summaryrefslogtreecommitdiffstats
path: root/security/nss/lib/freebl/mpi/mpmontg.c
blob: 36ae51b352998a614b6b660535054bb3ffad8299 (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
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
/* This Source Code Form is subject to the terms of the Mozilla Public
 * License, v. 2.0. If a copy of the MPL was not distributed with this
 * file, You can obtain one at http://mozilla.org/MPL/2.0/. */

/* This file implements moduluar exponentiation using Montgomery's
 * method for modular reduction.  This file implements the method
 * described as "Improvement 2" in the paper "A Cryptogrpahic Library for
 * the Motorola DSP56000" by Stephen R. Dusse' and Burton S. Kaliski Jr.
 * published in "Advances in Cryptology: Proceedings of EUROCRYPT '90"
 * "Lecture Notes in Computer Science" volume 473, 1991, pg 230-244,
 * published by Springer Verlag.
 */

#define MP_USING_CACHE_SAFE_MOD_EXP 1
#include <string.h>
#include "mpi-priv.h"
#include "mplogic.h"
#include "mpprime.h"
#ifdef MP_USING_MONT_MULF
#include "montmulf.h"
#endif
#include <stddef.h> /* ptrdiff_t */
#include <assert.h>

#define STATIC

#define MAX_ODD_INTS 32 /* 2 ** (WINDOW_BITS - 1) */

/*! computes T = REDC(T), 2^b == R
    \param T < RN
*/
mp_err
s_mp_redc(mp_int *T, mp_mont_modulus *mmm)
{
    mp_err res;
    mp_size i;

    i = (MP_USED(&mmm->N) << 1) + 1;
    MP_CHECKOK(s_mp_pad(T, i));
    for (i = 0; i < MP_USED(&mmm->N); ++i) {
        mp_digit m_i = MP_DIGIT(T, i) * mmm->n0prime;
        /* T += N * m_i * (MP_RADIX ** i); */
        s_mp_mul_d_add_offset(&mmm->N, m_i, T, i);
    }
    s_mp_clamp(T);

    /* T /= R */
    s_mp_rshd(T, MP_USED(&mmm->N));

    if ((res = s_mp_cmp(T, &mmm->N)) >= 0) {
        /* T = T - N */
        MP_CHECKOK(s_mp_sub(T, &mmm->N));
#ifdef DEBUG
        if ((res = mp_cmp(T, &mmm->N)) >= 0) {
            res = MP_UNDEF;
            goto CLEANUP;
        }
#endif
    }
    res = MP_OKAY;
CLEANUP:
    return res;
}

#if !defined(MP_MONT_USE_MP_MUL)

/*! c <- REDC( a * b ) mod N
    \param a < N  i.e. "reduced"
    \param b < N  i.e. "reduced"
    \param mmm modulus N and n0' of N
*/
mp_err
s_mp_mul_mont(const mp_int *a, const mp_int *b, mp_int *c,
              mp_mont_modulus *mmm)
{
    mp_digit *pb;
    mp_digit m_i;
    mp_err res;
    mp_size ib; /* "index b": index of current digit of B */
    mp_size useda, usedb;

    ARGCHK(a != NULL && b != NULL && c != NULL, MP_BADARG);

    if (MP_USED(a) < MP_USED(b)) {
        const mp_int *xch = b; /* switch a and b, to do fewer outer loops */
        b = a;
        a = xch;
    }

    MP_USED(c) = 1;
    MP_DIGIT(c, 0) = 0;
    ib = (MP_USED(&mmm->N) << 1) + 1;
    if ((res = s_mp_pad(c, ib)) != MP_OKAY)
        goto CLEANUP;

    useda = MP_USED(a);
    pb = MP_DIGITS(b);
    s_mpv_mul_d(MP_DIGITS(a), useda, *pb++, MP_DIGITS(c));
    s_mp_setz(MP_DIGITS(c) + useda + 1, ib - (useda + 1));
    m_i = MP_DIGIT(c, 0) * mmm->n0prime;
    s_mp_mul_d_add_offset(&mmm->N, m_i, c, 0);

    /* Outer loop:  Digits of b */
    usedb = MP_USED(b);
    for (ib = 1; ib < usedb; ib++) {
        mp_digit b_i = *pb++;

