5, h7, h9
   */
   const int64_t X24 = (1 << 24);
   const int64_t X25 = (1 << 25);
   const int64_t X26 = (1 << 26);

   carry0 = (h0 + X25) >> 26;
   h1 += carry0;
   h0 -= carry0 * X26;
   carry4 = (h4 + X25) >> 26;
   h5 += carry4;
   h4 -= carry4 * X26;
   /* |h0| <= 2^25 */
   /* |h4| <= 2^25 */
   /* |h1| <= 1.71*2^59 */
   /* |h5| <= 1.71*2^59 */

   carry1 = (h1 + X24) >> 25;
   h2 += carry1;
   h1 -= carry1 * X25;
   carry5 = (h5 + X24) >> 25;
   h6 += carry5;
   h5 -= carry5 * X25;
   /* |h1| <= 2^24; from now on fits into int32 */
   /* |h5| <= 2^24; from now on fits into int32 */
   /* |h2| <= 1.41*2^60 */
   /* |h6| <= 1.41*2^60 */

   carry2 = (h2 + X25) >> 26;
   h3 += carry2;
   h2 -= carry2 * X26;
   carry6 = (h6 + X25) >> 26;
   h7 += carry6;
   h6 -= carry6 * X26;
   /* |h2| <= 2^25; from now on fits into int32 unchanged */
   /* |h6| <= 2^25; from now on fits into int32 unchanged */
   /* |h3| <= 1.71*2^59 */
   /* |h7| <= 1.71*2^59 */

   carry3 = (h3 + X24) >> 25;
   h4 += carry3;
   h3 -= carry3 * X25;
   carry7 = (h7 + X24) >> 25;
   h8 += carry7;
   h7 -= carry7 * X25;
   /* |h3| <= 2^24; from now on fits into int32 unchanged */
   /* |h7| <= 2^24; from now on fits into int32 unchanged */
   /* |h4| <= 1.72*2^34 */
   /* |h8| <= 1.41*2^60 */

   carry4 = (h4 + X25) >> 26;
   h5 += carry4;
   h4 -= carry4 * X26;
   carry8 = (h8 + X25) >> 26;
   h9 += carry8;
   h8 -= carry8 * X26;
   /* |h4| <= 2^25; from now on fits into int32 unchanged */
   /* |h8| <= 2^25; from now on fits into int32 unchanged */
   /* |h5| <= 1.01*2^24 */
   /* |h9| <= 1.71*2^59 */

   carry9 = (h9 + X24) >> 25;
   h0 += carry9 * 19;
   h9 -= carry9 * X25;
   /* |h9| <= 2^24; from now on fits into int32 unchanged */
   /* |h0| <= 1.1*2^39 */

   carry0 = (h0 + X25) >> 26;
   h1 += carry0;
   h0 -= carry0 * X26;
   /* |h0| <= 2^25; from now on fits into int32 unchanged */
   /* |h1| <= 1.01*2^24 */

   return FE_25519(h0, h1, h2, h3, h4, h5, h6, h7, h8, h9);
   }

/*
h = f * f
Can overlap h with f.

Preconditions:
|f| bounded by 1.65*2^26,1.65*2^25,1.65*2^26,1.65*2^25,etc.

Postconditions:
|h| bounded by 1.01*2^25,1.01*2^24,1.01*2^25,1.01*2^24,etc.
*/

/*
See fe_mul.c for discussion of implementation strategy.
*/

//static
FE_25519 FE_25519::sqr_iter(const FE_25519& f, size_t iter)
   {
   const int64_t X24 = (1 << 24);
   const int64_t X25 = (1 << 25);
   const int64_t X26 = (1 << 26);

   int32_t f0 = f[0];
   int32_t f1 = f[1];
   int32_t f2 = f[2];
   int32_t f3 = f[3];
   int32_t f4 = f[4];
   int32_t f5 = f[5];
   int32_t f6 = f[6];
   int32_t f7 = f[7];
   int32_t f8 = f[8];
   int32_t f9 = f[9];

