summaryrefslogtreecommitdiffstats
path: root/security/nss/lib/freebl/rijndael.c
blob: e4ad60388fe5d59d38533fcf06d99a2a0691abf9 (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
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
/* 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/. */

#ifdef FREEBL_NO_DEPEND
#include "stubs.h"
#endif

#include "prinit.h"
#include "prenv.h"
#include "prerr.h"
#include "secerr.h"

#include "prtypes.h"
#include "blapi.h"
#include "rijndael.h"

#include "cts.h"
#include "ctr.h"
#include "gcm.h"
#include "mpi.h"

#ifdef USE_HW_AES
#include "intel-aes.h"
#endif
#ifdef INTEL_GCM
#include "intel-gcm.h"
#endif /* INTEL_GCM */

/*
 * There are currently five ways to build this code, varying in performance
 * and code size.
 *
 * RIJNDAEL_INCLUDE_TABLES         Include all tables from rijndael32.tab
 * RIJNDAEL_GENERATE_TABLES        Generate tables on first
 *                                 encryption/decryption, then store them;
 *                                 use the function gfm
 * RIJNDAEL_GENERATE_TABLES_MACRO  Same as above, but use macros to do
 *                                 the generation
 * RIJNDAEL_GENERATE_VALUES        Do not store tables, generate the table
 *                                 values "on-the-fly", using gfm
 * RIJNDAEL_GENERATE_VALUES_MACRO  Same as above, but use macros
 *
 * The default is RIJNDAEL_INCLUDE_TABLES.
 */

/*
 * When building RIJNDAEL_INCLUDE_TABLES, includes S**-1, Rcon, T[0..4],
 *                                                 T**-1[0..4], IMXC[0..4]
 * When building anything else, includes S, S**-1, Rcon
 */
#include "rijndael32.tab"

#if defined(RIJNDAEL_INCLUDE_TABLES)
/*
 * RIJNDAEL_INCLUDE_TABLES
 */
#define T0(i) _T0[i]
#define T1(i) _T1[i]
#define T2(i) _T2[i]
#define T3(i) _T3[i]
#define TInv0(i) _TInv0[i]
#define TInv1(i) _TInv1[i]
#define TInv2(i) _TInv2[i]
#define TInv3(i) _TInv3[i]
#define IMXC0(b) _IMXC0[b]
#define IMXC1(b) _IMXC1[b]
#define IMXC2(b) _IMXC2[b]
#define IMXC3(b) _IMXC3[b]
/* The S-box can be recovered from the T-tables */
#ifdef IS_LITTLE_ENDIAN
#define SBOX(b) ((PRUint8)_T3[b])
#else
#define SBOX(b) ((PRUint8)_T1[b])
#endif
#define SINV(b) (_SInv[b])

#else /* not RIJNDAEL_INCLUDE_TABLES */

/*
 * Code for generating T-table values.
 */

#ifdef IS_LITTLE_ENDIAN
#define WORD4(b0, b1, b2, b3) \
    ((((PRUint32)b3) << 24) | \
     (((PRUint32)b2) << 16) | \
     (((PRUint32)b1) << 8) |  \
     ((PRUint32)b0))
#else
#define WORD4(b0, b1, b2, b3) \
    ((((PRUint32)b0) << 24) | \
     (((PRUint32)b1) << 16) | \
     (((PRUint32)b2) << 8) |  \
     ((PRUint32)b3))
#endif

/*
 * Define the S and S**-1 tables (both have been stored)
 */
#define SBOX(b) (_S[b])
#define SINV(b) (_SInv[b])

/*
 * The function xtime, used for Galois field multiplication
 */
#define XTIME(a) \
    ((a & 0x80) ? ((a << 1) ^ 0x1b) : (a << 1))

/* Choose GFM method (macros or function) */
#if defined(RIJNDAEL_GENERATE_TABLES_MACRO) || \
    defined(RIJNDAEL_GENERATE_VALUES_MACRO)

/*
 * Galois field GF(2**8) multipliers, in macro form
 */
#define GFM01(a) \
    (a) /* a * 01 = a, the identity */
#define GFM02(a) \
    (XTIME(a) & 0xff) /* a * 02 = xtime(a) */
#define GFM04(a) \
    (GFM02(GFM02(a))) /* a * 04 = xtime**2(a) */
#define GFM08(a) \
    (GFM02(GFM04(a))) /* a * 08 = xtime**3(a) */
#define GFM03(a) \
    (GFM01(a) ^ GFM02(a)) /* a * 03 = a * (01 + 02) */
#define GFM09(a) \
    (GFM01(a) ^ GFM08(a)) /* a * 09 = a * (01 + 08) */
#define GFM0B(a) \
    (GFM01(a) ^ GFM02(a) ^ GFM08(a)) /* a * 0B = a * (01 + 02 + 08) */
#define GFM0D(a) \
    (GFM01(a) ^ GFM04(a) ^ GFM08(a)) /* a * 0D = a * (01 + 04 + 08) */
#define GFM0E(a) \
    (GFM02(a) ^ GFM04(a) ^ GFM08(a)) /* a * 0E = a * (02 + 04 + 08) */

#else /* RIJNDAEL_GENERATE_TABLES or RIJNDAEL_GENERATE_VALUES */

/* GF_MULTIPLY
 *
 * multiply two bytes represented in GF(2**8), mod (x**4 + 1)
 */
PRUint8
gfm(PRUint8 a, PRUint8 b)
{
    PRUint8 res = 0;
    while (b > 0) {
        res = (b & 0x01) ? res ^ a : res;
        a = XTIME(a);
        b >>= 1;
    }
    return res;
}

#define GFM01(a) \
    (a) /* a * 01 = a, the identity */
#define GFM02(a) \
    (XTIME(a) & 0xff) /* a * 02 = xtime(a) */
#define GFM03(a) \
    (gfm(a, 0x03)) /* a * 03 */
#define GFM09(a) \
    (gfm(a, 0x09)) /* a * 09 */
#define GFM0B(a) \
    (gfm(a, 0x0B)) /* a * 0B */
#define GFM0D(a) \
    (gfm(a, 0x0D)) /* a * 0D */
#define GFM0E(a) \
    (gfm(a, 0x0E)) /* a * 0E */

#endif /* choosing GFM function */

/*
 * The T-tables
 */
#define G_T0(i) \
    (WORD4(GFM02(SBOX(i)), GFM01(SBOX(i)), GFM01(SBOX(i)), GFM03(SBOX(i))))
#define G_T1(i) \
    (WORD4(GFM03(SBOX(i)), GFM02(SBOX(i)), GFM01(SBOX(i)), GFM01(SBOX(i))))
#define G_T2(i) \
    (WORD4(GFM01(SBOX(i)), GFM03(SBOX(i)), GFM02(SBOX(i)), GFM01(SBOX(i))))
#define G_T3(i) \
    (WORD4(GFM01(SBOX(i)), GFM01(SBOX(i)), GFM03(SBOX(i)), GFM02(SBOX(i))))

