summaryrefslogtreecommitdiffstats
path: root/arch/x86/crypto/sm4-aesni-avx2-asm_64.S
blob: 4732fe8bb65b6579e2ce8a4489dd7df8e89be74c (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
// SPDX-License-Identifier: GPL-2.0-or-later
/*
 * SM4 Cipher Algorithm, AES-NI/AVX2 optimized.
 * as specified in
 * https://tools.ietf.org/id/draft-ribose-cfrg-sm4-10.html
 *
 * Copyright (C) 2018 Markku-Juhani O. Saarinen <mjos@iki.fi>
 * Copyright (C) 2020 Jussi Kivilinna <jussi.kivilinna@iki.fi>
 * Copyright (c) 2021 Tianjia Zhang <tianjia.zhang@linux.alibaba.com>
 */

/* Based on SM4 AES-NI work by libgcrypt and Markku-Juhani O. Saarinen at:
 *  https://github.com/mjosaarinen/sm4ni
 */

#include <linux/linkage.h>
#include <asm/frame.h>

#define rRIP         (%rip)

/* vector registers */
#define RX0          %ymm0
#define RX1          %ymm1
#define MASK_4BIT    %ymm2
#define RTMP0        %ymm3
#define RTMP1        %ymm4
#define RTMP2        %ymm5
#define RTMP3        %ymm6
#define RTMP4        %ymm7

#define RA0          %ymm8
#define RA1          %ymm9
#define RA2          %ymm10
#define RA3          %ymm11

#define RB0          %ymm12
#define RB1          %ymm13
#define RB2          %ymm14
#define RB3          %ymm15

#define RNOT         %ymm0
#define RBSWAP       %ymm1

#define RX0x         %xmm0
#define RX1x         %xmm1
#define MASK_4BITx   %xmm2

#define RNOTx        %xmm0
#define RBSWAPx      %xmm1

#define RTMP0x       %xmm3
#define RTMP1x       %xmm4
#define RTMP2x       %xmm5
#define RTMP3x       %xmm6
#define RTMP4x       %xmm7


/* helper macros */

/* Transpose four 32-bit words between 128-bit vector lanes. */
#define transpose_4x4(x0, x1, x2, x3, t1, t2) \
	vpunpckhdq x1, x0, t2;                \
	vpunpckldq x1, x0, x0;                \
	                                      \
	vpunpckldq x3, x2, t1;                \
	vpunpckhdq x3, x2, x2;                \
	                                      \
	vpunpckhqdq t1, x0, x1;               \
	vpunpcklqdq t1, x0, x0;               \
	                                      \
	vpunpckhqdq x2, t2, x3;               \
	vpunpcklqdq x2, t2, x2;

/* post-SubByte transform. */
#define transform_pre(x, lo_t, hi_t, mask4bit, tmp0) \
	vpand x, mask4bit, tmp0;                     \
	vpandn x, mask4bit, x;                       \
	vpsrld $4, x, x;                             \
	                                             \
	vpshufb tmp0, lo_t, tmp0;                    \
	vpshufb x, hi_t, x;                          \
	vpxor tmp0, x, x;

/* post-SubByte transform. Note: x has been XOR'ed with mask4bit by
 * 'vaeslastenc' instruction. */
#define transform_post(x, lo_t, hi_t, mask4bit, tmp0) \
	vpandn mask4bit, x, tmp0;                     \
	vpsrld $4, x, x;                              \
	vpand x, mask4bit, x;                         \
	                                              \
	vpshufb tmp0, lo_t, tmp0;                     \
	vpshufb x, hi_t, x;                           \
	vpxor tmp0, x, x;


.section	.rodata.cst16, "aM", @progbits, 16
.align 16

/*
 * Following four affine transform look-up tables are from work by
 * Markku-Juhani O. Saarinen, at https://github.com/mjosaarinen/sm4ni
 *
 * These allow exposing SM4 S-Box from AES SubByte.
 */

/* pre-SubByte affine transform, from SM4 field to AES field. */
.Lpre_tf_lo_s:
	.quad 0x9197E2E474720701, 0xC7C1B4B222245157
.Lpre_tf_hi_s:
	.quad 0xE240AB09EB49A200, 0xF052B91BF95BB012