   for(size_t i = 0; i != iter; ++i)
      {
      const int32_t f0_2 = 2 * f0;
      const int32_t f1_2 = 2 * f1;
      const int32_t f2_2 = 2 * f2;
      const int32_t f3_2 = 2 * f3;
      const int32_t f4_2 = 2 * f4;
      const int32_t f5_2 = 2 * f5;
      const int32_t f6_2 = 2 * f6;
      const int32_t f7_2 = 2 * f7;
      const int32_t f5_38 = 38 * f5; /* 1.959375*2^30 */
      const int32_t f6_19 = 19 * f6; /* 1.959375*2^30 */
      const int32_t f7_38 = 38 * f7; /* 1.959375*2^30 */
      const int32_t f8_19 = 19 * f8; /* 1.959375*2^30 */
      const int32_t f9_38 = 38 * f9; /* 1.959375*2^30 */

      const int64_t f0f0    = f0   * static_cast<int64_t>(f0);
      const int64_t f0f1_2  = f0_2 * static_cast<int64_t>(f1);
      const int64_t f0f2_2  = f0_2 * static_cast<int64_t>(f2);
      const int64_t f0f3_2  = f0_2 * static_cast<int64_t>(f3);
      const int64_t f0f4_2  = f0_2 * static_cast<int64_t>(f4);
      const int64_t f0f5_2  = f0_2 * static_cast<int64_t>(f5);
      const int64_t f0f6_2  = f0_2 * static_cast<int64_t>(f6);
      const int64_t f0f7_2  = f0_2 * static_cast<int64_t>(f7);
      const int64_t f0f8_2  = f0_2 * static_cast<int64_t>(f8);
      const int64_t f0f9_2  = f0_2 * static_cast<int64_t>(f9);
      const int64_t f1f1_2  = f1_2 * static_cast<int64_t>(f1);
      const int64_t f1f2_2  = f1_2 * static_cast<int64_t>(f2);
      const int64_t f1f3_4  = f1_2 * static_cast<int64_t>(f3_2);
      const int64_t f1f4_2  = f1_2 * static_cast<int64_t>(f4);
      const int64_t f1f5_4  = f1_2 * static_cast<int64_t>(f5_2);
      const int64_t f1f6_2  = f1_2 * static_cast<int64_t>(f6);
      const int64_t f1f7_4  = f1_2 * static_cast<int64_t>(f7_2);
      const int64_t f1f8_2  = f1_2 * static_cast<int64_t>(f8);
      const int64_t f1f9_76 = f1_2 * static_cast<int64_t>(f9_38);
      const int64_t f2f2    = f2   * static_cast<int64_t>(f2);
      const int64_t f2f3_2  = f2_2 * static_cast<int64_t>(f3);
      const int64_t f2f4_2  = f2_2 * static_cast<int64_t>(f4);
      const int64_t f2f5_2  = f2_2 * static_cast<int64_t>(f5);
      const int64_t f2f6_2  = f2_2 * static_cast<int64_t>(f6);
      const int64_t f2f7_2  = f2_2 * static_cast<int64_t>(f7);
      const int64_t f2f8_38 = f2_2 * static_cast<int64_t>(f8_19);
      const int64_t f2f9_38 = f2   * static_cast<int64_t>(f9_38);
      const int64_t f3f3_2  = f3_2 * static_cast<int64_t>(f3);
      const int64_t f3f4_2  = f3_2 * static_cast<int64_t>(f4);
      const int64_t f3f5_4  = f3_2 * static_cast<int64_t>(f5_2);
      const int64_t f3f6_2  = f3_2 * static_cast<int64_t>(f6);
      const int64_t f3f7_76 = f3_2 * static_cast<int64_t>(f7_38);
      const int64_t f3f8_38 = f3_2 * static_cast<int64_t>(f8_19);
      const int64_t f3f9_76 = f3_2 * static_cast<int64_t>(f9_38);
      const int64_t f4f4    = f4   * static_cast<int64_t>(f4);
      const int64_t f4f5_2  = f4_2 * static_cast<int64_t>(f5);
      const int64_t f4f6_38 = f4_2 * static_cast<int64_t>(f6_19);
      const int64_t f4f7_38 = f4   * static_cast<int64_t>(f7_38);
      const int64_t f4f8_38 = f4_2 * static_cast<int64_t>(f8_19);
      const int64_t f4f9_38 = f4   * static_cast<int64_t>(f9_38);
      const int64_t f5f5_38 = f5   * static_cast<int64_t>(f5_38);
      const int64_t f5f6_38 = f5_2 * static_cast<int64_t>(f6_19);
      const int64_t f5f7_76 = f5_2 * static_cast<int64_t>(f7_38);
      const int64_t f5f8_38 = f5_2 * static_cast<int64_t>(f8_19);
      const int64_t f5f9_76 = f5_2 * static_cast<int64_t>(f9_38);
      const int64_t f6f6_19 = f6   * static_cast<int64_t>(f6_19);
      const int64_t f6f7_38 = f6   * static_cast<int64_t>(f7_38);
      const int64_t f6f8_38 = f6_2 * static_cast<int64_t>(f8_19);
      const int64_t f6f9_38 = f6   * static_cast<int64_t>(f9_38);
      const int64_t f7f7_38 = f7   * static_cast<int64_t>(f7_38);
      const int64_t f7f8_38 = f7_2 * static_cast<int64_t>(f8_19);
      const int64_t f7f9_76 = f7_2 * static_cast<int64_t>(f9_38);
      const int64_t f8f8_19 = f8   * static_cast<int64_t>(f8_19);
      const int64_t f8f9_38 = f8   * static_cast<int64_t>(f9_38);
      const int64_t f9f9_38 = f9   * static_cast<int64_t>(f9_38);