/*
 * The inverse T-tables
 */
#define G_TInv0(i) \
    (WORD4(GFM0E(SINV(i)), GFM09(SINV(i)), GFM0D(SINV(i)), GFM0B(SINV(i))))
#define G_TInv1(i) \
    (WORD4(GFM0B(SINV(i)), GFM0E(SINV(i)), GFM09(SINV(i)), GFM0D(SINV(i))))
#define G_TInv2(i) \
    (WORD4(GFM0D(SINV(i)), GFM0B(SINV(i)), GFM0E(SINV(i)), GFM09(SINV(i))))
#define G_TInv3(i) \
    (WORD4(GFM09(SINV(i)), GFM0D(SINV(i)), GFM0B(SINV(i)), GFM0E(SINV(i))))

/*
 * The inverse mix column tables
 */
#define G_IMXC0(i) \
    (WORD4(GFM0E(i), GFM09(i), GFM0D(i), GFM0B(i)))
#define G_IMXC1(i) \
    (WORD4(GFM0B(i), GFM0E(i), GFM09(i), GFM0D(i)))
#define G_IMXC2(i) \
    (WORD4(GFM0D(i), GFM0B(i), GFM0E(i), GFM09(i)))
#define G_IMXC3(i) \
    (WORD4(GFM09(i), GFM0D(i), GFM0B(i), GFM0E(i)))

/* Now choose the T-table indexing method */
#if defined(RIJNDAEL_GENERATE_VALUES)
/* generate values for the tables with a function*/
static PRUint32
gen_TInvXi(PRUint8 tx, PRUint8 i)
{
    PRUint8 si01, si02, si03, si04, si08, si09, si0B, si0D, si0E;
    si01 = SINV(i);
    si02 = XTIME(si01);
    si04 = XTIME(si02);
    si08 = XTIME(si04);
    si03 = si02 ^ si01;
    si09 = si08 ^ si01;
    si0B = si08 ^ si03;
    si0D = si09 ^ si04;
    si0E = si08 ^ si04 ^ si02;
    switch (tx) {
        case 0:
            return WORD4(si0E, si09, si0D, si0B);
        case 1:
            return WORD4(si0B, si0E, si09, si0D);
        case 2:
            return WORD4(si0D, si0B, si0E, si09);
        case 3:
            return WORD4(si09, si0D, si0B, si0E);
    }
    return -1;
}
#define T0(i) G_T0(i)
#define T1(i) G_T1(i)
#define T2(i) G_T2(i)
#define T3(i) G_T3(i)
#define TInv0(i) gen_TInvXi(0, i)
#define TInv1(i) gen_TInvXi(1, i)
#define TInv2(i) gen_TInvXi(2, i)
#define TInv3(i) gen_TInvXi(3, i)
#define IMXC0(b) G_IMXC0(b)
#define IMXC1(b) G_IMXC1(b)
#define IMXC2(b) G_IMXC2(b)
#define IMXC3(b) G_IMXC3(b)
#elif defined(RIJNDAEL_GENERATE_VALUES_MACRO)
/* generate values for the tables with macros */
#define T0(i) G_T0(i)
#define T1(i) G_T1(i)
#define T2(i) G_T2(i)
#define T3(i) G_T3(i)
#define TInv0(i) G_TInv0(i)
#define TInv1(i) G_TInv1(i)
#define TInv2(i) G_TInv2(i)
#define TInv3(i) G_TInv3(i)
#define IMXC0(b) G_IMXC0(b)
#define IMXC1(b) G_IMXC1(b)
#define IMXC2(b) G_IMXC2(b)
#define IMXC3(b) G_IMXC3(b)
#else /* RIJNDAEL_GENERATE_TABLES or RIJNDAEL_GENERATE_TABLES_MACRO */
/* Generate T and T**-1 table values and store, then index */
/* The inverse mix column tables are still generated */
#define T0(i) rijndaelTables->T0[i]
#define T1(i) rijndaelTables->T1[i]
#define T2(i) rijndaelTables->T2[i]
#define T3(i) rijndaelTables->T3[i]
#define TInv0(i) rijndaelTables->TInv0[i]
#define TInv1(i) rijndaelTables->TInv1[i]
#define TInv2(i) rijndaelTables->TInv2[i]
#define TInv3(i) rijndaelTables->TInv3[i]
#define IMXC0(b) G_IMXC0(b)
#define IMXC1(b) G_IMXC1(b)
#define IMXC2(b) G_IMXC2(b)
#define IMXC3(b) G_IMXC3(b)
#endif /* choose T-table indexing method */

#endif /* not RIJNDAEL_INCLUDE_TABLES */

#if defined(RIJNDAEL_GENERATE_TABLES) || \
    defined(RIJNDAEL_GENERATE_TABLES_MACRO)

/* Code to generate and store the tables */

struct rijndael_tables_str {
    PRUint32 T0[256];
    PRUint32 T1[256];
    PRUint32 T2[256];
    PRUint32 T3[256];
    PRUint32 TInv0[256];
    PRUint32 TInv1[256];
    PRUint32 TInv2[256];
    PRUint32 TInv3[256];
};

static struct rijndael_tables_str *rijndaelTables = NULL;
static PRCallOnceType coRTInit = { 0, 0, 0 };
static PRStatus
init_rijndael_tables(void)
{
    PRUint32 i;
    PRUint8 si01, si02, si03, si04, si08, si09, si0B, si0D, si0E;
    struct rijndael_tables_str *rts;
    rts = (struct rijndael_tables_str *)
        PORT_Alloc(sizeof(struct rijndael_tables_str));
    if (!rts)
        return PR_FAILURE;
    for (i = 0; i < 256; i++) {
        /* The forward values */
        si01 = SBOX(i);
        si02 = XTIME(si01);
        si03 = si02 ^ si01;
        rts->T0[i] = WORD4(si02, si01, si01, si03);
        rts->T1[i] = WORD4(si03, si02, si01, si01);
        rts->T2[i] = WORD4(si01, si03, si02, si01);
        rts->T3[i] = WORD4(si01, si01, si03, si02);
        /* The inverse values */
        si01 = SINV(i);
        si02 = XTIME(si01);
        si04 = XTIME(si02);
        si08 = XTIME(si04);
        si03 = si02 ^ si01;
        si09 = si08 ^ si01;
        si0B = si08 ^ si03;
        si0D = si09 ^ si04;
        si0E = si08 ^ si04 ^ si02;
        rts->TInv0[i] = WORD4(si0E, si09, si0D, si0B);
        rts->TInv1[i] = WORD4(si0B, si0E, si09, si0D);
        rts->TInv2[i] = WORD4(si0D, si0B, si0E, si09);
        rts->TInv3[i] = WORD4(si09, si0D, si0B, si0E);
    }
    /* wait until all the values are in to set */
    rijndaelTables = rts;
    return PR_SUCCESS;
}

#endif /* code to generate tables */

/**************************************************************************
 *
 * Stuff related to the Rijndael key schedule
 *
 *************************************************************************/