/* post-SubByte affine transform, from AES field to SM4 field. */
.Lpost_tf_lo_s:
	.quad 0x5B67F2CEA19D0834, 0xEDD14478172BBE82
.Lpost_tf_hi_s:
	.quad 0xAE7201DD73AFDC00, 0x11CDBE62CC1063BF

/* For isolating SubBytes from AESENCLAST, inverse shift row */
.Linv_shift_row:
	.byte 0x00, 0x0d, 0x0a, 0x07, 0x04, 0x01, 0x0e, 0x0b
	.byte 0x08, 0x05, 0x02, 0x0f, 0x0c, 0x09, 0x06, 0x03

/* Inverse shift row + Rotate left by 8 bits on 32-bit words with vpshufb */
.Linv_shift_row_rol_8:
	.byte 0x07, 0x00, 0x0d, 0x0a, 0x0b, 0x04, 0x01, 0x0e
	.byte 0x0f, 0x08, 0x05, 0x02, 0x03, 0x0c, 0x09, 0x06

/* Inverse shift row + Rotate left by 16 bits on 32-bit words with vpshufb */
.Linv_shift_row_rol_16:
	.byte 0x0a, 0x07, 0x00, 0x0d, 0x0e, 0x0b, 0x04, 0x01
	.byte 0x02, 0x0f, 0x08, 0x05, 0x06, 0x03, 0x0c, 0x09

/* Inverse shift row + Rotate left by 24 bits on 32-bit words with vpshufb */
.Linv_shift_row_rol_24:
	.byte 0x0d, 0x0a, 0x07, 0x00, 0x01, 0x0e, 0x0b, 0x04
	.byte 0x05, 0x02, 0x0f, 0x08, 0x09, 0x06, 0x03, 0x0c

/* For CTR-mode IV byteswap */
.Lbswap128_mask:
	.byte 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0

/* For input word byte-swap */
.Lbswap32_mask:
	.byte 3, 2, 1, 0, 7, 6, 5, 4, 11, 10, 9, 8, 15, 14, 13, 12

.align 4
/* 4-bit mask */
.L0f0f0f0f:
	.long 0x0f0f0f0f

/* 12 bytes, only for padding */
.Lpadding_deadbeef:
	.long 0xdeadbeef, 0xdeadbeef, 0xdeadbeef

.text
.align 16

.align 8
SYM_FUNC_START_LOCAL(__sm4_crypt_blk16)
	/* input:
	 *	%rdi: round key array, CTX
	 *	RA0, RA1, RA2, RA3, RB0, RB1, RB2, RB3: sixteen parallel
	 *						plaintext blocks
	 * output:
	 *	RA0, RA1, RA2, RA3, RB0, RB1, RB2, RB3: sixteen parallel
	 * 						ciphertext blocks
	 */
	FRAME_BEGIN

	vbroadcasti128 .Lbswap32_mask rRIP, RTMP2;
	vpshufb RTMP2, RA0, RA0;
	vpshufb RTMP2, RA1, RA1;
	vpshufb RTMP2, RA2, RA2;
	vpshufb RTMP2, RA3, RA3;
	vpshufb RTMP2, RB0, RB0;
	vpshufb RTMP2, RB1, RB1;
	vpshufb RTMP2, RB2, RB2;
	vpshufb RTMP2, RB3, RB3;

	vpbroadcastd .L0f0f0f0f rRIP, MASK_4BIT;
	transpose_4x4(RA0, RA1, RA2, RA3, RTMP0, RTMP1);
	transpose_4x4(RB0, RB1, RB2, RB3, RTMP0, RTMP1);