      int64_t h0 = f0f0  +f1f9_76+f2f8_38+f3f7_76+f4f6_38+f5f5_38;
      int64_t h1 = f0f1_2+f2f9_38+f3f8_38+f4f7_38+f5f6_38;
      int64_t h2 = f0f2_2+f1f1_2 +f3f9_76+f4f8_38+f5f7_76+f6f6_19;
      int64_t h3 = f0f3_2+f1f2_2 +f4f9_38+f5f8_38+f6f7_38;
      int64_t h4 = f0f4_2+f1f3_4 +f2f2   +f5f9_76+f6f8_38+f7f7_38;
      int64_t h5 = f0f5_2+f1f4_2 +f2f3_2 +f6f9_38+f7f8_38;
      int64_t h6 = f0f6_2+f1f5_4 +f2f4_2 +f3f3_2 +f7f9_76+f8f8_19;
      int64_t h7 = f0f7_2+f1f6_2 +f2f5_2 +f3f4_2 +f8f9_38;
      int64_t h8 = f0f8_2+f1f7_4 +f2f6_2 +f3f5_4 +f4f4   +f9f9_38;
      int64_t h9 = f0f9_2+f1f8_2 +f2f7_2 +f3f6_2 +f4f5_2;

      int64_t carry0;
      int64_t carry1;
      int64_t carry2;
      int64_t carry3;
      int64_t carry4;
      int64_t carry5;
      int64_t carry6;
      int64_t carry7;
      int64_t carry8;
      int64_t carry9;

      carry0 = (h0 + X25) >> 26;
      h1 += carry0;
      h0 -= carry0 * X26;
      carry4 = (h4 + X25) >> 26;
      h5 += carry4;
      h4 -= carry4 * X26;

      carry1 = (h1 + X24) >> 25;
      h2 += carry1;
      h1 -= carry1 * X25;
      carry5 = (h5 + X24) >> 25;
      h6 += carry5;
      h5 -= carry5 * X25;

      carry2 = (h2 + X25) >> 26;
      h3 += carry2;
      h2 -= carry2 * X26;
      carry6 = (h6 + X25) >> 26;
      h7 += carry6;
      h6 -= carry6 * X26;

      carry3 = (h3 + X24) >> 25;
      h4 += carry3;
      h3 -= carry3 * X25;
      carry7 = (h7 + X24) >> 25;
      h8 += carry7;
      h7 -= carry7 * X25;

      carry4 = (h4 + X25) >> 26;
      h5 += carry4;
      h4 -= carry4 * X26;
      carry8 = (h8 + X25) >> 26;
      h9 += carry8;
      h8 -= carry8 * X26;

      carry9 = (h9 + X24) >> 25;
      h0 += carry9 * 19;
      h9 -= carry9 * X25;
      carry0 = (h0 + X25) >> 26;
      h1 += carry0;
      h0 -= carry0 * X26;

      f0 = h0;
      f1 = h1;
      f2 = h2;
      f3 = h3;
      f4 = h4;
      f5 = h5;
      f6 = h6;
      f7 = h7;
      f8 = h8;
      f9 = h9;
      }

   return FE_25519(f0, f1, f2, f3, f4, f5, f6, f7, f8, f9);
   }

/*
h = 2 * f * f
Can overlap h with f.