        /* Inner product:  Digits of a */
        if (b_i)
            s_mpv_mul_d_add_prop(MP_DIGITS(a), useda, b_i, MP_DIGITS(c) + ib);
        m_i = MP_DIGIT(c, ib) * mmm->n0prime;
        s_mp_mul_d_add_offset(&mmm->N, m_i, c, ib);
    }
    if (usedb < MP_USED(&mmm->N)) {
        for (usedb = MP_USED(&mmm->N); ib < usedb; ++ib) {
            m_i = MP_DIGIT(c, ib) * mmm->n0prime;
            s_mp_mul_d_add_offset(&mmm->N, m_i, c, ib);
        }
    }
    s_mp_clamp(c);
    s_mp_rshd(c, MP_USED(&mmm->N)); /* c /= R */
    if (s_mp_cmp(c, &mmm->N) >= 0) {
        MP_CHECKOK(s_mp_sub(c, &mmm->N));
    }
    res = MP_OKAY;

CLEANUP:
    return res;
}
#endif

STATIC
mp_err
s_mp_to_mont(const mp_int *x, mp_mont_modulus *mmm, mp_int *xMont)
{
    mp_err res;

    /* xMont = x * R mod N   where  N is modulus */
    MP_CHECKOK(mp_copy(x, xMont));
    MP_CHECKOK(s_mp_lshd(xMont, MP_USED(&mmm->N))); /* xMont = x << b */
    MP_CHECKOK(mp_div(xMont, &mmm->N, 0, xMont));   /*         mod N */
CLEANUP:
    return res;
}

#ifdef MP_USING_MONT_MULF

/* the floating point multiply is already cache safe,
 * don't turn on cache safe unless we specifically
 * force it */
#ifndef MP_FORCE_CACHE_SAFE
#undef MP_USING_CACHE_SAFE_MOD_EXP
#endif

unsigned int mp_using_mont_mulf = 1;

/* computes montgomery square of the integer in mResult */
#define SQR                                              \
    conv_i32_to_d32_and_d16(dm1, d16Tmp, mResult, nLen); \
    mont_mulf_noconv(mResult, dm1, d16Tmp,               \
                     dTmp, dn, MP_DIGITS(modulus), nLen, dn0)

/* computes montgomery product of x and the integer in mResult */
#define MUL(x)                                   \
    conv_i32_to_d32(dm1, mResult, nLen);         \
    mont_mulf_noconv(mResult, dm1, oddPowers[x], \
                     dTmp, dn, MP_DIGITS(modulus), nLen, dn0)

/* Do modular exponentiation using floating point multiply code. */
mp_err
mp_exptmod_f(const mp_int *montBase,
             const mp_int *exponent,
             const mp_int *modulus,
             mp_int *result,
             mp_mont_modulus *mmm,
             int nLen,
             mp_size bits_in_exponent,
             mp_size window_bits,
             mp_size odd_ints)
{
    mp_digit *mResult;
    double *dBuf = 0, *dm1, *dn, *dSqr, *d16Tmp, *dTmp;
    double dn0;
    mp_size i;
    mp_err res;
    int expOff;
    int dSize = 0, oddPowSize, dTmpSize;
    mp_int accum1;
    double *oddPowers[MAX_ODD_INTS];

    /* function for computing n0prime only works if n0 is odd */

    MP_DIGITS(&accum1) = 0;

    for (i = 0; i < MAX_ODD_INTS; ++i)
        oddPowers[i] = 0;

    MP_CHECKOK(mp_init_size(&accum1, 3 * nLen + 2));

    mp_set(&accum1, 1);
    MP_CHECKOK(s_mp_to_mont(&accum1, mmm, &accum1));
    MP_CHECKOK(s_mp_pad(&accum1, nLen));

    oddPowSize = 2 * nLen + 1;
    dTmpSize = 2 * oddPowSize;
    dSize = sizeof(double) * (nLen * 4 + 1 +
                              ((odd_ints + 1) * oddPowSize) + dTmpSize);
    dBuf = malloc(dSize);
    if (!dBuf) {
        res = MP_MEM;
        goto CLEANUP;
    }
    dm1 = dBuf;           /* array of d32 */
    dn = dBuf + nLen;     /* array of d32 */
    dSqr = dn + nLen;     /* array of d32 */
    d16Tmp = dSqr + nLen; /* array of d16 */
    dTmp = d16Tmp + oddPowSize;

    for (i = 0; i < odd_ints; ++i) {
        oddPowers[i] = dTmp;
        dTmp += oddPowSize;
    }
    mResult = (mp_digit *)(dTmp + dTmpSize); /* size is nLen + 1 */

    /* Make dn and dn0 */
    conv_i32_to_d32(dn, MP_DIGITS(modulus), nLen);
    dn0 = (double)(mmm->n0prime & 0xffff);