#define ROUND(round, s0, s1, s2, s3, r0, r1, r2, r3)                \
	vpbroadcastd (4*(round))(%rdi), RX0;                        \
	vbroadcasti128 .Lpre_tf_lo_s rRIP, RTMP4;                   \
	vbroadcasti128 .Lpre_tf_hi_s rRIP, RTMP1;                   \
	vmovdqa RX0, RX1;                                           \
	vpxor s1, RX0, RX0;                                         \
	vpxor s2, RX0, RX0;                                         \
	vpxor s3, RX0, RX0; /* s1 ^ s2 ^ s3 ^ rk */                 \
	vbroadcasti128 .Lpost_tf_lo_s rRIP, RTMP2;                  \
	vbroadcasti128 .Lpost_tf_hi_s rRIP, RTMP3;                  \
	vpxor r1, RX1, RX1;                                         \
	vpxor r2, RX1, RX1;                                         \
	vpxor r3, RX1, RX1; /* r1 ^ r2 ^ r3 ^ rk */                 \
	                                                            \
	/* sbox, non-linear part */                                 \
	transform_pre(RX0, RTMP4, RTMP1, MASK_4BIT, RTMP0);         \
	transform_pre(RX1, RTMP4, RTMP1, MASK_4BIT, RTMP0);         \
	vextracti128 $1, RX0, RTMP4x;                               \
	vextracti128 $1, RX1, RTMP0x;                               \
	vaesenclast MASK_4BITx, RX0x, RX0x;                         \
	vaesenclast MASK_4BITx, RTMP4x, RTMP4x;                     \
	vaesenclast MASK_4BITx, RX1x, RX1x;                         \
	vaesenclast MASK_4BITx, RTMP0x, RTMP0x;                     \
	vinserti128 $1, RTMP4x, RX0, RX0;                           \
	vbroadcasti128 .Linv_shift_row rRIP, RTMP4;                 \
	vinserti128 $1, RTMP0x, RX1, RX1;                           \
	transform_post(RX0, RTMP2, RTMP3, MASK_4BIT, RTMP0);        \
	transform_post(RX1, RTMP2, RTMP3, MASK_4BIT, RTMP0);        \
	                                                            \
	/* linear part */                                           \
	vpshufb RTMP4, RX0, RTMP0;                                  \
	vpxor RTMP0, s0, s0; /* s0 ^ x */                           \
	vpshufb RTMP4, RX1, RTMP2;                                  \
	vbroadcasti128 .Linv_shift_row_rol_8 rRIP, RTMP4;           \
	vpxor RTMP2, r0, r0; /* r0 ^ x */                           \
	vpshufb RTMP4, RX0, RTMP1;                                  \
	vpxor RTMP1, RTMP0, RTMP0; /* x ^ rol(x,8) */               \
	vpshufb RTMP4, RX1, RTMP3;                                  \
	vbroadcasti128 .Linv_shift_row_rol_16 rRIP, RTMP4;          \
	vpxor RTMP3, RTMP2, RTMP2; /* x ^ rol(x,8) */               \
	vpshufb RTMP4, RX0, RTMP1;                                  \
	vpxor RTMP1, RTMP0, RTMP0; /* x ^ rol(x,8) ^ rol(x,16) */   \
	vpshufb RTMP4, RX1, RTMP3;                                  \
	vbroadcasti128 .Linv_shift_row_rol_24 rRIP, RTMP4;          \
	vpxor RTMP3, RTMP2, RTMP2; /* x ^ rol(x,8) ^ rol(x,16) */   \
	vpshufb RTMP4, RX0, RTMP1;                                  \
	vpxor RTMP1, s0, s0; /* s0 ^ x ^ rol(x,24) */               \
	vpslld $2, RTMP0, RTMP1;                                    \
	vpsrld $30, RTMP0, RTMP0;                                   \
	vpxor RTMP0, s0, s0;                                        \
	/* s0 ^ x ^ rol(x,2) ^ rol(x,10) ^ rol(x,18) ^ rol(x,24) */ \
	vpxor RTMP1, s0, s0;                                        \
	vpshufb RTMP4, RX1, RTMP3;                                  \
	vpxor RTMP3, r0, r0; /* r0 ^ x ^ rol(x,24) */               \
	vpslld $2, RTMP2, RTMP3;                                    \
	vpsrld $30, RTMP2, RTMP2;                                   \
	vpxor RTMP2, r0, r0;                                        \
	/* r0 ^ x ^ rol(x,2) ^ rol(x,10) ^ rol(x,18) ^ rol(x,24) */ \
	vpxor RTMP3, r0, r0;

	leaq (32*4)(%rdi), %rax;
.align 16
.Lroundloop_blk8:
	ROUND(0, RA0, RA1, RA2, RA3, RB0, RB1, RB2, RB3);
	ROUND(1, RA1, RA2, RA3, RA0, RB1, RB2, RB3, RB0);
	ROUND(2, RA2, RA3, RA0, RA1, RB2, RB3, RB0, RB1);
	ROUND(3, RA3, RA0, RA1, RA2, RB3, RB0, RB1, RB2);
	leaq (4*4)(%rdi), %rdi;
	cmpq %rax, %rdi;
	jne .Lroundloop_blk8;