Preconditions:
|f| bounded by 1.65*2^26,1.65*2^25,1.65*2^26,1.65*2^25,etc.

Postconditions:
|h| bounded by 1.01*2^25,1.01*2^24,1.01*2^25,1.01*2^24,etc.
*/

/*
See fe_mul.c for discussion of implementation strategy.
*/

//static
FE_25519 FE_25519::sqr2(const FE_25519& f)
   {
   const int64_t X24 = (1 << 24);
   const int64_t X25 = (1 << 25);
   const int64_t X26 = (1 << 26);

   int32_t f0 = f[0];
   int32_t f1 = f[1];
   int32_t f2 = f[2];
   int32_t f3 = f[3];
   int32_t f4 = f[4];
   int32_t f5 = f[5];
   int32_t f6 = f[6];
   int32_t f7 = f[7];
   int32_t f8 = f[8];
   int32_t f9 = f[9];
   int32_t f0_2 = 2 * f0;
   int32_t f1_2 = 2 * f1;
   int32_t f2_2 = 2 * f2;
   int32_t f3_2 = 2 * f3;
   int32_t f4_2 = 2 * f4;
   int32_t f5_2 = 2 * f5;
   int32_t f6_2 = 2 * f6;
   int32_t f7_2 = 2 * f7;
   int32_t f5_38 = 38 * f5; /* 1.959375*2^30 */
   int32_t f6_19 = 19 * f6; /* 1.959375*2^30 */
   int32_t f7_38 = 38 * f7; /* 1.959375*2^30 */
   int32_t f8_19 = 19 * f8; /* 1.959375*2^30 */
   int32_t f9_38 = 38 * f9; /* 1.959375*2^30 */
   int64_t f0f0    = f0   * static_cast<int64_t>(f0);
   int64_t f0f1_2  = f0_2 * static_cast<int64_t>(f1);
   int64_t f0f2_2  = f0_2 * static_cast<int64_t>(f2);
   int64_t f0f3_2  = f0_2 * static_cast<int64_t>(f3);
   int64_t f0f4_2  = f0_2 * static_cast<int64_t>(f4);
   int64_t f0f5_2  = f0_2 * static_cast<int64_t>(f5);
   int64_t f0f6_2  = f0_2 * static_cast<int64_t>(f6);
   int64_t f0f7_2  = f0_2 * static_cast<int64_t>(f7);
   int64_t f0f8_2  = f0_2 * static_cast<int64_t>(f8);
   int64_t f0f9_2  = f0_2 * static_cast<int64_t>(f9);
   int64_t f1f1_2  = f1_2 * static_cast<int64_t>(f1);
   int64_t f1f2_2  = f1_2 * static_cast<int64_t>(f2);
   int64_t f1f3_4  = f1_2 * static_cast<int64_t>(f3_2);
   int64_t f1f4_2  = f1_2 * static_cast<int64_t>(f4);
   int64_t f1f5_4  = f1_2 * static_cast<int64_t>(f5_2);
   int64_t f1f6_2  = f1_2 * static_cast<int64_t>(f6);
   int64_t f1f7_4  = f1_2 * static_cast<int64_t>(f7_2);
   int64_t f1f8_2  = f1_2 * static_cast<int64_t>(f8);
   int64_t f1f9_76 = f1_2 * static_cast<int64_t>(f9_38);
   int64_t f2f2    = f2   * static_cast<int64_t>(f2);
   int64_t f2f3_2  = f2_2 * static_cast<int64_t>(f3);
   int64_t f2f4_2  = f2_2 * static_cast<int64_t>(f4);
   int64_t f2f5_2  = f2_2 * static_cast<int64_t>(f5);
   int64_t f2f6_2  = f2_2 * static_cast<int64_t>(f6);
   int64_t f2f7_2  = f2_2 * static_cast<int64_t>(f7);
   int64_t f2f8_38 = f2_2 * static_cast<int64_t>(f8_19);