    /* Make dSqr */
    conv_i32_to_d32_and_d16(dm1, oddPowers[0], MP_DIGITS(montBase), nLen);
    mont_mulf_noconv(mResult, dm1, oddPowers[0],
                     dTmp, dn, MP_DIGITS(modulus), nLen, dn0);
    conv_i32_to_d32(dSqr, mResult, nLen);

    for (i = 1; i < odd_ints; ++i) {
        mont_mulf_noconv(mResult, dSqr, oddPowers[i - 1],
                         dTmp, dn, MP_DIGITS(modulus), nLen, dn0);
        conv_i32_to_d16(oddPowers[i], mResult, nLen);
    }

    s_mp_copy(MP_DIGITS(&accum1), mResult, nLen); /* from, to, len */

    for (expOff = bits_in_exponent - window_bits; expOff >= 0; expOff -= window_bits) {
        mp_size smallExp;
        MP_CHECKOK(mpl_get_bits(exponent, expOff, window_bits));
        smallExp = (mp_size)res;

        if (window_bits == 1) {
            if (!smallExp) {
                SQR;
            } else if (smallExp & 1) {
                SQR;
                MUL(0);
            } else {
                abort();
            }
        } else if (window_bits == 4) {
            if (!smallExp) {
                SQR;
                SQR;
                SQR;
                SQR;
            } else if (smallExp & 1) {
                SQR;
                SQR;
                SQR;
                SQR;
                MUL(smallExp / 2);
            } else if (smallExp & 2) {
                SQR;
                SQR;
                SQR;
                MUL(smallExp / 4);
                SQR;
            } else if (smallExp & 4) {
                SQR;
                SQR;
                MUL(smallExp / 8);
                SQR;
                SQR;
            } else if (smallExp & 8) {
                SQR;
                MUL(smallExp / 16);
                SQR;
                SQR;
                SQR;
            } else {
                abort();
            }
        } else if (window_bits == 5) {
            if (!smallExp) {
                SQR;
                SQR;
                SQR;
                SQR;
                SQR;
            } else if (smallExp & 1) {
                SQR;
                SQR;
                SQR;
                SQR;
                SQR;
                MUL(smallExp / 2);
            } else if (smallExp & 2) {
                SQR;
                SQR;
                SQR;
                SQR;
                MUL(smallExp / 4);
                SQR;
            } else if (smallExp & 4) {
                SQR;
                SQR;
                SQR;
                MUL(smallExp / 8);
                SQR;
                SQR;
            } else if (smallExp & 8) {
                SQR;
                SQR;
                MUL(smallExp / 16);
                SQR;
                SQR;
                SQR;
            } else if (smallExp & 0x10) {
                SQR;
                MUL(smallExp / 32);
                SQR;
                SQR;
                SQR;
                SQR;
            } else {
                abort();
            }
        } else if (window_bits == 6) {
            if (!smallExp) {
                SQR;
                SQR;
                SQR;
                SQR;
                SQR;
                SQR;
            } else if (smallExp & 1) {
                SQR;
                SQR;
                SQR;
                SQR;
                SQR;
                SQR;
                MUL(smallExp / 2);
            } else if (smallExp & 2) {
                SQR;
                SQR;
                SQR;
                SQR;
                SQR;
                MUL(smallExp / 4);
                SQR;
            } else if (smallExp & 4) {
                SQR;
                SQR;
                SQR;
                SQR;
                MUL(smallExp / 8);
                SQR;
                SQR;
            } else if (smallExp & 8) {
                SQR;
                SQR;
                SQR;
                MUL(smallExp / 16);
                SQR;
                SQR;
                SQR;
            } else if (smallExp & 0x10) {
                SQR;
                SQR;
                MUL(smallExp / 32);
                SQR;
                SQR;
                SQR;
                SQR;
            } else if (smallExp & 0x20) {
                SQR;
                MUL(smallExp / 64);
                SQR;
                SQR;
                SQR;
                SQR;
                SQR;
            } else {
                abort();
            }
        } else {
            abort();
        }
    }

    s_mp_copy(mResult, MP_DIGITS(&accum1), nLen); /* from, to, len */

    res = s_mp_redc(&accum1, mmm);
    mp_exch(&accum1, result);

CLEANUP:
    mp_clear(&accum1);
    if (dBuf) {
        if (dSize)
            memset(dBuf, 0, dSize);
        free(dBuf);
    }

    return res;
}
#undef SQR
#undef MUL
#endif

#define SQR(a, b)             \
    MP_CHECKOK(mp_sqr(a, b)); \
    MP_CHECKOK(s_mp_redc(b, mmm))

#if defined(MP_MONT_USE_MP_MUL)
#define MUL(x, a, b)                           \
    MP_CHECKOK(mp_mul(a, oddPowers + (x), b)); \
    MP_CHECKOK(s_mp_redc(b, mmm))
#else
#define MUL(x, a, b) \
    MP_CHECKOK(s_mp_mul_mont(a, oddPowers + (x), b, mmm))
#endif