#define SUBBYTE(w)                                \
    ((((PRUint32)SBOX((w >> 24) & 0xff)) << 24) | \
     (((PRUint32)SBOX((w >> 16) & 0xff)) << 16) | \
     (((PRUint32)SBOX((w >> 8) & 0xff)) << 8) |   \
     (((PRUint32)SBOX((w)&0xff))))

#ifdef IS_LITTLE_ENDIAN
#define ROTBYTE(b) \
    ((b >> 8) | (b << 24))
#else
#define ROTBYTE(b) \
    ((b << 8) | (b >> 24))
#endif

/* rijndael_key_expansion7
 *
 * Generate the expanded key from the key input by the user.
 * XXX
 * Nk == 7 (224 key bits) is a weird case.  Since Nk > 6, an added SubByte
 * transformation is done periodically.  The period is every 4 bytes, and
 * since 7%4 != 0 this happens at different times for each key word (unlike
 * Nk == 8 where it happens twice in every key word, in the same positions).
 * For now, I'm implementing this case "dumbly", w/o any unrolling.
 */
static void
rijndael_key_expansion7(AESContext *cx, const unsigned char *key, unsigned int Nk)
{
    unsigned int i;
    PRUint32 *W;
    PRUint32 *pW;
    PRUint32 tmp;
    W = cx->expandedKey;
    /* 1.  the first Nk words contain the cipher key */
    memcpy(W, key, Nk * 4);
    i = Nk;
    /* 2.  loop until full expanded key is obtained */
    pW = W + i - 1;
    for (; i < cx->Nb * (cx->Nr + 1); ++i) {
        tmp = *pW++;
        if (i % Nk == 0)
            tmp = SUBBYTE(ROTBYTE(tmp)) ^ Rcon[i / Nk - 1];
        else if (i % Nk == 4)
            tmp = SUBBYTE(tmp);
        *pW = W[i - Nk] ^ tmp;
    }
}

#if defined(NSS_X86_OR_X64)
#define EXPAND_KEY128(k, rcon, res)                   \
    tmp_key = _mm_aeskeygenassist_si128(k, rcon);     \
    tmp_key = _mm_shuffle_epi32(tmp_key, 0xFF);       \
    tmp = _mm_xor_si128(k, _mm_slli_si128(k, 4));     \
    tmp = _mm_xor_si128(tmp, _mm_slli_si128(tmp, 4)); \
    tmp = _mm_xor_si128(tmp, _mm_slli_si128(tmp, 4)); \
    res = _mm_xor_si128(tmp, tmp_key)

static void
native_key_expansion128(AESContext *cx, const unsigned char *key)
{
    __m128i *keySchedule = cx->keySchedule;
    pre_align __m128i tmp_key post_align;
    pre_align __m128i tmp post_align;
    keySchedule[0] = _mm_loadu_si128((__m128i *)key);
    EXPAND_KEY128(keySchedule[0], 0x01, keySchedule[1]);
    EXPAND_KEY128(keySchedule[1], 0x02, keySchedule[2]);
    EXPAND_KEY128(keySchedule[2], 0x04, keySchedule[3]);
    EXPAND_KEY128(keySchedule[3], 0x08, keySchedule[4]);
    EXPAND_KEY128(keySchedule[4], 0x10, keySchedule[5]);
    EXPAND_KEY128(keySchedule[5], 0x20, keySchedule[6]);
    EXPAND_KEY128(keySchedule[6], 0x40, keySchedule[7]);
    EXPAND_KEY128(keySchedule[7], 0x80, keySchedule[8]);
    EXPAND_KEY128(keySchedule[8], 0x1B, keySchedule[9]);
    EXPAND_KEY128(keySchedule[9], 0x36, keySchedule[10]);
}

#define EXPAND_KEY192_PART1(res, k0, kt, rcon)                                \
    tmp2 = _mm_slli_si128(k0, 4);                                             \
    tmp1 = _mm_xor_si128(k0, tmp2);                                           \
    tmp2 = _mm_slli_si128(tmp2, 4);                                           \
    tmp1 = _mm_xor_si128(_mm_xor_si128(tmp1, tmp2), _mm_slli_si128(tmp2, 4)); \
    tmp2 = _mm_aeskeygenassist_si128(kt, rcon);                               \
    res = _mm_xor_si128(tmp1, _mm_shuffle_epi32(tmp2, 0x55))

#define EXPAND_KEY192_PART2(res, k1, k2)             \
    tmp2 = _mm_xor_si128(k1, _mm_slli_si128(k1, 4)); \
    res = _mm_xor_si128(tmp2, _mm_shuffle_epi32(k2, 0xFF))

#define EXPAND_KEY192(k0, res1, res2, res3, carry, rcon1, rcon2)         \
    EXPAND_KEY192_PART1(tmp3, k0, res1, rcon1);                          \
    EXPAND_KEY192_PART2(carry, res1, tmp3);                              \
    res1 = _mm_castpd_si128(_mm_shuffle_pd(_mm_castsi128_pd(res1),       \
                                           _mm_castsi128_pd(tmp3), 0));  \
    res2 = _mm_castpd_si128(_mm_shuffle_pd(_mm_castsi128_pd(tmp3),       \
                                           _mm_castsi128_pd(carry), 1)); \
    EXPAND_KEY192_PART1(res3, tmp3, carry, rcon2)

static void
native_key_expansion192(AESContext *cx, const unsigned char *key)
{
    __m128i *keySchedule = cx->keySchedule;
    pre_align __m128i tmp1 post_align;
    pre_align __m128i tmp2 post_align;
    pre_align __m128i tmp3 post_align;
    pre_align __m128i carry post_align;
    keySchedule[0] = _mm_loadu_si128((__m128i *)key);
    keySchedule[1] = _mm_loadu_si128((__m128i *)(key + 16));
    EXPAND_KEY192(keySchedule[0], keySchedule[1], keySchedule[2],
                  keySchedule[3], carry, 0x1, 0x2);
    EXPAND_KEY192_PART2(keySchedule[4], carry, keySchedule[3]);
    EXPAND_KEY192(keySchedule[3], keySchedule[4], keySchedule[5],
                  keySchedule[6], carry, 0x4, 0x8);
    EXPAND_KEY192_PART2(keySchedule[7], carry, keySchedule[6]);
    EXPAND_KEY192(keySchedule[6], keySchedule[7], keySchedule[8],
                  keySchedule[9], carry, 0x10, 0x20);
    EXPAND_KEY192_PART2(keySchedule[10], carry, keySchedule[9]);
    EXPAND_KEY192(keySchedule[9], keySchedule[10], keySchedule[11],
                  keySchedule[12], carry, 0x40, 0x80);
}

#define EXPAND_KEY256_PART(res, rconx, k1x, k2x, X)                           \
    tmp_key = _mm_shuffle_epi32(_mm_aeskeygenassist_si128(k2x, rconx), X);    \
    tmp2 = _mm_slli_si128(k1x, 4);                                            \
    tmp1 = _mm_xor_si128(k1x, tmp2);                                          \
    tmp2 = _mm_slli_si128(tmp2, 4);                                           \
    tmp1 = _mm_xor_si128(_mm_xor_si128(tmp1, tmp2), _mm_slli_si128(tmp2, 4)); \
    res = _mm_xor_si128(tmp1, tmp_key);