   int64_t f2f9_38 = f2   * static_cast<int64_t>(f9_38);
   int64_t f3f3_2  = f3_2 * static_cast<int64_t>(f3);
   int64_t f3f4_2  = f3_2 * static_cast<int64_t>(f4);
   int64_t f3f5_4  = f3_2 * static_cast<int64_t>(f5_2);
   int64_t f3f6_2  = f3_2 * static_cast<int64_t>(f6);
   int64_t f3f7_76 = f3_2 * static_cast<int64_t>(f7_38);
   int64_t f3f8_38 = f3_2 * static_cast<int64_t>(f8_19);
   int64_t f3f9_76 = f3_2 * static_cast<int64_t>(f9_38);
   int64_t f4f4    = f4   * static_cast<int64_t>(f4);
   int64_t f4f5_2  = f4_2 * static_cast<int64_t>(f5);
   int64_t f4f6_38 = f4_2 * static_cast<int64_t>(f6_19);
   int64_t f4f7_38 = f4   * static_cast<int64_t>(f7_38);
   int64_t f4f8_38 = f4_2 * static_cast<int64_t>(f8_19);
   int64_t f4f9_38 = f4   * static_cast<int64_t>(f9_38);
   int64_t f5f5_38 = f5   * static_cast<int64_t>(f5_38);
   int64_t f5f6_38 = f5_2 * static_cast<int64_t>(f6_19);
   int64_t f5f7_76 = f5_2 * static_cast<int64_t>(f7_38);
   int64_t f5f8_38 = f5_2 * static_cast<int64_t>(f8_19);
   int64_t f5f9_76 = f5_2 * static_cast<int64_t>(f9_38);
   int64_t f6f6_19 = f6   * static_cast<int64_t>(f6_19);
   int64_t f6f7_38 = f6   * static_cast<int64_t>(f7_38);
   int64_t f6f8_38 = f6_2 * static_cast<int64_t>(f8_19);
   int64_t f6f9_38 = f6   * static_cast<int64_t>(f9_38);
   int64_t f7f7_38 = f7   * static_cast<int64_t>(f7_38);
   int64_t f7f8_38 = f7_2 * static_cast<int64_t>(f8_19);
   int64_t f7f9_76 = f7_2 * static_cast<int64_t>(f9_38);
   int64_t f8f8_19 = f8   * static_cast<int64_t>(f8_19);
   int64_t f8f9_38 = f8   * static_cast<int64_t>(f9_38);
   int64_t f9f9_38 = f9   * static_cast<int64_t>(f9_38);
   int64_t h0 = f0f0  +f1f9_76+f2f8_38+f3f7_76+f4f6_38+f5f5_38;
   int64_t h1 = f0f1_2+f2f9_38+f3f8_38+f4f7_38+f5f6_38;
   int64_t h2 = f0f2_2+f1f1_2 +f3f9_76+f4f8_38+f5f7_76+f6f6_19;
   int64_t h3 = f0f3_2+f1f2_2 +f4f9_38+f5f8_38+f6f7_38;
   int64_t h4 = f0f4_2+f1f3_4 +f2f2   +f5f9_76+f6f8_38+f7f7_38;
   int64_t h5 = f0f5_2+f1f4_2 +f2f3_2 +f6f9_38+f7f8_38;
   int64_t h6 = f0f6_2+f1f5_4 +f2f4_2 +f3f3_2 +f7f9_76+f8f8_19;
   int64_t h7 = f0f7_2+f1f6_2 +f2f5_2 +f3f4_2 +f8f9_38;
   int64_t h8 = f0f8_2+f1f7_4 +f2f6_2 +f3f5_4 +f4f4   +f9f9_38;
   int64_t h9 = f0f9_2+f1f8_2 +f2f7_2 +f3f6_2 +f4f5_2;
   int64_t carry0;
   int64_t carry1;
   int64_t carry2;
   int64_t carry3;
   int64_t carry4;
   int64_t carry5;
   int64_t carry6;
   int64_t carry7;
   int64_t carry8;
   int64_t carry9;

