aboutsummaryrefslogtreecommitdiffstats
path: root/contrib/libs/openssl/asm/aarch64/crypto/modes/ghashv8-armx.S
blob: 17ba12c5072edb798fb9dbe98e55e22aa94dc025 (plain) (blame)
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
#include "arm_arch.h"

#if __ARM_MAX_ARCH__>=7
.text
.arch	armv8-a+crypto
.globl	gcm_init_v8
.type	gcm_init_v8,%function
.align	4
gcm_init_v8:
	ld1	{v17.2d},[x1]		//load input H
	movi	v19.16b,#0xe1
	shl	v19.2d,v19.2d,#57		//0xc2.0
	ext	v3.16b,v17.16b,v17.16b,#8
	ushr	v18.2d,v19.2d,#63
	dup	v17.4s,v17.s[1]
	ext	v16.16b,v18.16b,v19.16b,#8		//t0=0xc2....01
	ushr	v18.2d,v3.2d,#63
	sshr	v17.4s,v17.4s,#31		//broadcast carry bit
	and	v18.16b,v18.16b,v16.16b
	shl	v3.2d,v3.2d,#1
	ext	v18.16b,v18.16b,v18.16b,#8
	and	v16.16b,v16.16b,v17.16b
	orr	v3.16b,v3.16b,v18.16b		//H<<<=1
	eor	v20.16b,v3.16b,v16.16b		//twisted H
	st1	{v20.2d},[x0],#16		//store Htable[0]

	//calculate H^2
	ext	v16.16b,v20.16b,v20.16b,#8		//Karatsuba pre-processing
	pmull	v0.1q,v20.1d,v20.1d
	eor	v16.16b,v16.16b,v20.16b
	pmull2	v2.1q,v20.2d,v20.2d
	pmull	v1.1q,v16.1d,v16.1d

	ext	v17.16b,v0.16b,v2.16b,#8		//Karatsuba post-processing
	eor	v18.16b,v0.16b,v2.16b
	eor	v1.16b,v1.16b,v17.16b
	eor	v1.16b,v1.16b,v18.16b
	pmull	v18.1q,v0.1d,v19.1d		//1st phase

	ins	v2.d[0],v1.d[1]
	ins	v1.d[1],v0.d[0]
	eor	v0.16b,v1.16b,v18.16b

	ext	v18.16b,v0.16b,v0.16b,#8		//2nd phase
	pmull	v0.1q,v0.1d,v19.1d
	eor	v18.16b,v18.16b,v2.16b
	eor	v22.16b,v0.16b,v18.16b

	ext	v17.16b,v22.16b,v22.16b,#8		//Karatsuba pre-processing
	eor	v17.16b,v17.16b,v22.16b
	ext	v21.16b,v16.16b,v17.16b,#8		//pack Karatsuba pre-processed
	st1	{v21.2d,v22.2d},[x0],#32	//store Htable[1..2]
	//calculate H^3 and H^4
	pmull	v0.1q,v20.1d, v22.1d
	pmull	v5.1q,v22.1d,v22.1d
	pmull2	v2.1q,v20.2d, v22.2d
	pmull2	v7.1q,v22.2d,v22.2d
	pmull	v1.1q,v16.1d,v17.1d
	pmull	v6.1q,v17.1d,v17.1d

	ext	v16.16b,v0.16b,v2.16b,#8		//Karatsuba post-processing
	ext	v17.16b,v5.16b,v7.16b,#8
	eor	v18.16b,v0.16b,v2.16b
	eor	v1.16b,v1.16b,v16.16b
	eor	v4.16b,v5.16b,v7.16b
	eor	v6.16b,v6.16b,v17.16b
	eor	v1.16b,v1.16b,v18.16b
	pmull	v18.1q,v0.1d,v19.1d		//1st phase
	eor	v6.16b,v6.16b,v4.16b
	pmull	v4.1q,v5.1d,v19.1d