#define EXPAND_KEY256(res1, res2, k1, k2, rcon)   \
    EXPAND_KEY256_PART(res1, rcon, k1, k2, 0xFF); \
    EXPAND_KEY256_PART(res2, 0x00, k2, res1, 0xAA)

static void
native_key_expansion256(AESContext *cx, const unsigned char *key)
{
    __m128i *keySchedule = cx->keySchedule;
    pre_align __m128i tmp_key post_align;
    pre_align __m128i tmp1 post_align;
    pre_align __m128i tmp2 post_align;
    keySchedule[0] = _mm_loadu_si128((__m128i *)key);
    keySchedule[1] = _mm_loadu_si128((__m128i *)(key + 16));
    EXPAND_KEY256(keySchedule[2], keySchedule[3], keySchedule[0],
                  keySchedule[1], 0x01);
    EXPAND_KEY256(keySchedule[4], keySchedule[5], keySchedule[2],
                  keySchedule[3], 0x02);
    EXPAND_KEY256(keySchedule[6], keySchedule[7], keySchedule[4],
                  keySchedule[5], 0x04);
    EXPAND_KEY256(keySchedule[8], keySchedule[9], keySchedule[6],
                  keySchedule[7], 0x08);
    EXPAND_KEY256(keySchedule[10], keySchedule[11], keySchedule[8],
                  keySchedule[9], 0x10);
    EXPAND_KEY256(keySchedule[12], keySchedule[13], keySchedule[10],
                  keySchedule[11], 0x20);
    EXPAND_KEY256_PART(keySchedule[14], 0x40, keySchedule[12],
                       keySchedule[13], 0xFF);
}

#endif /* NSS_X86_OR_X64 */

/*
 * AES key expansion using aes-ni instructions.
 */
static void
native_key_expansion(AESContext *cx, const unsigned char *key, unsigned int Nk)
{
#ifdef NSS_X86_OR_X64
    switch (Nk) {
        case 4:
            native_key_expansion128(cx, key);
            return;
        case 6:
            native_key_expansion192(cx, key);
            return;
        case 8:
            native_key_expansion256(cx, key);
            return;
        default:
            /* This shouldn't happen. */
            PORT_Assert(0);
    }
#else
    PORT_Assert(0);
#endif /* NSS_X86_OR_X64 */
}

static void
native_encryptBlock(AESContext *cx,
                    unsigned char *output,
                    const unsigned char *input)
{
#ifdef NSS_X86_OR_X64
    int i;
    pre_align __m128i m post_align = _mm_loadu_si128((__m128i *)input);
    m = _mm_xor_si128(m, cx->keySchedule[0]);
    for (i = 1; i < cx->Nr; ++i) {
        m = _mm_aesenc_si128(m, cx->keySchedule[i]);
    }
    m = _mm_aesenclast_si128(m, cx->keySchedule[cx->Nr]);
    _mm_storeu_si128((__m128i *)output, m);
#else
    PORT_Assert(0);
#endif /* NSS_X86_OR_X64 */
}

/* rijndael_key_expansion
 *
 * Generate the expanded key from the key input by the user.
 */
static void
rijndael_key_expansion(AESContext *cx, const unsigned char *key, unsigned int Nk)
{
    unsigned int i;
    PRUint32 *W;
    PRUint32 *pW;
    PRUint32 tmp;
    unsigned int round_key_words = cx->Nb * (cx->Nr + 1);
    if (Nk == 7) {
        rijndael_key_expansion7(cx, key, Nk);
        return;
    }
    W = cx->expandedKey;
    /* The first Nk words contain the input cipher key */
    memcpy(W, key, Nk * 4);
    i = Nk;
    pW = W + i - 1;
    /* Loop over all sets of Nk words, except the last */
    while (i < round_key_words - Nk) {
        tmp = *pW++;
        tmp = SUBBYTE(ROTBYTE(tmp)) ^ Rcon[i / Nk - 1];
        *pW = W[i++ - Nk] ^ tmp;
        tmp = *pW++;
        *pW = W[i++ - Nk] ^ tmp;
        tmp = *pW++;
        *pW = W[i++ - Nk] ^ tmp;
        tmp = *pW++;
        *pW = W[i++ - Nk] ^ tmp;
        if (Nk == 4)
            continue;
        switch (Nk) {
            case 8:
                tmp = *pW++;
                tmp = SUBBYTE(tmp);
                *pW = W[i++ - Nk] ^ tmp;
            case 7:
                tmp = *pW++;
                *pW = W[i++ - Nk] ^ tmp;
            case 6:
                tmp = *pW++;
                *pW = W[i++ - Nk] ^ tmp;
            case 5:
                tmp = *pW++;
                *pW = W[i++ - Nk] ^ tmp;
        }
    }
    /* Generate the last word */
    tmp = *pW++;
    tmp = SUBBYTE(ROTBYTE(tmp)) ^ Rcon[i / Nk - 1];
    *pW = W[i++ - Nk] ^ tmp;
    /* There may be overflow here, if Nk % (Nb * (Nr + 1)) > 0.  However,
     * since the above loop generated all but the last Nk key words, there
     * is no more need for the SubByte transformation.
     */
    if (Nk < 8) {
        for (; i < round_key_words; ++i) {
            tmp = *pW++;
            *pW = W[i - Nk] ^ tmp;
        }
    } else {
        /* except in the case when Nk == 8.  Then one more SubByte may have
         * to be performed, at i % Nk == 4.
         */
        for (; i < round_key_words; ++i) {
            tmp = *pW++;
            if (i % Nk == 4)
                tmp = SUBBYTE(tmp);
            *pW = W[i - Nk] ^ tmp;
        }
    }
}