   h0 += h0;
   h1 += h1;
   h2 += h2;
   h3 += h3;
   h4 += h4;
   h5 += h5;
   h6 += h6;
   h7 += h7;
   h8 += h8;
   h9 += h9;

   carry0 = (h0 + X25) >> 26;
   h1 += carry0;
   h0 -= carry0 * X26;
   carry4 = (h4 + X25) >> 26;
   h5 += carry4;
   h4 -= carry4 * X26;

   carry1 = (h1 + X24) >> 25;
   h2 += carry1;
   h1 -= carry1 * X25;
   carry5 = (h5 + X24) >> 25;
   h6 += carry5;
   h5 -= carry5 * X25;

   carry2 = (h2 + X25) >> 26;
   h3 += carry2;
   h2 -= carry2 * X26;
   carry6 = (h6 + X25) >> 26;
   h7 += carry6;
   h6 -= carry6 * X26;

   carry3 = (h3 + X24) >> 25;
   h4 += carry3;
   h3 -= carry3 * X25;
   carry7 = (h7 + X24) >> 25;
   h8 += carry7;
   h7 -= carry7 * X25;

   carry4 = (h4 + X25) >> 26;
   h5 += carry4;
   h4 -= carry4 * X26;
   carry8 = (h8 + X25) >> 26;
   h9 += carry8;
   h8 -= carry8 * X26;

   carry9 = (h9 + X24) >> 25;
   h0 += carry9 * 19;
   h9 -= carry9 * X25;

   carry0 = (h0 + X25) >> 26;
   h1 += carry0;
   h0 -= carry0 * X26;

   return FE_25519(h0, h1, h2, h3, h4, h5, h6, h7, h8, h9);
   }

/*
Ignores top bit of h.
*/

void FE_25519::from_bytes(const uint8_t s[32])
   {
   const int64_t X24 = (1 << 24);
   const int64_t X25 = (1 << 25);
   const int64_t X26 = (1 << 26);

   int64_t h0 = load_4(s);
   int64_t h1 = load_3(s + 4) << 6;
   int64_t h2 = load_3(s + 7) << 5;
   int64_t h3 = load_3(s + 10) << 3;
   int64_t h4 = load_3(s + 13) << 2;
   int64_t h5 = load_4(s + 16);
   int64_t h6 = load_3(s + 20) << 7;
   int64_t h7 = load_3(s + 23) << 5;
   int64_t h8 = load_3(s + 26) << 4;
   int64_t h9 = (load_3(s + 29) & 0x7fffff) << 2;

   const int64_t carry9 = (h9 + X24) >> 25;
   h0 += carry9 * 19;
   h9 -= carry9 * X25;
   const int64_t carry1 = (h1 + X24) >> 25;
   h2 += carry1;
   h1 -= carry1 * X25;
   const int64_t carry3 = (h3 + X24) >> 25;
   h4 += carry3;
   h3 -= carry3 * X25;
   const int64_t carry5 = (h5 + X24) >> 25;
   h6 += carry5;
   h5 -= carry5 * X25;
   const int64_t carry7 = (h7 + X24) >> 25;
   h8 += carry7;
   h7 -= carry7 * X25;

   const int64_t carry0 = (h0 + X25) >> 26;
   h1 += carry0;
   h0 -= carry0 * X26;
   const int64_t carry2 = (h2 + X25) >> 26;
   h3 += carry2;
   h2 -= carry2 * X26;
   const int64_t carry4 = (h4 + X25) >> 26;
   h5 += carry4;
   h4 -= carry4 * X26;
   const int64_t carry6 = (h6 + X25) >> 26;
   h7 += carry6;
   h6 -= carry6 * X26;
   const int64_t carry8 = (h8 + X25) >> 26;
   h9 += carry8;
   h8 -= carry8 * X26;

   m_fe[0] = h0;
   m_fe[1] = h1;
   m_fe[2] = h2;
   m_fe[3] = h3;
   m_fe[4] = h4;
   m_fe[5] = h5;
   m_fe[6] = h6;
   m_fe[7] = h7;
   m_fe[8] = h8;
   m_fe[9] = h9;
   }

/*
Preconditions:
|h| bounded by 1.1*2^26,1.1*2^25,1.1*2^26,1.1*2^25,etc.

Write p=2^255-19; q=floor(h/p).
Basic claim: q = floor(2^(-255)(h + 19 2^(-25)h9 + 2^(-1))).

Proof:
Have |h|<=p so |q|<=1 so |19^2 2^(-255) q|<1/4.
Also have |h-2^230 h9|<2^231 so |19 2^(-255)(h-2^230 h9)|<1/4.

Write y=2^(-1)-19^2 2^(-255)q-19 2^(-255)(h-2^230 h9).
Then 0<y<1.

Write r=h-pq.
Have 0<=r<=p-1=2^255-20.
Thus 0<=r+19(2^-255)r<r+19(2^-255)2^255<=2^255-1.

Write x=r+19(2^-255)r+y.
Then 0<x<2^255 so floor(2^(-255)x) = 0 so floor(q+2^(-255)x) = q.