#define SWAPPA  \
    ptmp = pa1; \
    pa1 = pa2;  \
    pa2 = ptmp

/* Do modular exponentiation using integer multiply code. */
mp_err
mp_exptmod_i(const mp_int *montBase,
             const mp_int *exponent,
             const mp_int *modulus,
             mp_int *result,
             mp_mont_modulus *mmm,
             int nLen,
             mp_size bits_in_exponent,
             mp_size window_bits,
             mp_size odd_ints)
{
    mp_int *pa1, *pa2, *ptmp;
    mp_size i;
    mp_err res;
    int expOff;
    mp_int accum1, accum2, power2, oddPowers[MAX_ODD_INTS];

    /* power2 = base ** 2; oddPowers[i] = base ** (2*i + 1); */
    /* oddPowers[i] = base ** (2*i + 1); */

    MP_DIGITS(&accum1) = 0;
    MP_DIGITS(&accum2) = 0;
    MP_DIGITS(&power2) = 0;
    for (i = 0; i < MAX_ODD_INTS; ++i) {
        MP_DIGITS(oddPowers + i) = 0;
    }

    MP_CHECKOK(mp_init_size(&accum1, 3 * nLen + 2));
    MP_CHECKOK(mp_init_size(&accum2, 3 * nLen + 2));

    MP_CHECKOK(mp_init_copy(&oddPowers[0], montBase));

    MP_CHECKOK(mp_init_size(&power2, nLen + 2 * MP_USED(montBase) + 2));
    MP_CHECKOK(mp_sqr(montBase, &power2)); /* power2 = montBase ** 2 */
    MP_CHECKOK(s_mp_redc(&power2, mmm));

    for (i = 1; i < odd_ints; ++i) {
        MP_CHECKOK(mp_init_size(oddPowers + i, nLen + 2 * MP_USED(&power2) + 2));
        MP_CHECKOK(mp_mul(oddPowers + (i - 1), &power2, oddPowers + i));
        MP_CHECKOK(s_mp_redc(oddPowers + i, mmm));
    }

    /* set accumulator to montgomery residue of 1 */
    mp_set(&accum1, 1);
    MP_CHECKOK(s_mp_to_mont(&accum1, mmm, &accum1));
    pa1 = &accum1;
    pa2 = &accum2;

    for (expOff = bits_in_exponent - window_bits; expOff >= 0; expOff -= window_bits) {
        mp_size smallExp;
        MP_CHECKOK(mpl_get_bits(exponent, expOff, window_bits));
        smallExp = (mp_size)res;

        if (window_bits == 1) {
            if (!smallExp) {
                SQR(pa1, pa2);
                SWAPPA;
            } else if (smallExp & 1) {
                SQR(pa1, pa2);
                MUL(0, pa2, pa1);
            } else {
                abort();
            }
        } else if (window_bits == 4) {
            if (!smallExp) {
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
            } else if (smallExp & 1) {
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                MUL(smallExp / 2, pa1, pa2);
                SWAPPA;
            } else if (smallExp & 2) {
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                MUL(smallExp / 4, pa2, pa1);
                SQR(pa1, pa2);
                SWAPPA;
            } else if (smallExp & 4) {
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                MUL(smallExp / 8, pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SWAPPA;
            } else if (smallExp & 8) {
                SQR(pa1, pa2);
                MUL(smallExp / 16, pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SWAPPA;
            } else {
                abort();
            }
        } else if (window_bits == 5) {
            if (!smallExp) {
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SWAPPA;
            } else if (smallExp & 1) {
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                MUL(smallExp / 2, pa2, pa1);
            } else if (smallExp & 2) {
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                MUL(smallExp / 4, pa1, pa2);
                SQR(pa2, pa1);
            } else if (smallExp & 4) {
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                MUL(smallExp / 8, pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
            } else if (smallExp & 8) {
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                MUL(smallExp / 16, pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
            } else if (smallExp & 0x10) {
                SQR(pa1, pa2);
                MUL(smallExp / 32, pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
            } else {
                abort();
            }
        } else if (window_bits == 6) {
            if (!smallExp) {
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
            } else if (smallExp & 1) {
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                MUL(smallExp / 2, pa1, pa2);
                SWAPPA;
            } else if (smallExp & 2) {
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                MUL(smallExp / 4, pa2, pa1);
                SQR(pa1, pa2);
                SWAPPA;
            } else if (smallExp & 4) {
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                MUL(smallExp / 8, pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SWAPPA;
            } else if (smallExp & 8) {
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                MUL(smallExp / 16, pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SWAPPA;
            } else if (smallExp & 0x10) {
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                MUL(smallExp / 32, pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SWAPPA;
            } else if (smallExp & 0x20) {
                SQR(pa1, pa2);
                MUL(smallExp / 64, pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SWAPPA;
            } else {
                abort();
            }
        } else {
            abort();
        }
    }

    res = s_mp_redc(pa1, mmm);
    mp_exch(pa1, result);