	ins	v2.d[0],v1.d[1]
	ins	v7.d[0],v6.d[1]
	ins	v1.d[1],v0.d[0]
	ins	v6.d[1],v5.d[0]
	eor	v0.16b,v1.16b,v18.16b
	eor	v5.16b,v6.16b,v4.16b

	ext	v18.16b,v0.16b,v0.16b,#8		//2nd phase
	ext	v4.16b,v5.16b,v5.16b,#8
	pmull	v0.1q,v0.1d,v19.1d
	pmull	v5.1q,v5.1d,v19.1d
	eor	v18.16b,v18.16b,v2.16b
	eor	v4.16b,v4.16b,v7.16b
	eor	v20.16b, v0.16b,v18.16b		//H^3
	eor	v22.16b,v5.16b,v4.16b		//H^4

	ext	v16.16b,v20.16b, v20.16b,#8		//Karatsuba pre-processing
	ext	v17.16b,v22.16b,v22.16b,#8
	eor	v16.16b,v16.16b,v20.16b
	eor	v17.16b,v17.16b,v22.16b
	ext	v21.16b,v16.16b,v17.16b,#8		//pack Karatsuba pre-processed
	st1	{v20.2d,v21.2d,v22.2d},[x0]		//store Htable[3..5]
	ret
.size	gcm_init_v8,.-gcm_init_v8
.globl	gcm_gmult_v8
.type	gcm_gmult_v8,%function
.align	4
gcm_gmult_v8:
	ld1	{v17.2d},[x0]		//load Xi
	movi	v19.16b,#0xe1
	ld1	{v20.2d,v21.2d},[x1]	//load twisted H, ...
	shl	v19.2d,v19.2d,#57
#ifndef __ARMEB__
	rev64	v17.16b,v17.16b
#endif
	ext	v3.16b,v17.16b,v17.16b,#8

	pmull	v0.1q,v20.1d,v3.1d		//H.lo·Xi.lo
	eor	v17.16b,v17.16b,v3.16b		//Karatsuba pre-processing
	pmull2	v2.1q,v20.2d,v3.2d		//H.hi·Xi.hi
	pmull	v1.1q,v21.1d,v17.1d		//(H.lo+H.hi)·(Xi.lo+Xi.hi)

	ext	v17.16b,v0.16b,v2.16b,#8		//Karatsuba post-processing
	eor	v18.16b,v0.16b,v2.16b
	eor	v1.16b,v1.16b,v17.16b
	eor	v1.16b,v1.16b,v18.16b
	pmull	v18.1q,v0.1d,v19.1d		//1st phase of reduction

	ins	v2.d[0],v1.d[1]
	ins	v1.d[1],v0.d[0]
	eor	v0.16b,v1.16b,v18.16b

	ext	v18.16b,v0.16b,v0.16b,#8		//2nd phase of reduction
	pmull	v0.1q,v0.1d,v19.1d
	eor	v18.16b,v18.16b,v2.16b
	eor	v0.16b,v0.16b,v18.16b

#ifndef __ARMEB__
	rev64	v0.16b,v0.16b
#endif
	ext	v0.16b,v0.16b,v0.16b,#8
	st1	{v0.2d},[x0]		//write out Xi