#undef ROUND

	vbroadcasti128 .Lbswap128_mask rRIP, RTMP2;

	transpose_4x4(RA0, RA1, RA2, RA3, RTMP0, RTMP1);
	transpose_4x4(RB0, RB1, RB2, RB3, RTMP0, RTMP1);
	vpshufb RTMP2, RA0, RA0;
	vpshufb RTMP2, RA1, RA1;
	vpshufb RTMP2, RA2, RA2;
	vpshufb RTMP2, RA3, RA3;
	vpshufb RTMP2, RB0, RB0;
	vpshufb RTMP2, RB1, RB1;
	vpshufb RTMP2, RB2, RB2;
	vpshufb RTMP2, RB3, RB3;

	FRAME_END
	RET;
SYM_FUNC_END(__sm4_crypt_blk16)

#define inc_le128(x, minus_one, tmp) \
	vpcmpeqq minus_one, x, tmp;  \
	vpsubq minus_one, x, x;      \
	vpslldq $8, tmp, tmp;        \
	vpsubq tmp, x, x;

/*
 * void sm4_aesni_avx2_ctr_enc_blk16(const u32 *rk, u8 *dst,
 *                                   const u8 *src, u8 *iv)
 */
.align 8
SYM_FUNC_START(sm4_aesni_avx2_ctr_enc_blk16)
	/* input:
	 *	%rdi: round key array, CTX
	 *	%rsi: dst (16 blocks)
	 *	%rdx: src (16 blocks)
	 *	%rcx: iv (big endian, 128bit)
	 */
	FRAME_BEGIN

	movq 8(%rcx), %rax;
	bswapq %rax;

	vzeroupper;

	vbroadcasti128 .Lbswap128_mask rRIP, RTMP3;
	vpcmpeqd RNOT, RNOT, RNOT;
	vpsrldq $8, RNOT, RNOT;   /* ab: -1:0 ; cd: -1:0 */
	vpaddq RNOT, RNOT, RTMP2; /* ab: -2:0 ; cd: -2:0 */

	/* load IV and byteswap */
	vmovdqu (%rcx), RTMP4x;
	vpshufb RTMP3x, RTMP4x, RTMP4x;
	vmovdqa RTMP4x, RTMP0x;
	inc_le128(RTMP4x, RNOTx, RTMP1x);
	vinserti128 $1, RTMP4x, RTMP0, RTMP0;
	vpshufb RTMP3, RTMP0, RA0; /* +1 ; +0 */

	/* check need for handling 64-bit overflow and carry */
	cmpq $(0xffffffffffffffff - 16), %rax;
	ja .Lhandle_ctr_carry;

	/* construct IVs */
	vpsubq RTMP2, RTMP0, RTMP0; /* +3 ; +2 */
	vpshufb RTMP3, RTMP0, RA1;
	vpsubq RTMP2, RTMP0, RTMP0; /* +5 ; +4 */
	vpshufb RTMP3, RTMP0, RA2;
	vpsubq RTMP2, RTMP0, RTMP0; /* +7 ; +6 */
	vpshufb RTMP3, RTMP0, RA3;
	vpsubq RTMP2, RTMP0, RTMP0; /* +9 ; +8 */
	vpshufb RTMP3, RTMP0, RB0;
	vpsubq RTMP2, RTMP0, RTMP0; /* +11 ; +10 */
	vpshufb RTMP3, RTMP0, RB1;
	vpsubq RTMP2, RTMP0, RTMP0; /* +13 ; +12 */
	vpshufb RTMP3, RTMP0, RB2;
	vpsubq RTMP2, RTMP0, RTMP0; /* +15 ; +14 */
	vpshufb RTMP3, RTMP0, RB3;
	vpsubq RTMP2, RTMP0, RTMP0; /* +16 */
	vpshufb RTMP3x, RTMP0x, RTMP0x;

	jmp .Lctr_carry_done;