CLEANUP:
    mp_clear(&accum1);
    mp_clear(&accum2);
    mp_clear(&power2);
    for (i = 0; i < odd_ints; ++i) {
        mp_clear(oddPowers + i);
    }
    return res;
}
#undef SQR
#undef MUL

#ifdef MP_USING_CACHE_SAFE_MOD_EXP
unsigned int mp_using_cache_safe_exp = 1;
#endif

mp_err
mp_set_safe_modexp(int value)
{
#ifdef MP_USING_CACHE_SAFE_MOD_EXP
    mp_using_cache_safe_exp = value;
    return MP_OKAY;
#else
    if (value == 0) {
        return MP_OKAY;
    }
    return MP_BADARG;
#endif
}

#ifdef MP_USING_CACHE_SAFE_MOD_EXP
#define WEAVE_WORD_SIZE 4

/*
 * mpi_to_weave takes an array of bignums, a matrix in which each bignum
 * occupies all the columns of a row, and transposes it into a matrix in
 * which each bignum occupies a column of every row.  The first row of the
 * input matrix becomes the first column of the output matrix.  The n'th
 * row of input becomes the n'th column of output.  The input data is said
 * to be "interleaved" or "woven" into the output matrix.
 *
 * The array of bignums is left in this woven form.  Each time a single
 * bignum value is needed, it is recreated by fetching the n'th column,
 * forming a single row which is the new bignum.
 *
 * The purpose of this interleaving is make it impossible to determine which
 * of the bignums is being used in any one operation by examining the pattern
 * of cache misses.
 *
 * The weaving function does not transpose the entire input matrix in one call.
 * It transposes 4 rows of mp_ints into their respective columns of output.
 *
 * This implementation treats each mp_int bignum as an array of mp_digits,
 * It stores those bytes as a column of mp_digits in the output matrix.  It
 * doesn't care if the machine uses big-endian or little-endian byte ordering
 * within mp_digits.
 *
 * "bignums" is an array of mp_ints.
 * It points to four rows, four mp_ints, a subset of a larger array of mp_ints.
 *
 * "weaved" is the weaved output matrix.
 * The first byte of bignums[0] is stored in weaved[0].
 *
 * "nBignums" is the total number of bignums in the array of which "bignums"
 * is a part.
 *
 * "nDigits" is the size in mp_digits of each mp_int in the "bignums" array.
 * mp_ints that use less than nDigits digits are logically padded with zeros
 * while being stored in the weaved array.
 */
mp_err
mpi_to_weave(const mp_int *bignums,
             mp_digit *weaved,
             mp_size nDigits,  /* in each mp_int of input */
             mp_size nBignums) /* in the entire source array */
{
    mp_size i;
    mp_digit *endDest = weaved + (nDigits * nBignums);

    for (i = 0; i < WEAVE_WORD_SIZE; i++) {
        mp_size used = MP_USED(&bignums[i]);
        mp_digit *pSrc = MP_DIGITS(&bignums[i]);
        mp_digit *endSrc = pSrc + used;
        mp_digit *pDest = weaved + i;

        ARGCHK(MP_SIGN(&bignums[i]) == MP_ZPOS, MP_BADARG);
        ARGCHK(used <= nDigits, MP_BADARG);

        for (; pSrc < endSrc; pSrc++) {
            *pDest = *pSrc;
            pDest += nBignums;
        }
        while (pDest < endDest) {
            *pDest = 0;
            pDest += nBignums;
        }
    }

    return MP_OKAY;
}

/*
 * These functions return 0xffffffff if the output is true, and 0 otherwise.
 */
#define CONST_TIME_MSB(x) (0L - ((x) >> (8 * sizeof(x) - 1)))
#define CONST_TIME_EQ_Z(x) CONST_TIME_MSB(~(x) & ((x)-1))
#define CONST_TIME_EQ(a, b) CONST_TIME_EQ_Z((a) ^ (b))

/* Reverse the operation above for one mp_int.
 * Reconstruct one mp_int from its column in the weaved array.
 * Every read accesses every element of the weaved array, in order to
 * avoid timing attacks based on patterns of memory accesses.
 */
mp_err
weave_to_mpi(mp_int *a,              /* out, result */
             const mp_digit *weaved, /* in, byte matrix */
             mp_size index,          /* which column to read */
             mp_size nDigits,        /* number of mp_digits in each bignum */
             mp_size nBignums)       /* width of the matrix */
{
    /* these are indices, but need to be the same size as mp_digit
     * because of the CONST_TIME operations */
    mp_digit i, j;
    mp_digit d;
    mp_digit *pDest = MP_DIGITS(a);