/* rijndael_invkey_expansion
 *
 * Generate the expanded key for the inverse cipher from the key input by
 * the user.
 */
static void
rijndael_invkey_expansion(AESContext *cx, const unsigned char *key, unsigned int Nk)
{
    unsigned int r;
    PRUint32 *roundkeyw;
    PRUint8 *b;
    int Nb = cx->Nb;
    /* begins like usual key expansion ... */
    rijndael_key_expansion(cx, key, Nk);
    /* ... but has the additional step of InvMixColumn,
     * excepting the first and last round keys.
     */
    roundkeyw = cx->expandedKey + cx->Nb;
    for (r = 1; r < cx->Nr; ++r) {
        /* each key word, roundkeyw, represents a column in the key
         * matrix.  Each column is multiplied by the InvMixColumn matrix.
         *   [ 0E 0B 0D 09 ]   [ b0 ]
         *   [ 09 0E 0B 0D ] * [ b1 ]
         *   [ 0D 09 0E 0B ]   [ b2 ]
         *   [ 0B 0D 09 0E ]   [ b3 ]
         */
        b = (PRUint8 *)roundkeyw;
        *roundkeyw++ = IMXC0(b[0]) ^ IMXC1(b[1]) ^ IMXC2(b[2]) ^ IMXC3(b[3]);
        b = (PRUint8 *)roundkeyw;
        *roundkeyw++ = IMXC0(b[0]) ^ IMXC1(b[1]) ^ IMXC2(b[2]) ^ IMXC3(b[3]);
        b = (PRUint8 *)roundkeyw;
        *roundkeyw++ = IMXC0(b[0]) ^ IMXC1(b[1]) ^ IMXC2(b[2]) ^ IMXC3(b[3]);
        b = (PRUint8 *)roundkeyw;
        *roundkeyw++ = IMXC0(b[0]) ^ IMXC1(b[1]) ^ IMXC2(b[2]) ^ IMXC3(b[3]);
        if (Nb <= 4)
            continue;
        switch (Nb) {
            case 8:
                b = (PRUint8 *)roundkeyw;
                *roundkeyw++ = IMXC0(b[0]) ^ IMXC1(b[1]) ^
                               IMXC2(b[2]) ^ IMXC3(b[3]);
            case 7:
                b = (PRUint8 *)roundkeyw;
                *roundkeyw++ = IMXC0(b[0]) ^ IMXC1(b[1]) ^
                               IMXC2(b[2]) ^ IMXC3(b[3]);
            case 6:
                b = (PRUint8 *)roundkeyw;
                *roundkeyw++ = IMXC0(b[0]) ^ IMXC1(b[1]) ^
                               IMXC2(b[2]) ^ IMXC3(b[3]);
            case 5:
                b = (PRUint8 *)roundkeyw;
                *roundkeyw++ = IMXC0(b[0]) ^ IMXC1(b[1]) ^
                               IMXC2(b[2]) ^ IMXC3(b[3]);
        }
    }
}

/**************************************************************************
 *
 * Stuff related to Rijndael encryption/decryption.
 *
 *************************************************************************/

#ifdef IS_LITTLE_ENDIAN
#define BYTE0WORD(w) ((w)&0x000000ff)
#define BYTE1WORD(w) ((w)&0x0000ff00)
#define BYTE2WORD(w) ((w)&0x00ff0000)
#define BYTE3WORD(w) ((w)&0xff000000)
#else
#define BYTE0WORD(w) ((w)&0xff000000)
#define BYTE1WORD(w) ((w)&0x00ff0000)
#define BYTE2WORD(w) ((w)&0x0000ff00)
#define BYTE3WORD(w) ((w)&0x000000ff)
#endif

typedef union {
    PRUint32 w[4];
    PRUint8 b[16];
} rijndael_state;

#define COLUMN_0(state) state.w[0]
#define COLUMN_1(state) state.w[1]
#define COLUMN_2(state) state.w[2]
#define COLUMN_3(state) state.w[3]

#define STATE_BYTE(i) state.b[i]

static void NO_SANITIZE_ALIGNMENT
rijndael_encryptBlock128(AESContext *cx,
                         unsigned char *output,
                         const unsigned char *input)
{
    unsigned int r;
    PRUint32 *roundkeyw;
    rijndael_state state;
    PRUint32 C0, C1, C2, C3;
#if defined(NSS_X86_OR_X64)
#define pIn input
#define pOut output
#else
    unsigned char *pIn, *pOut;
    PRUint32 inBuf[4], outBuf[4];

    if ((ptrdiff_t)input & 0x3) {
        memcpy(inBuf, input, sizeof inBuf);
        pIn = (unsigned char *)inBuf;
    } else {
        pIn = (unsigned char *)input;
    }
    if ((ptrdiff_t)output & 0x3) {
        pOut = (unsigned char *)outBuf;
    } else {
        pOut = (unsigned char *)output;
    }
#endif
    roundkeyw = cx->expandedKey;
    /* Step 1: Add Round Key 0 to initial state */
    COLUMN_0(state) = *((PRUint32 *)(pIn)) ^ *roundkeyw++;
    COLUMN_1(state) = *((PRUint32 *)(pIn + 4)) ^ *roundkeyw++;
    COLUMN_2(state) = *((PRUint32 *)(pIn + 8)) ^ *roundkeyw++;
    COLUMN_3(state) = *((PRUint32 *)(pIn + 12)) ^ *roundkeyw++;
    /* Step 2: Loop over rounds [1..NR-1] */
    for (r = 1; r < cx->Nr; ++r) {
        /* Do ShiftRow, ByteSub, and MixColumn all at once */
        C0 = T0(STATE_BYTE(0)) ^
             T1(STATE_BYTE(5)) ^
             T2(STATE_BYTE(10)) ^
             T3(STATE_BYTE(15));
        C1 = T0(STATE_BYTE(4)) ^
             T1(STATE_BYTE(9)) ^
             T2(STATE_BYTE(14)) ^
             T3(STATE_BYTE(3));
        C2 = T0(STATE_BYTE(8)) ^
             T1(STATE_BYTE(13)) ^
             T2(STATE_BYTE(2)) ^
             T3(STATE_BYTE(7));
        C3 = T0(STATE_BYTE(12)) ^
             T1(STATE_BYTE(1)) ^
             T2(STATE_BYTE(6)) ^
             T3(STATE_BYTE(11));
        /* Round key addition */
        COLUMN_0(state) = C0 ^ *roundkeyw++;
        COLUMN_1(state) = C1 ^ *roundkeyw++;
        COLUMN_2(state) = C2 ^ *roundkeyw++;
        COLUMN_3(state) = C3 ^ *roundkeyw++;
    }
    /* Step 3: Do the last round */
    /* Final round does not employ MixColumn */
    C0 = ((BYTE0WORD(T2(STATE_BYTE(0)))) |
          (BYTE1WORD(T3(STATE_BYTE(5)))) |
          (BYTE2WORD(T0(STATE_BYTE(10)))) |
          (BYTE3WORD(T1(STATE_BYTE(15))))) ^
         *roundkeyw++;
    C1 = ((BYTE0WORD(T2(STATE_BYTE(4)))) |
          (BYTE1WORD(T3(STATE_BYTE(9)))) |
          (BYTE2WORD(T0(STATE_BYTE(14)))) |
          (BYTE3WORD(T1(STATE_BYTE(3))))) ^
         *roundkeyw++;
    C2 = ((BYTE0WORD(T2(STATE_BYTE(8)))) |
          (BYTE1WORD(T3(STATE_BYTE(13)))) |
          (BYTE2WORD(T0(STATE_BYTE(2)))) |
          (BYTE3WORD(T1(STATE_BYTE(7))))) ^
         *roundkeyw++;
    C3 = ((BYTE0WORD(T2(STATE_BYTE(12)))) |
          (BYTE1WORD(T3(STATE_BYTE(1)))) |
          (BYTE2WORD(T0(STATE_BYTE(6)))) |
          (BYTE3WORD(T1(STATE_BYTE(11))))) ^
         *roundkeyw++;
    *((PRUint32 *)pOut) = C0;
    *((PRUint32 *)(pOut + 4)) = C1;
    *((PRUint32 *)(pOut + 8)) = C2;
    *((PRUint32 *)(pOut + 12)) = C3;
#if defined(NSS_X86_OR_X64)
#undef pIn
#undef pOut
#else
    if ((ptrdiff_t)output & 0x3) {
        memcpy(output, outBuf, sizeof outBuf);
    }
#endif
}