	ret
.size	gcm_gmult_v8,.-gcm_gmult_v8
.globl	gcm_ghash_v8
.type	gcm_ghash_v8,%function
.align	4
gcm_ghash_v8:
	cmp	x3,#64
	b.hs	.Lgcm_ghash_v8_4x
	ld1	{v0.2d},[x0]		//load [rotated] Xi
						//"[rotated]" means that
						//loaded value would have
						//to be rotated in order to
						//make it appear as in
						//algorithm specification
	subs	x3,x3,#32		//see if x3 is 32 or larger
	mov	x12,#16		//x12 is used as post-
						//increment for input pointer;
						//as loop is modulo-scheduled
						//x12 is zeroed just in time
						//to preclude overstepping
						//inp[len], which means that
						//last block[s] are actually
						//loaded twice, but last
						//copy is not processed
	ld1	{v20.2d,v21.2d},[x1],#32	//load twisted H, ..., H^2
	movi	v19.16b,#0xe1
	ld1	{v22.2d},[x1]
	csel	x12,xzr,x12,eq			//is it time to zero x12?
	ext	v0.16b,v0.16b,v0.16b,#8		//rotate Xi
	ld1	{v16.2d},[x2],#16	//load [rotated] I[0]
	shl	v19.2d,v19.2d,#57		//compose 0xc2.0 constant
#ifndef __ARMEB__
	rev64	v16.16b,v16.16b
	rev64	v0.16b,v0.16b
#endif
	ext	v3.16b,v16.16b,v16.16b,#8		//rotate I[0]
	b.lo	.Lodd_tail_v8		//x3 was less than 32
	ld1	{v17.2d},[x2],x12	//load [rotated] I[1]
#ifndef __ARMEB__
	rev64	v17.16b,v17.16b
#endif
	ext	v7.16b,v17.16b,v17.16b,#8
	eor	v3.16b,v3.16b,v0.16b		//I[i]^=Xi
	pmull	v4.1q,v20.1d,v7.1d		//H·Ii+1
	eor	v17.16b,v17.16b,v7.16b		//Karatsuba pre-processing
	pmull2	v6.1q,v20.2d,v7.2d
	b	.Loop_mod2x_v8

.align	4
.Loop_mod2x_v8:
	ext	v18.16b,v3.16b,v3.16b,#8
	subs	x3,x3,#32		//is there more data?
	pmull	v0.1q,v22.1d,v3.1d		//H^2.lo·Xi.lo
	csel	x12,xzr,x12,lo			//is it time to zero x12?

	pmull	v5.1q,v21.1d,v17.1d
	eor	v18.16b,v18.16b,v3.16b		//Karatsuba pre-processing
	pmull2	v2.1q,v22.2d,v3.2d		//H^2.hi·Xi.hi
	eor	v0.16b,v0.16b,v4.16b		//accumulate
	pmull2	v1.1q,v21.2d,v18.2d		//(H^2.lo+H^2.hi)·(Xi.lo+Xi.hi)
	ld1	{v16.2d},[x2],x12	//load [rotated] I[i+2]

	eor	v2.16b,v2.16b,v6.16b
	csel	x12,xzr,x12,eq			//is it time to zero x12?
	eor	v1.16b,v1.16b,v5.16b

	ext	v17.16b,v0.16b,v2.16b,#8		//Karatsuba post-processing
	eor	v18.16b,v0.16b,v2.16b
	eor	v1.16b,v1.16b,v17.16b
	ld1	{v17.2d},[x2],x12	//load [rotated] I[i+3]
#ifndef __ARMEB__
	rev64	v16.16b,v16.16b
#endif
	eor	v1.16b,v1.16b,v18.16b
	pmull	v18.1q,v0.1d,v19.1d		//1st phase of reduction

#ifndef __ARMEB__
	rev64	v17.16b,v17.16b
#endif
	ins	v2.d[0],v1.d[1]
	ins	v1.d[1],v0.d[0]
	ext	v7.16b,v17.16b,v17.16b,#8
	ext	v3.16b,v16.16b,v16.16b,#8
	eor	v0.16b,v1.16b,v18.16b
	pmull	v4.1q,v20.1d,v7.1d		//H·Ii+1
	eor	v3.16b,v3.16b,v2.16b		//accumulate v3.16b early

	ext	v18.16b,v0.16b,v0.16b,#8		//2nd phase of reduction
	pmull	v0.1q,v0.1d,v19.1d
	eor	v3.16b,v3.16b,v18.16b
	eor	v17.16b,v17.16b,v7.16b		//Karatsuba pre-processing
	eor	v3.16b,v3.16b,v0.16b
	pmull2	v6.1q,v20.2d,v7.2d
	b.hs	.Loop_mod2x_v8		//there was at least 32 more bytes