.Lhandle_ctr_carry:
	/* construct IVs */
	inc_le128(RTMP0, RNOT, RTMP1);
	inc_le128(RTMP0, RNOT, RTMP1);
	vpshufb RTMP3, RTMP0, RA1; /* +3 ; +2 */
	inc_le128(RTMP0, RNOT, RTMP1);
	inc_le128(RTMP0, RNOT, RTMP1);
	vpshufb RTMP3, RTMP0, RA2; /* +5 ; +4 */
	inc_le128(RTMP0, RNOT, RTMP1);
	inc_le128(RTMP0, RNOT, RTMP1);
	vpshufb RTMP3, RTMP0, RA3; /* +7 ; +6 */
	inc_le128(RTMP0, RNOT, RTMP1);
	inc_le128(RTMP0, RNOT, RTMP1);
	vpshufb RTMP3, RTMP0, RB0; /* +9 ; +8 */
	inc_le128(RTMP0, RNOT, RTMP1);
	inc_le128(RTMP0, RNOT, RTMP1);
	vpshufb RTMP3, RTMP0, RB1; /* +11 ; +10 */
	inc_le128(RTMP0, RNOT, RTMP1);
	inc_le128(RTMP0, RNOT, RTMP1);
	vpshufb RTMP3, RTMP0, RB2; /* +13 ; +12 */
	inc_le128(RTMP0, RNOT, RTMP1);
	inc_le128(RTMP0, RNOT, RTMP1);
	vpshufb RTMP3, RTMP0, RB3; /* +15 ; +14 */
	inc_le128(RTMP0, RNOT, RTMP1);
	vextracti128 $1, RTMP0, RTMP0x;
	vpshufb RTMP3x, RTMP0x, RTMP0x; /* +16 */

.align 4
.Lctr_carry_done:
	/* store new IV */
	vmovdqu RTMP0x, (%rcx);

	call __sm4_crypt_blk16;

	vpxor (0 * 32)(%rdx), RA0, RA0;
	vpxor (1 * 32)(%rdx), RA1, RA1;
	vpxor (2 * 32)(%rdx), RA2, RA2;
	vpxor (3 * 32)(%rdx), RA3, RA3;
	vpxor (4 * 32)(%rdx), RB0, RB0;
	vpxor (5 * 32)(%rdx), RB1, RB1;
	vpxor (6 * 32)(%rdx), RB2, RB2;
	vpxor (7 * 32)(%rdx), RB3, RB3;

	vmovdqu RA0, (0 * 32)(%rsi);
	vmovdqu RA1, (1 * 32)(%rsi);
	vmovdqu RA2, (2 * 32)(%rsi);
	vmovdqu RA3, (3 * 32)(%rsi);
	vmovdqu RB0, (4 * 32)(%rsi);
	vmovdqu RB1, (5 * 32)(%rsi);
	vmovdqu RB2, (6 * 32)(%rsi);
	vmovdqu RB3, (7 * 32)(%rsi);

	vzeroall;
	FRAME_END
	RET;
SYM_FUNC_END(sm4_aesni_avx2_ctr_enc_blk16)

/*
 * void sm4_aesni_avx2_cbc_dec_blk16(const u32 *rk, u8 *dst,
 *                                   const u8 *src, u8 *iv)
 */
.align 8
SYM_FUNC_START(sm4_aesni_avx2_cbc_dec_blk16)
	/* input:
	 *	%rdi: round key array, CTX
	 *	%rsi: dst (16 blocks)
	 *	%rdx: src (16 blocks)
	 *	%rcx: iv
	 */
	FRAME_BEGIN

	vzeroupper;

	vmovdqu (0 * 32)(%rdx), RA0;
	vmovdqu (1 * 32)(%rdx), RA1;
	vmovdqu (2 * 32)(%rdx), RA2;
	vmovdqu (3 * 32)(%rdx), RA3;
	vmovdqu (4 * 32)(%rdx), RB0;
	vmovdqu (5 * 32)(%rdx), RB1;
	vmovdqu (6 * 32)(%rdx), RB2;
	vmovdqu (7 * 32)(%rdx), RB3;