    MP_SIGN(a) = MP_ZPOS;
    MP_USED(a) = nDigits;

    assert(weaved != NULL);

    /* Fetch the proper column in constant time, indexing over the whole array */
    for (i = 0; i < nDigits; ++i) {
        d = 0;
        for (j = 0; j < nBignums; ++j) {
            d |= weaved[i * nBignums + j] & CONST_TIME_EQ(j, index);
        }
        pDest[i] = d;
    }

    s_mp_clamp(a);
    return MP_OKAY;
}

#define SQR(a, b)             \
    MP_CHECKOK(mp_sqr(a, b)); \
    MP_CHECKOK(s_mp_redc(b, mmm))

#if defined(MP_MONT_USE_MP_MUL)
#define MUL_NOWEAVE(x, a, b)     \
    MP_CHECKOK(mp_mul(a, x, b)); \
    MP_CHECKOK(s_mp_redc(b, mmm))
#else
#define MUL_NOWEAVE(x, a, b) \
    MP_CHECKOK(s_mp_mul_mont(a, x, b, mmm))
#endif

#define MUL(x, a, b)                                               \
    MP_CHECKOK(weave_to_mpi(&tmp, powers, (x), nLen, num_powers)); \
    MUL_NOWEAVE(&tmp, a, b)

#define SWAPPA  \
    ptmp = pa1; \
    pa1 = pa2;  \
    pa2 = ptmp
#define MP_ALIGN(x, y) ((((ptrdiff_t)(x)) + ((y)-1)) & (((ptrdiff_t)0) - (y)))

/* Do modular exponentiation using integer multiply code. */
mp_err
mp_exptmod_safe_i(const mp_int *montBase,
                  const mp_int *exponent,
                  const mp_int *modulus,
                  mp_int *result,
                  mp_mont_modulus *mmm,
                  int nLen,
                  mp_size bits_in_exponent,
                  mp_size window_bits,
                  mp_size num_powers)
{
    mp_int *pa1, *pa2, *ptmp;
    mp_size i;
    mp_size first_window;
    mp_err res;
    int expOff;
    mp_int accum1, accum2, accum[WEAVE_WORD_SIZE];
    mp_int tmp;
    mp_digit *powersArray = NULL;
    mp_digit *powers = NULL;

    MP_DIGITS(&accum1) = 0;
    MP_DIGITS(&accum2) = 0;
    MP_DIGITS(&accum[0]) = 0;
    MP_DIGITS(&accum[1]) = 0;
    MP_DIGITS(&accum[2]) = 0;
    MP_DIGITS(&accum[3]) = 0;
    MP_DIGITS(&tmp) = 0;

    /* grab the first window value. This allows us to preload accumulator1
   * and save a conversion, some squares and a multiple*/
    MP_CHECKOK(mpl_get_bits(exponent,
                            bits_in_exponent - window_bits, window_bits));
    first_window = (mp_size)res;

    MP_CHECKOK(mp_init_size(&accum1, 3 * nLen + 2));
    MP_CHECKOK(mp_init_size(&accum2, 3 * nLen + 2));

    /* build the first WEAVE_WORD powers inline */
    /* if WEAVE_WORD_SIZE is not 4, this code will have to change */
    if (num_powers > 2) {
        MP_CHECKOK(mp_init_size(&accum[0], 3 * nLen + 2));
        MP_CHECKOK(mp_init_size(&accum[1], 3 * nLen + 2));
        MP_CHECKOK(mp_init_size(&accum[2], 3 * nLen + 2));
        MP_CHECKOK(mp_init_size(&accum[3], 3 * nLen + 2));
        mp_set(&accum[0], 1);
        MP_CHECKOK(s_mp_to_mont(&accum[0], mmm, &accum[0]));
        MP_CHECKOK(mp_copy(montBase, &accum[1]));
        SQR(montBase, &accum[2]);
        MUL_NOWEAVE(montBase, &accum[2], &accum[3]);
        powersArray = (mp_digit *)malloc(num_powers * (nLen * sizeof(mp_digit) + 1));
        if (!powersArray) {
            res = MP_MEM;
            goto CLEANUP;
        }
        /* powers[i] = base ** (i); */
        powers = (mp_digit *)MP_ALIGN(powersArray, num_powers);
        MP_CHECKOK(mpi_to_weave(accum, powers, nLen, num_powers));
        if (first_window < 4) {
            MP_CHECKOK(mp_copy(&accum[first_window], &accum1));
            first_window = num_powers;
        }
    } else {
        if (first_window == 0) {
            mp_set(&accum1, 1);
            MP_CHECKOK(s_mp_to_mont(&accum1, mmm, &accum1));
        } else {
            /* assert first_window == 1? */
            MP_CHECKOK(mp_copy(montBase, &accum1));
        }
    }