	eor	v2.16b,v2.16b,v18.16b
	ext	v3.16b,v16.16b,v16.16b,#8		//re-construct v3.16b
	adds	x3,x3,#32		//re-construct x3
	eor	v0.16b,v0.16b,v2.16b		//re-construct v0.16b
	b.eq	.Ldone_v8		//is x3 zero?
.Lodd_tail_v8:
	ext	v18.16b,v0.16b,v0.16b,#8
	eor	v3.16b,v3.16b,v0.16b		//inp^=Xi
	eor	v17.16b,v16.16b,v18.16b		//v17.16b is rotated inp^Xi

	pmull	v0.1q,v20.1d,v3.1d		//H.lo·Xi.lo
	eor	v17.16b,v17.16b,v3.16b		//Karatsuba pre-processing
	pmull2	v2.1q,v20.2d,v3.2d		//H.hi·Xi.hi
	pmull	v1.1q,v21.1d,v17.1d		//(H.lo+H.hi)·(Xi.lo+Xi.hi)

	ext	v17.16b,v0.16b,v2.16b,#8		//Karatsuba post-processing
	eor	v18.16b,v0.16b,v2.16b
	eor	v1.16b,v1.16b,v17.16b
	eor	v1.16b,v1.16b,v18.16b
	pmull	v18.1q,v0.1d,v19.1d		//1st phase of reduction

	ins	v2.d[0],v1.d[1]
	ins	v1.d[1],v0.d[0]
	eor	v0.16b,v1.16b,v18.16b

	ext	v18.16b,v0.16b,v0.16b,#8		//2nd phase of reduction
	pmull	v0.1q,v0.1d,v19.1d
	eor	v18.16b,v18.16b,v2.16b
	eor	v0.16b,v0.16b,v18.16b

.Ldone_v8:
#ifndef __ARMEB__
	rev64	v0.16b,v0.16b
#endif
	ext	v0.16b,v0.16b,v0.16b,#8
	st1	{v0.2d},[x0]		//write out Xi

	ret
.size	gcm_ghash_v8,.-gcm_ghash_v8
.type	gcm_ghash_v8_4x,%function
.align	4
gcm_ghash_v8_4x:
.Lgcm_ghash_v8_4x:
	ld1	{v0.2d},[x0]		//load [rotated] Xi
	ld1	{v20.2d,v21.2d,v22.2d},[x1],#48	//load twisted H, ..., H^2
	movi	v19.16b,#0xe1
	ld1	{v26.2d,v27.2d,v28.2d},[x1]	//load twisted H^3, ..., H^4
	shl	v19.2d,v19.2d,#57		//compose 0xc2.0 constant

	ld1	{v4.2d,v5.2d,v6.2d,v7.2d},[x2],#64
#ifndef __ARMEB__
	rev64	v0.16b,v0.16b
	rev64	v5.16b,v5.16b
	rev64	v6.16b,v6.16b
	rev64	v7.16b,v7.16b
	rev64	v4.16b,v4.16b
#endif
	ext	v25.16b,v7.16b,v7.16b,#8
	ext	v24.16b,v6.16b,v6.16b,#8
	ext	v23.16b,v5.16b,v5.16b,#8

	pmull	v29.1q,v20.1d,v25.1d		//H·Ii+3
	eor	v7.16b,v7.16b,v25.16b
	pmull2	v31.1q,v20.2d,v25.2d
	pmull	v30.1q,v21.1d,v7.1d

	pmull	v16.1q,v22.1d,v24.1d		//H^2·Ii+2
	eor	v6.16b,v6.16b,v24.16b
	pmull2	v24.1q,v22.2d,v24.2d
	pmull2	v6.1q,v21.2d,v6.2d