	call __sm4_crypt_blk16;

	vmovdqu (%rcx), RNOTx;
	vinserti128 $1, (%rdx), RNOT, RNOT;
	vpxor RNOT, RA0, RA0;
	vpxor (0 * 32 + 16)(%rdx), RA1, RA1;
	vpxor (1 * 32 + 16)(%rdx), RA2, RA2;
	vpxor (2 * 32 + 16)(%rdx), RA3, RA3;
	vpxor (3 * 32 + 16)(%rdx), RB0, RB0;
	vpxor (4 * 32 + 16)(%rdx), RB1, RB1;
	vpxor (5 * 32 + 16)(%rdx), RB2, RB2;
	vpxor (6 * 32 + 16)(%rdx), RB3, RB3;
	vmovdqu (7 * 32 + 16)(%rdx), RNOTx;
	vmovdqu RNOTx, (%rcx); /* store new IV */

	vmovdqu RA0, (0 * 32)(%rsi);
	vmovdqu RA1, (1 * 32)(%rsi);
	vmovdqu RA2, (2 * 32)(%rsi);
	vmovdqu RA3, (3 * 32)(%rsi);
	vmovdqu RB0, (4 * 32)(%rsi);
	vmovdqu RB1, (5 * 32)(%rsi);
	vmovdqu RB2, (6 * 32)(%rsi);
	vmovdqu RB3, (7 * 32)(%rsi);

	vzeroall;
	FRAME_END
	RET;
SYM_FUNC_END(sm4_aesni_avx2_cbc_dec_blk16)

/*
 * void sm4_aesni_avx2_cfb_dec_blk16(const u32 *rk, u8 *dst,
 *                                   const u8 *src, u8 *iv)
 */
.align 8
SYM_FUNC_START(sm4_aesni_avx2_cfb_dec_blk16)
	/* input:
	 *	%rdi: round key array, CTX
	 *	%rsi: dst (16 blocks)
	 *	%rdx: src (16 blocks)
	 *	%rcx: iv
	 */
	FRAME_BEGIN

	vzeroupper;

	/* Load input */
	vmovdqu (%rcx), RNOTx;
	vinserti128 $1, (%rdx), RNOT, RA0;
	vmovdqu (0 * 32 + 16)(%rdx), RA1;
	vmovdqu (1 * 32 + 16)(%rdx), RA2;
	vmovdqu (2 * 32 + 16)(%rdx), RA3;
	vmovdqu (3 * 32 + 16)(%rdx), RB0;
	vmovdqu (4 * 32 + 16)(%rdx), RB1;
	vmovdqu (5 * 32 + 16)(%rdx), RB2;
	vmovdqu (6 * 32 + 16)(%rdx), RB3;

	/* Update IV */
	vmovdqu (7 * 32 + 16)(%rdx), RNOTx;
	vmovdqu RNOTx, (%rcx);

	call __sm4_crypt_blk16;

	vpxor (0 * 32)(%rdx), RA0, RA0;
	vpxor (1 * 32)(%rdx), RA1, RA1;
	vpxor (2 * 32)(%rdx), RA2, RA2;
	vpxor (3 * 32)(%rdx), RA3, RA3;
	vpxor (4 * 32)(%rdx), RB0, RB0;
	vpxor (5 * 32)(%rdx), RB1, RB1;
	vpxor (6 * 32)(%rdx), RB2, RB2;
	vpxor (7 * 32)(%rdx), RB3, RB3;

	vmovdqu RA0, (0 * 32)(%rsi);
	vmovdqu RA1, (1 * 32)(%rsi);
	vmovdqu RA2, (2 * 32)(%rsi);
	vmovdqu RA3, (3 * 32)(%rsi);
	vmovdqu RB0, (4 * 32)(%rsi);
	vmovdqu RB1, (5 * 32)(%rsi);
	vmovdqu RB2, (6 * 32)(%rsi);
	vmovdqu RB3, (7 * 32)(%rsi);

	vzeroall;
	FRAME_END
	RET;
SYM_FUNC_END(sm4_aesni_avx2_cfb_dec_blk16)