    /*
     * calculate all the powers in the powers array.
     * this adds 2**(k-1)-2 square operations over just calculating the
     * odd powers where k is the window size in the two other mp_modexpt
     * implementations in this file. We will get some of that
     * back by not needing the first 'k' squares and one multiply for the
     * first window.
     * Given the value of 4 for WEAVE_WORD_SIZE, this loop will only execute if
     * num_powers > 2, in which case powers will have been allocated.
     */
    for (i = WEAVE_WORD_SIZE; i < num_powers; i++) {
        int acc_index = i & (WEAVE_WORD_SIZE - 1); /* i % WEAVE_WORD_SIZE */
        if (i & 1) {
            MUL_NOWEAVE(montBase, &accum[acc_index - 1], &accum[acc_index]);
            /* we've filled the array do our 'per array' processing */
            if (acc_index == (WEAVE_WORD_SIZE - 1)) {
                MP_CHECKOK(mpi_to_weave(accum, powers + i - (WEAVE_WORD_SIZE - 1),
                                        nLen, num_powers));

                if (first_window <= i) {
                    MP_CHECKOK(mp_copy(&accum[first_window & (WEAVE_WORD_SIZE - 1)],
                                       &accum1));
                    first_window = num_powers;
                }
            }
        } else {
            /* up to 8 we can find 2^i-1 in the accum array, but at 8 we our source
             * and target are the same so we need to copy.. After that, the
             * value is overwritten, so we need to fetch it from the stored
             * weave array */
            if (i > 2 * WEAVE_WORD_SIZE) {
                MP_CHECKOK(weave_to_mpi(&accum2, powers, i / 2, nLen, num_powers));
                SQR(&accum2, &accum[acc_index]);
            } else {
                int half_power_index = (i / 2) & (WEAVE_WORD_SIZE - 1);
                if (half_power_index == acc_index) {
                    /* copy is cheaper than weave_to_mpi */
                    MP_CHECKOK(mp_copy(&accum[half_power_index], &accum2));
                    SQR(&accum2, &accum[acc_index]);
                } else {
                    SQR(&accum[half_power_index], &accum[acc_index]);
                }
            }
        }
    }
/* if the accum1 isn't set, Then there is something wrong with our logic
   * above and is an internal programming error.
   */
#if MP_ARGCHK == 2
    assert(MP_USED(&accum1) != 0);
#endif

    /* set accumulator to montgomery residue of 1 */
    pa1 = &accum1;
    pa2 = &accum2;

    /* tmp is not used if window_bits == 1. */
    if (window_bits != 1) {
        MP_CHECKOK(mp_init_size(&tmp, 3 * nLen + 2));
    }

    for (expOff = bits_in_exponent - window_bits * 2; expOff >= 0; expOff -= window_bits) {
        mp_size smallExp;
        MP_CHECKOK(mpl_get_bits(exponent, expOff, window_bits));
        smallExp = (mp_size)res;

        /* handle unroll the loops */
        switch (window_bits) {
            case 1:
                if (!smallExp) {
                    SQR(pa1, pa2);
                    SWAPPA;
                } else if (smallExp & 1) {
                    SQR(pa1, pa2);
                    MUL_NOWEAVE(montBase, pa2, pa1);
                } else {
                    abort();
                }
                break;
            case 6:
                SQR(pa1, pa2);
                SQR(pa2, pa1);
            /* fall through */
            case 4:
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                MUL(smallExp, pa1, pa2);
                SWAPPA;
                break;
            case 5:
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                SQR(pa2, pa1);
                SQR(pa1, pa2);
                MUL(smallExp, pa2, pa1);
                break;
            default:
                abort(); /* could do a loop? */
        }
    }

    res = s_mp_redc(pa1, mmm);
    mp_exch(pa1, result);

CLEANUP:
    mp_clear(&accum1);
    mp_clear(&accum2);
    mp_clear(&accum[0]);
    mp_clear(&accum[1]);
    mp_clear(&accum[2]);
    mp_clear(&accum[3]);
    mp_clear(&tmp);
    /* zero required by FIPS here, can't use PORT_ZFree
     * because mpi doesn't link with util */
    if (powers) {
        PORT_Memset(powers, 0, num_powers * sizeof(mp_digit));
    }
    free(powersArray);
    return res;
}
#undef SQR
#undef MUL
#endif

mp_err
mp_exptmod(const mp_int *inBase, const mp_int *exponent,
           const mp_int *modulus, mp_int *result)
{
    const mp_int *base;
    mp_size bits_in_exponent, i, window_bits, odd_ints;
    mp_err res;
    int nLen;
    mp_int montBase, goodBase;
    mp_mont_modulus mmm;
#ifdef MP_USING_CACHE_SAFE_MOD_EXP
    static unsigned int max_window_bits;
#endif