	eor	v29.16b,v29.16b,v16.16b
	eor	v31.16b,v31.16b,v24.16b
	eor	v30.16b,v30.16b,v6.16b

	pmull	v7.1q,v26.1d,v23.1d		//H^3·Ii+1
	eor	v5.16b,v5.16b,v23.16b
	pmull2	v23.1q,v26.2d,v23.2d
	pmull	v5.1q,v27.1d,v5.1d

	eor	v29.16b,v29.16b,v7.16b
	eor	v31.16b,v31.16b,v23.16b
	eor	v30.16b,v30.16b,v5.16b

	subs	x3,x3,#128
	b.lo	.Ltail4x

	b	.Loop4x

.align	4
.Loop4x:
	eor	v16.16b,v4.16b,v0.16b
	ld1	{v4.2d,v5.2d,v6.2d,v7.2d},[x2],#64
	ext	v3.16b,v16.16b,v16.16b,#8
#ifndef __ARMEB__
	rev64	v5.16b,v5.16b
	rev64	v6.16b,v6.16b
	rev64	v7.16b,v7.16b
	rev64	v4.16b,v4.16b
#endif

	pmull	v0.1q,v28.1d,v3.1d		//H^4·(Xi+Ii)
	eor	v16.16b,v16.16b,v3.16b
	pmull2	v2.1q,v28.2d,v3.2d
	ext	v25.16b,v7.16b,v7.16b,#8
	pmull2	v1.1q,v27.2d,v16.2d

	eor	v0.16b,v0.16b,v29.16b
	eor	v2.16b,v2.16b,v31.16b
	ext	v24.16b,v6.16b,v6.16b,#8
	eor	v1.16b,v1.16b,v30.16b
	ext	v23.16b,v5.16b,v5.16b,#8

	ext	v17.16b,v0.16b,v2.16b,#8		//Karatsuba post-processing
	eor	v18.16b,v0.16b,v2.16b
	pmull	v29.1q,v20.1d,v25.1d		//H·Ii+3
	eor	v7.16b,v7.16b,v25.16b
	eor	v1.16b,v1.16b,v17.16b
	pmull2	v31.1q,v20.2d,v25.2d
	eor	v1.16b,v1.16b,v18.16b
	pmull	v30.1q,v21.1d,v7.1d

	pmull	v18.1q,v0.1d,v19.1d		//1st phase of reduction
	ins	v2.d[0],v1.d[1]
	ins	v1.d[1],v0.d[0]
	pmull	v16.1q,v22.1d,v24.1d		//H^2·Ii+2
	eor	v6.16b,v6.16b,v24.16b
	pmull2	v24.1q,v22.2d,v24.2d
	eor	v0.16b,v1.16b,v18.16b
	pmull2	v6.1q,v21.2d,v6.2d

	eor	v29.16b,v29.16b,v16.16b
	eor	v31.16b,v31.16b,v24.16b
	eor	v30.16b,v30.16b,v6.16b

	ext	v18.16b,v0.16b,v0.16b,#8		//2nd phase of reduction
	pmull	v0.1q,v0.1d,v19.1d
	pmull	v7.1q,v26.1d,v23.1d		//H^3·Ii+1
	eor	v5.16b,v5.16b,v23.16b
	eor	v18.16b,v18.16b,v2.16b
	pmull2	v23.1q,v26.2d,v23.2d
	pmull	v5.1q,v27.1d,v5.1d

	eor	v0.16b,v0.16b,v18.16b
	eor	v29.16b,v29.16b,v7.16b
	eor	v31.16b,v31.16b,v23.16b
	ext	v0.16b,v0.16b,v0.16b,#8
	eor	v30.16b,v30.16b,v5.16b