static SECStatus NO_SANITIZE_ALIGNMENT
rijndael_decryptBlock128(AESContext *cx,
                         unsigned char *output,
                         const unsigned char *input)
{
    int r;
    PRUint32 *roundkeyw;
    rijndael_state state;
    PRUint32 C0, C1, C2, C3;
#if defined(NSS_X86_OR_X64)
#define pIn input
#define pOut output
#else
    unsigned char *pIn, *pOut;
    PRUint32 inBuf[4], outBuf[4];

    if ((ptrdiff_t)input & 0x3) {
        memcpy(inBuf, input, sizeof inBuf);
        pIn = (unsigned char *)inBuf;
    } else {
        pIn = (unsigned char *)input;
    }
    if ((ptrdiff_t)output & 0x3) {
        pOut = (unsigned char *)outBuf;
    } else {
        pOut = (unsigned char *)output;
    }
#endif
    roundkeyw = cx->expandedKey + cx->Nb * cx->Nr + 3;
    /* reverse the final key addition */
    COLUMN_3(state) = *((PRUint32 *)(pIn + 12)) ^ *roundkeyw--;
    COLUMN_2(state) = *((PRUint32 *)(pIn + 8)) ^ *roundkeyw--;
    COLUMN_1(state) = *((PRUint32 *)(pIn + 4)) ^ *roundkeyw--;
    COLUMN_0(state) = *((PRUint32 *)(pIn)) ^ *roundkeyw--;
    /* Loop over rounds in reverse [NR..1] */
    for (r = cx->Nr; r > 1; --r) {
        /* Invert the (InvByteSub*InvMixColumn)(InvShiftRow(state)) */
        C0 = TInv0(STATE_BYTE(0)) ^
             TInv1(STATE_BYTE(13)) ^
             TInv2(STATE_BYTE(10)) ^
             TInv3(STATE_BYTE(7));
        C1 = TInv0(STATE_BYTE(4)) ^
             TInv1(STATE_BYTE(1)) ^
             TInv2(STATE_BYTE(14)) ^
             TInv3(STATE_BYTE(11));
        C2 = TInv0(STATE_BYTE(8)) ^
             TInv1(STATE_BYTE(5)) ^
             TInv2(STATE_BYTE(2)) ^
             TInv3(STATE_BYTE(15));
        C3 = TInv0(STATE_BYTE(12)) ^
             TInv1(STATE_BYTE(9)) ^
             TInv2(STATE_BYTE(6)) ^
             TInv3(STATE_BYTE(3));
        /* Invert the key addition step */
        COLUMN_3(state) = C3 ^ *roundkeyw--;
        COLUMN_2(state) = C2 ^ *roundkeyw--;
        COLUMN_1(state) = C1 ^ *roundkeyw--;
        COLUMN_0(state) = C0 ^ *roundkeyw--;
    }
    /* inverse sub */
    pOut[0] = SINV(STATE_BYTE(0));
    pOut[1] = SINV(STATE_BYTE(13));
    pOut[2] = SINV(STATE_BYTE(10));
    pOut[3] = SINV(STATE_BYTE(7));
    pOut[4] = SINV(STATE_BYTE(4));
    pOut[5] = SINV(STATE_BYTE(1));
    pOut[6] = SINV(STATE_BYTE(14));
    pOut[7] = SINV(STATE_BYTE(11));
    pOut[8] = SINV(STATE_BYTE(8));
    pOut[9] = SINV(STATE_BYTE(5));
    pOut[10] = SINV(STATE_BYTE(2));
    pOut[11] = SINV(STATE_BYTE(15));
    pOut[12] = SINV(STATE_BYTE(12));
    pOut[13] = SINV(STATE_BYTE(9));
    pOut[14] = SINV(STATE_BYTE(6));
    pOut[15] = SINV(STATE_BYTE(3));
    /* final key addition */
    *((PRUint32 *)(pOut + 12)) ^= *roundkeyw--;
    *((PRUint32 *)(pOut + 8)) ^= *roundkeyw--;
    *((PRUint32 *)(pOut + 4)) ^= *roundkeyw--;
    *((PRUint32 *)pOut) ^= *roundkeyw--;
#if defined(NSS_X86_OR_X64)
#undef pIn
#undef pOut
#else
    if ((ptrdiff_t)output & 0x3) {
        memcpy(output, outBuf, sizeof outBuf);
    }
#endif
    return SECSuccess;
}

/**************************************************************************
 *
 *  Rijndael modes of operation (ECB and CBC)
 *
 *************************************************************************/

static SECStatus
rijndael_encryptECB(AESContext *cx, unsigned char *output,
                    unsigned int *outputLen, unsigned int maxOutputLen,
                    const unsigned char *input, unsigned int inputLen)
{
    AESBlockFunc *encryptor;

    if (aesni_support()) {
        /* Use hardware acceleration for normal AES parameters. */
        encryptor = &native_encryptBlock;
    } else {
        encryptor = &rijndael_encryptBlock128;
    }
    while (inputLen > 0) {
        (*encryptor)(cx, output, input);
        output += AES_BLOCK_SIZE;
        input += AES_BLOCK_SIZE;
        inputLen -= AES_BLOCK_SIZE;
    }
    return SECSuccess;
}

static SECStatus
rijndael_encryptCBC(AESContext *cx, unsigned char *output,
                    unsigned int *outputLen, unsigned int maxOutputLen,
                    const unsigned char *input, unsigned int inputLen)
{
    unsigned int j;
    unsigned char *lastblock;
    unsigned char inblock[AES_BLOCK_SIZE * 8];

    if (!inputLen)
        return SECSuccess;
    lastblock = cx->iv;
    while (inputLen > 0) {
        /* XOR with the last block (IV if first block) */
        for (j = 0; j < AES_BLOCK_SIZE; ++j) {
            inblock[j] = input[j] ^ lastblock[j];
        }
        /* encrypt */
        rijndael_encryptBlock128(cx, output, inblock);
        /* move to the next block */
        lastblock = output;
        output += AES_BLOCK_SIZE;
        input += AES_BLOCK_SIZE;
        inputLen -= AES_BLOCK_SIZE;
    }
    memcpy(cx->iv, lastblock, AES_BLOCK_SIZE);
    return SECSuccess;
}

static SECStatus
rijndael_decryptECB(AESContext *cx, unsigned char *output,
                    unsigned int *outputLen, unsigned int maxOutputLen,
                    const unsigned char *input, unsigned int inputLen)
{
    while (inputLen > 0) {
        if (rijndael_decryptBlock128(cx, output, input) != SECSuccess) {
            return SECFailure;
        }
        output += AES_BLOCK_SIZE;
        input += AES_BLOCK_SIZE;
        inputLen -= AES_BLOCK_SIZE;
    }
    return SECSuccess;
}