Have q+2^(-255)x = 2^(-255)(h + 19 2^(-25) h9 + 2^(-1))
so floor(2^(-255)(h + 19 2^(-25) h9 + 2^(-1))) = q.
*/

void FE_25519::to_bytes(uint8_t s[32]) const
   {
   const int64_t X25 = (1 << 25);
   const int64_t X26 = (1 << 26);

   int32_t h0 = m_fe[0];
   int32_t h1 = m_fe[1];
   int32_t h2 = m_fe[2];
   int32_t h3 = m_fe[3];
   int32_t h4 = m_fe[4];
   int32_t h5 = m_fe[5];
   int32_t h6 = m_fe[6];
   int32_t h7 = m_fe[7];
   int32_t h8 = m_fe[8];
   int32_t h9 = m_fe[9];
   int32_t q;
   int32_t carry0;
   int32_t carry1;
   int32_t carry2;
   int32_t carry3;
   int32_t carry4;
   int32_t carry5;
   int32_t carry6;
   int32_t carry7;
   int32_t carry8;
   int32_t carry9;

   q = (19 * h9 + ((static_cast<int32_t>(1) << 24))) >> 25;
   q = (h0 + q) >> 26;
   q = (h1 + q) >> 25;
   q = (h2 + q) >> 26;
   q = (h3 + q) >> 25;
   q = (h4 + q) >> 26;
   q = (h5 + q) >> 25;
   q = (h6 + q) >> 26;
   q = (h7 + q) >> 25;
   q = (h8 + q) >> 26;
   q = (h9 + q) >> 25;

   /* Goal: Output h-(2^255-19)q, which is between 0 and 2^255-20. */
   h0 += 19 * q;
   /* Goal: Output h-2^255 q, which is between 0 and 2^255-20. */

   carry0 = h0 >> 26;
   h1 += carry0;
   h0 -= carry0 * X26;
   carry1 = h1 >> 25;
   h2 += carry1;
   h1 -= carry1 * X25;
   carry2 = h2 >> 26;
   h3 += carry2;
   h2 -= carry2 * X26;
   carry3 = h3 >> 25;
   h4 += carry3;
   h3 -= carry3 * X25;
   carry4 = h4 >> 26;
   h5 += carry4;
   h4 -= carry4 * X26;
   carry5 = h5 >> 25;
   h6 += carry5;
   h5 -= carry5 * X25;
   carry6 = h6 >> 26;
   h7 += carry6;
   h6 -= carry6 * X26;
   carry7 = h7 >> 25;
   h8 += carry7;
   h7 -= carry7 * X25;
   carry8 = h8 >> 26;
   h9 += carry8;
   h8 -= carry8 * X26;
   carry9 = h9 >> 25;
   h9 -= carry9 * X25;
   /* h10 = carry9 */

   /*
   Goal: Output h0+...+2^255 h10-2^255 q, which is between 0 and 2^255-20.
   Have h0+...+2^230 h9 between 0 and 2^255-1;
   evidently 2^255 h10-2^255 q = 0.
   Goal: Output h0+...+2^230 h9.
   */

   s[0] = h0 >> 0;
   s[1] = h0 >> 8;
   s[2] = h0 >> 16;
   s[3] = (h0 >> 24) | (h1 << 2);
   s[4] = h1 >> 6;
   s[5] = h1 >> 14;
   s[6] = (h1 >> 22) | (h2 << 3);
   s[7] = h2 >> 5;
   s[8] = h2 >> 13;
   s[9] = (h2 >> 21) | (h3 << 5);
   s[10] = h3 >> 3;
   s[11] = h3 >> 11;
   s[12] = (h3 >> 19) | (h4 << 6);
   s[13] = h4 >> 2;
   s[14] = h4 >> 10;
   s[15] = h4 >> 18;
   s[16] = h5 >> 0;
   s[17] = h5 >> 8;
   s[18] = h5 >> 16;
   s[19] = (h5 >> 24) | (h6 << 1);
   s[20] = h6 >> 7;
   s[21] = h6 >> 15;
   s[22] = (h6 >> 23) | (h7 << 3);
   s[23] = h7 >> 5;
   s[24] = h7 >> 13;
   s[25] = (h7 >> 21) | (h8 << 4);
   s[26] = h8 >> 4;
   s[27] = h8 >> 12;
   s[28] = (h8 >> 20) | (h9 << 6);
   s[29] = h9 >> 2;
   s[30] = h9 >> 10;
   s[31] = h9 >> 18;
   }

}