	subs	x3,x3,#64
	b.hs	.Loop4x

.Ltail4x:
	eor	v16.16b,v4.16b,v0.16b
	ext	v3.16b,v16.16b,v16.16b,#8

	pmull	v0.1q,v28.1d,v3.1d		//H^4·(Xi+Ii)
	eor	v16.16b,v16.16b,v3.16b
	pmull2	v2.1q,v28.2d,v3.2d
	pmull2	v1.1q,v27.2d,v16.2d

	eor	v0.16b,v0.16b,v29.16b
	eor	v2.16b,v2.16b,v31.16b
	eor	v1.16b,v1.16b,v30.16b

	adds	x3,x3,#64
	b.eq	.Ldone4x

	cmp	x3,#32
	b.lo	.Lone
	b.eq	.Ltwo
.Lthree:
	ext	v17.16b,v0.16b,v2.16b,#8		//Karatsuba post-processing
	eor	v18.16b,v0.16b,v2.16b
	eor	v1.16b,v1.16b,v17.16b
	ld1	{v4.2d,v5.2d,v6.2d},[x2]
	eor	v1.16b,v1.16b,v18.16b
#ifndef	__ARMEB__
	rev64	v5.16b,v5.16b
	rev64	v6.16b,v6.16b
	rev64	v4.16b,v4.16b
#endif

	pmull	v18.1q,v0.1d,v19.1d		//1st phase of reduction
	ins	v2.d[0],v1.d[1]
	ins	v1.d[1],v0.d[0]
	ext	v24.16b,v6.16b,v6.16b,#8
	ext	v23.16b,v5.16b,v5.16b,#8
	eor	v0.16b,v1.16b,v18.16b

	pmull	v29.1q,v20.1d,v24.1d		//H·Ii+2
	eor	v6.16b,v6.16b,v24.16b

	ext	v18.16b,v0.16b,v0.16b,#8		//2nd phase of reduction
	pmull	v0.1q,v0.1d,v19.1d
	eor	v18.16b,v18.16b,v2.16b
	pmull2	v31.1q,v20.2d,v24.2d
	pmull	v30.1q,v21.1d,v6.1d
	eor	v0.16b,v0.16b,v18.16b
	pmull	v7.1q,v22.1d,v23.1d		//H^2·Ii+1
	eor	v5.16b,v5.16b,v23.16b
	ext	v0.16b,v0.16b,v0.16b,#8

	pmull2	v23.1q,v22.2d,v23.2d
	eor	v16.16b,v4.16b,v0.16b
	pmull2	v5.1q,v21.2d,v5.2d
	ext	v3.16b,v16.16b,v16.16b,#8

	eor	v29.16b,v29.16b,v7.16b
	eor	v31.16b,v31.16b,v23.16b
	eor	v30.16b,v30.16b,v5.16b

	pmull	v0.1q,v26.1d,v3.1d		//H^3·(Xi+Ii)
	eor	v16.16b,v16.16b,v3.16b
	pmull2	v2.1q,v26.2d,v3.2d
	pmull	v1.1q,v27.1d,v16.1d

	eor	v0.16b,v0.16b,v29.16b
	eor	v2.16b,v2.16b,v31.16b
	eor	v1.16b,v1.16b,v30.16b
	b	.Ldone4x

.align	4
.Ltwo:
	ext	v17.16b,v0.16b,v2.16b,#8		//Karatsuba post-processing
	eor	v18.16b,v0.16b,v2.16b
	eor	v1.16b,v1.16b,v17.16b
	ld1	{v4.2d,v5.2d},[x2]
	eor	v1.16b,v1.16b,v18.16b
#ifndef	__ARMEB__
	rev64	v5.16b,v5.16b
	rev64	v4.16b,v4.16b
#endif

	pmull	v18.1q,v0.1d,v19.1d		//1st phase of reduction
	ins	v2.d[0],v1.d[1]
	ins	v1.d[1],v0.d[0]
	ext	v23.16b,v5.16b,v5.16b,#8
	eor	v0.16b,v1.16b,v18.16b