static SECStatus
rijndael_decryptCBC(AESContext *cx, unsigned char *output,
                    unsigned int *outputLen, unsigned int maxOutputLen,
                    const unsigned char *input, unsigned int inputLen)
{
    const unsigned char *in;
    unsigned char *out;
    unsigned int j;
    unsigned char newIV[AES_BLOCK_SIZE];

    if (!inputLen)
        return SECSuccess;
    PORT_Assert(output - input >= 0 || input - output >= (int)inputLen);
    in = input + (inputLen - AES_BLOCK_SIZE);
    memcpy(newIV, in, AES_BLOCK_SIZE);
    out = output + (inputLen - AES_BLOCK_SIZE);
    while (inputLen > AES_BLOCK_SIZE) {
        if (rijndael_decryptBlock128(cx, out, in) != SECSuccess) {
            return SECFailure;
        }
        for (j = 0; j < AES_BLOCK_SIZE; ++j)
            out[j] ^= in[(int)(j - AES_BLOCK_SIZE)];
        out -= AES_BLOCK_SIZE;
        in -= AES_BLOCK_SIZE;
        inputLen -= AES_BLOCK_SIZE;
    }
    if (in == input) {
        if (rijndael_decryptBlock128(cx, out, in) != SECSuccess) {
            return SECFailure;
        }
        for (j = 0; j < AES_BLOCK_SIZE; ++j)
            out[j] ^= cx->iv[j];
    }
    memcpy(cx->iv, newIV, AES_BLOCK_SIZE);
    return SECSuccess;
}

/************************************************************************
 *
 * BLAPI Interface functions
 *
 * The following functions implement the encryption routines defined in
 * BLAPI for the AES cipher, Rijndael.
 *
 ***********************************************************************/

AESContext *
AES_AllocateContext(void)
{
    /* aligned_alloc is C11 so we have to do it the old way. */
    AESContext *ctx = PORT_ZAlloc(sizeof(AESContext) + 15);
    if (ctx == NULL) {
        PORT_SetError(SEC_ERROR_NO_MEMORY);
        return NULL;
    }
    ctx->mem = ctx;
    return (AESContext *)(((uintptr_t)ctx + 15) & ~(uintptr_t)0x0F);
}

/*
** Initialize a new AES context suitable for AES encryption/decryption in
** the ECB or CBC mode.
**  "mode" the mode of operation, which must be NSS_AES or NSS_AES_CBC
*/
static SECStatus
aes_InitContext(AESContext *cx, const unsigned char *key, unsigned int keysize,
                const unsigned char *iv, int mode, unsigned int encrypt)
{
    unsigned int Nk;
    PRBool use_hw_aes;
    /* According to AES, block lengths are 128 and key lengths are 128, 192, or
     * 256 bits. We support other key sizes as well [128, 256] as long as the
     * length in bytes is divisible by 4.
     */

    if (key == NULL ||
        keysize < AES_BLOCK_SIZE ||
        keysize > 32 ||
        keysize % 4 != 0) {
        PORT_SetError(SEC_ERROR_INVALID_ARGS);
        return SECFailure;
    }
    if (mode != NSS_AES && mode != NSS_AES_CBC) {
        PORT_SetError(SEC_ERROR_INVALID_ARGS);
        return SECFailure;
    }
    if (mode == NSS_AES_CBC && iv == NULL) {
        PORT_SetError(SEC_ERROR_INVALID_ARGS);
        return SECFailure;
    }
    if (!cx) {
        PORT_SetError(SEC_ERROR_INVALID_ARGS);
        return SECFailure;
    }
    use_hw_aes = aesni_support() && (keysize % 8) == 0;
    /* Nb = (block size in bits) / 32 */
    cx->Nb = AES_BLOCK_SIZE / 4;
    /* Nk = (key size in bits) / 32 */
    Nk = keysize / 4;
    /* Obtain number of rounds from "table" */
    cx->Nr = RIJNDAEL_NUM_ROUNDS(Nk, cx->Nb);
    /* copy in the iv, if neccessary */
    if (mode == NSS_AES_CBC) {
        memcpy(cx->iv, iv, AES_BLOCK_SIZE);
#ifdef USE_HW_AES
        if (use_hw_aes) {
            cx->worker = (freeblCipherFunc)
                intel_aes_cbc_worker(encrypt, keysize);
        } else
#endif
        {
            cx->worker = (freeblCipherFunc)(encrypt
                                                ? &rijndael_encryptCBC
                                                : &rijndael_decryptCBC);
        }
    } else {
#ifdef USE_HW_AES
        if (use_hw_aes) {
            cx->worker = (freeblCipherFunc)
                intel_aes_ecb_worker(encrypt, keysize);
        } else
#endif
        {
            cx->worker = (freeblCipherFunc)(encrypt
                                                ? &rijndael_encryptECB
                                                : &rijndael_decryptECB);
        }
    }
    PORT_Assert((cx->Nb * (cx->Nr + 1)) <= RIJNDAEL_MAX_EXP_KEY_SIZE);
    if ((cx->Nb * (cx->Nr + 1)) > RIJNDAEL_MAX_EXP_KEY_SIZE) {
        PORT_SetError(SEC_ERROR_LIBRARY_FAILURE);
        return SECFailure;
    }
#ifdef USE_HW_AES
    if (use_hw_aes) {
        intel_aes_init(encrypt, keysize);
    } else
#endif
    {

#if defined(RIJNDAEL_GENERATE_TABLES) || \
    defined(RIJNDAEL_GENERATE_TABLES_MACRO)
        if (rijndaelTables == NULL) {
            if (PR_CallOnce(&coRTInit, init_rijndael_tables) != PR_SUCCESS) {
                return SECFailure;
            }
        }
#endif
        /* Generate expanded key */
        if (encrypt) {
            if (use_hw_aes && (cx->mode == NSS_AES_GCM || cx->mode == NSS_AES ||
                               cx->mode == NSS_AES_CTR)) {
                PORT_Assert(keysize == 16 || keysize == 24 || keysize == 32);
                /* Prepare hardware key for normal AES parameters. */
                native_key_expansion(cx, key, Nk);
            } else {
                rijndael_key_expansion(cx, key, Nk);
            }
        } else {
            rijndael_invkey_expansion(cx, key, Nk);
        }
    }
    cx->worker_cx = cx;
    cx->destroy = NULL;
    cx->isBlock = PR_TRUE;
    return SECSuccess;
}

SECStatus
AES_InitContext(AESContext *cx, const unsigned char *key, unsigned int keysize,
                const unsigned char *iv, int mode, unsigned int encrypt,
                unsigned int blocksize)
{
    int basemode = mode;
    PRBool baseencrypt = encrypt;
    SECStatus rv;

    if (blocksize != AES_BLOCK_SIZE) {
        PORT_SetError(SEC_ERROR_INVALID_ARGS);
        return SECFailure;
    }

    switch (mode) {
        case NSS_AES_CTS:
            basemode = NSS_AES_CBC;
            break;
        case NSS_AES_GCM:
        case NSS_AES_CTR:
            basemode = NSS_AES;
            baseencrypt = PR_TRUE;
            break;
    }
    /* Make sure enough is initialized so we can safely call Destroy. */
    cx->worker_cx = NULL;
    cx->destroy = NULL;
    cx->mode = mode;
    rv = aes_InitContext(cx, key, keysize, iv, basemode, baseencrypt);
    if (rv != SECSuccess) {
        AES_DestroyContext(cx, PR_FALSE);
        return rv;
    }