    /* function for computing n0prime only works if n0 is odd */
    if (!mp_isodd(modulus))
        return s_mp_exptmod(inBase, exponent, modulus, result);

    MP_DIGITS(&montBase) = 0;
    MP_DIGITS(&goodBase) = 0;

    if (mp_cmp(inBase, modulus) < 0) {
        base = inBase;
    } else {
        MP_CHECKOK(mp_init(&goodBase));
        base = &goodBase;
        MP_CHECKOK(mp_mod(inBase, modulus, &goodBase));
    }

    nLen = MP_USED(modulus);
    MP_CHECKOK(mp_init_size(&montBase, 2 * nLen + 2));

    mmm.N = *modulus; /* a copy of the mp_int struct */

    /* compute n0', given n0, n0' = -(n0 ** -1) mod MP_RADIX
    **        where n0 = least significant mp_digit of N, the modulus.
    */
    mmm.n0prime = 0 - s_mp_invmod_radix(MP_DIGIT(modulus, 0));

    MP_CHECKOK(s_mp_to_mont(base, &mmm, &montBase));

    bits_in_exponent = mpl_significant_bits(exponent);
#ifdef MP_USING_CACHE_SAFE_MOD_EXP
    if (mp_using_cache_safe_exp) {
        if (bits_in_exponent > 780)
            window_bits = 6;
        else if (bits_in_exponent > 256)
            window_bits = 5;
        else if (bits_in_exponent > 20)
            window_bits = 4;
        /* RSA public key exponents are typically under 20 bits (common values
         * are: 3, 17, 65537) and a 4-bit window is inefficient
         */
        else
            window_bits = 1;
    } else
#endif
        if (bits_in_exponent > 480)
        window_bits = 6;
    else if (bits_in_exponent > 160)
        window_bits = 5;
    else if (bits_in_exponent > 20)
        window_bits = 4;
    /* RSA public key exponents are typically under 20 bits (common values
     * are: 3, 17, 65537) and a 4-bit window is inefficient
     */
    else
        window_bits = 1;

#ifdef MP_USING_CACHE_SAFE_MOD_EXP
    /*
     * clamp the window size based on
     * the cache line size.
     */
    if (!max_window_bits) {
        unsigned long cache_size = s_mpi_getProcessorLineSize();
        /* processor has no cache, use 'fast' code always */
        if (cache_size == 0) {
            mp_using_cache_safe_exp = 0;
        }
        if ((cache_size == 0) || (cache_size >= 64)) {
            max_window_bits = 6;
        } else if (cache_size >= 32) {
            max_window_bits = 5;
        } else if (cache_size >= 16) {
            max_window_bits = 4;
        } else
            max_window_bits = 1; /* should this be an assert? */
    }

    /* clamp the window size down before we caclulate bits_in_exponent */
    if (mp_using_cache_safe_exp) {
        if (window_bits > max_window_bits) {
            window_bits = max_window_bits;
        }
    }
#endif

    odd_ints = 1 << (window_bits - 1);
    i = bits_in_exponent % window_bits;
    if (i != 0) {
        bits_in_exponent += window_bits - i;
    }

#ifdef MP_USING_MONT_MULF
    if (mp_using_mont_mulf) {
        MP_CHECKOK(s_mp_pad(&montBase, nLen));
        res = mp_exptmod_f(&montBase, exponent, modulus, result, &mmm, nLen,
                           bits_in_exponent, window_bits, odd_ints);
    } else
#endif
#ifdef MP_USING_CACHE_SAFE_MOD_EXP
        if (mp_using_cache_safe_exp) {
        res = mp_exptmod_safe_i(&montBase, exponent, modulus, result, &mmm, nLen,
                                bits_in_exponent, window_bits, 1 << window_bits);
    } else
#endif
        res = mp_exptmod_i(&montBase, exponent, modulus, result, &mmm, nLen,
                           bits_in_exponent, window_bits, odd_ints);

CLEANUP:
    mp_clear(&montBase);
    mp_clear(&goodBase);
    /* Don't mp_clear mmm.N because it is merely a copy of modulus.
    ** Just zap it.
    */
    memset(&mmm, 0, sizeof mmm);
    return res;
}