	ext	v18.16b,v0.16b,v0.16b,#8		//2nd phase of reduction
	pmull	v0.1q,v0.1d,v19.1d
	eor	v18.16b,v18.16b,v2.16b
	eor	v0.16b,v0.16b,v18.16b
	ext	v0.16b,v0.16b,v0.16b,#8

	pmull	v29.1q,v20.1d,v23.1d		//H·Ii+1
	eor	v5.16b,v5.16b,v23.16b

	eor	v16.16b,v4.16b,v0.16b
	ext	v3.16b,v16.16b,v16.16b,#8

	pmull2	v31.1q,v20.2d,v23.2d
	pmull	v30.1q,v21.1d,v5.1d

	pmull	v0.1q,v22.1d,v3.1d		//H^2·(Xi+Ii)
	eor	v16.16b,v16.16b,v3.16b
	pmull2	v2.1q,v22.2d,v3.2d
	pmull2	v1.1q,v21.2d,v16.2d

	eor	v0.16b,v0.16b,v29.16b
	eor	v2.16b,v2.16b,v31.16b
	eor	v1.16b,v1.16b,v30.16b
	b	.Ldone4x

.align	4
.Lone:
	ext	v17.16b,v0.16b,v2.16b,#8		//Karatsuba post-processing
	eor	v18.16b,v0.16b,v2.16b
	eor	v1.16b,v1.16b,v17.16b
	ld1	{v4.2d},[x2]
	eor	v1.16b,v1.16b,v18.16b
#ifndef	__ARMEB__
	rev64	v4.16b,v4.16b
#endif

	pmull	v18.1q,v0.1d,v19.1d		//1st phase of reduction
	ins	v2.d[0],v1.d[1]
	ins	v1.d[1],v0.d[0]
	eor	v0.16b,v1.16b,v18.16b

	ext	v18.16b,v0.16b,v0.16b,#8		//2nd phase of reduction
	pmull	v0.1q,v0.1d,v19.1d
	eor	v18.16b,v18.16b,v2.16b
	eor	v0.16b,v0.16b,v18.16b
	ext	v0.16b,v0.16b,v0.16b,#8

	eor	v16.16b,v4.16b,v0.16b
	ext	v3.16b,v16.16b,v16.16b,#8

	pmull	v0.1q,v20.1d,v3.1d
	eor	v16.16b,v16.16b,v3.16b
	pmull2	v2.1q,v20.2d,v3.2d
	pmull	v1.1q,v21.1d,v16.1d

.Ldone4x:
	ext	v17.16b,v0.16b,v2.16b,#8		//Karatsuba post-processing
	eor	v18.16b,v0.16b,v2.16b
	eor	v1.16b,v1.16b,v17.16b
	eor	v1.16b,v1.16b,v18.16b

	pmull	v18.1q,v0.1d,v19.1d		//1st phase of reduction
	ins	v2.d[0],v1.d[1]
	ins	v1.d[1],v0.d[0]
	eor	v0.16b,v1.16b,v18.16b

	ext	v18.16b,v0.16b,v0.16b,#8		//2nd phase of reduction
	pmull	v0.1q,v0.1d,v19.1d
	eor	v18.16b,v18.16b,v2.16b
	eor	v0.16b,v0.16b,v18.16b
	ext	v0.16b,v0.16b,v0.16b,#8

#ifndef __ARMEB__
	rev64	v0.16b,v0.16b
#endif
	st1	{v0.2d},[x0]		//write out Xi

	ret
.size	gcm_ghash_v8_4x,.-gcm_ghash_v8_4x
.byte	71,72,65,83,72,32,102,111,114,32,65,82,77,118,56,44,32,67,82,89,80,84,79,71,65,77,83,32,98,121,32,60,97,112,112,114,111,64,111,112,101,110,115,115,108,46,111,114,103,62,0
.align	2
.align	2
#endif