    /* finally, set up any mode specific contexts */
    switch (mode) {
        case NSS_AES_CTS:
            cx->worker_cx = CTS_CreateContext(cx, cx->worker, iv);
            cx->worker = (freeblCipherFunc)(encrypt ? CTS_EncryptUpdate : CTS_DecryptUpdate);
            cx->destroy = (freeblDestroyFunc)CTS_DestroyContext;
            cx->isBlock = PR_FALSE;
            break;
        case NSS_AES_GCM:
#if defined(INTEL_GCM) && defined(USE_HW_AES)
            if (aesni_support() && (keysize % 8) == 0 && avx_support() &&
                clmul_support()) {
                cx->worker_cx = intel_AES_GCM_CreateContext(cx, cx->worker, iv);
                cx->worker = (freeblCipherFunc)(encrypt ? intel_AES_GCM_EncryptUpdate
                                                        : intel_AES_GCM_DecryptUpdate);
                cx->destroy = (freeblDestroyFunc)intel_AES_GCM_DestroyContext;
                cx->isBlock = PR_FALSE;
            } else
#endif
            {
                cx->worker_cx = GCM_CreateContext(cx, cx->worker, iv);
                cx->worker = (freeblCipherFunc)(encrypt ? GCM_EncryptUpdate
                                                        : GCM_DecryptUpdate);
                cx->destroy = (freeblDestroyFunc)GCM_DestroyContext;
                cx->isBlock = PR_FALSE;
            }
            break;
        case NSS_AES_CTR:
            cx->worker_cx = CTR_CreateContext(cx, cx->worker, iv);
#if defined(USE_HW_AES) && defined(_MSC_VER)
            if (aesni_support() && (keysize % 8) == 0) {
                cx->worker = (freeblCipherFunc)CTR_Update_HW_AES;
            } else
#endif
            {
                cx->worker = (freeblCipherFunc)CTR_Update;
            }
            cx->destroy = (freeblDestroyFunc)CTR_DestroyContext;
            cx->isBlock = PR_FALSE;
            break;
        default:
            /* everything has already been set up by aes_InitContext, just
             * return */
            return SECSuccess;
    }
    /* check to see if we succeeded in getting the worker context */
    if (cx->worker_cx == NULL) {
        /* no, just destroy the existing context */
        cx->destroy = NULL; /* paranoia, though you can see a dozen lines */
                            /* below that this isn't necessary */
        AES_DestroyContext(cx, PR_FALSE);
        return SECFailure;
    }
    return SECSuccess;
}

/* AES_CreateContext
 *
 * create a new context for Rijndael operations
 */
AESContext *
AES_CreateContext(const unsigned char *key, const unsigned char *iv,
                  int mode, int encrypt,
                  unsigned int keysize, unsigned int blocksize)
{
    AESContext *cx = AES_AllocateContext();
    if (cx) {
        SECStatus rv = AES_InitContext(cx, key, keysize, iv, mode, encrypt,
                                       blocksize);
        if (rv != SECSuccess) {
            AES_DestroyContext(cx, PR_TRUE);
            cx = NULL;
        }
    }
    return cx;
}

/*
 * AES_DestroyContext
 *
 * Zero an AES cipher context.  If freeit is true, also free the pointer
 * to the context.
 */
void
AES_DestroyContext(AESContext *cx, PRBool freeit)
{
    if (cx->worker_cx && cx->destroy) {
        (*cx->destroy)(cx->worker_cx, PR_TRUE);
        cx->worker_cx = NULL;
        cx->destroy = NULL;
    }
    if (freeit) {
        PORT_Free(cx->mem);
    }
}

/*
 * AES_Encrypt
 *
 * Encrypt an arbitrary-length buffer.  The output buffer must already be
 * allocated to at least inputLen.
 */
SECStatus
AES_Encrypt(AESContext *cx, unsigned char *output,
            unsigned int *outputLen, unsigned int maxOutputLen,
            const unsigned char *input, unsigned int inputLen)
{
    /* Check args */
    if (cx == NULL || output == NULL || (input == NULL && inputLen != 0)) {
        PORT_SetError(SEC_ERROR_INVALID_ARGS);
        return SECFailure;
    }
    if (cx->isBlock && (inputLen % AES_BLOCK_SIZE != 0)) {
        PORT_SetError(SEC_ERROR_INPUT_LEN);
        return SECFailure;
    }
    if (maxOutputLen < inputLen) {
        PORT_SetError(SEC_ERROR_OUTPUT_LEN);
        return SECFailure;
    }
    *outputLen = inputLen;
#if UINT_MAX > MP_32BIT_MAX
    /*
     * we can guarentee that GSM won't overlfow if we limit the input to
     * 2^36 bytes. For simplicity, we are limiting it to 2^32 for now.
     *
     * We do it here to cover both hardware and software GCM operations.
     */
    {
        PR_STATIC_ASSERT(sizeof(unsigned int) > 4);
    }
    if ((cx->mode == NSS_AES_GCM) && (inputLen > MP_32BIT_MAX)) {
        PORT_SetError(SEC_ERROR_OUTPUT_LEN);
        return SECFailure;
    }
#else
    /* if we can't pass in a 32_bit number, then no such check needed */
    {
        PR_STATIC_ASSERT(sizeof(unsigned int) <= 4);
    }
#endif

    return (*cx->worker)(cx->worker_cx, output, outputLen, maxOutputLen,
                         input, inputLen, AES_BLOCK_SIZE);
}

/*
 * AES_Decrypt
 *
 * Decrypt and arbitrary-length buffer.  The output buffer must already be
 * allocated to at least inputLen.
 */
SECStatus
AES_Decrypt(AESContext *cx, unsigned char *output,
            unsigned int *outputLen, unsigned int maxOutputLen,
            const unsigned char *input, unsigned int inputLen)
{
    /* Check args */
    if (cx == NULL || output == NULL || (input == NULL && inputLen != 0)) {
        PORT_SetError(SEC_ERROR_INVALID_ARGS);
        return SECFailure;
    }
    if (cx->isBlock && (inputLen % AES_BLOCK_SIZE != 0)) {
        PORT_SetError(SEC_ERROR_INPUT_LEN);
        return SECFailure;
    }
    if (maxOutputLen < inputLen) {
        PORT_SetError(SEC_ERROR_OUTPUT_LEN);
        return SECFailure;
    }
    *outputLen = inputLen;
    return (*cx->worker)(cx->worker_cx, output, outputLen, maxOutputLen,
                         input, inputLen, AES_BLOCK_SIZE);
}