summaryrefslogtreecommitdiffstats
path: root/src/spdk/intel-ipsec-mb/avx512/sha1_x16_avx512.asm
blob: d67046ce5500c940a637182e840a1547cbec6081 (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
;;
;; Copyright (c) 2012-2018, Intel Corporation
;;
;; Redistribution and use in source and binary forms, with or without
;; modification, are permitted provided that the following conditions are met:
;;
;;     * Redistributions of source code must retain the above copyright notice,
;;       this list of conditions and the following disclaimer.
;;     * Redistributions in binary form must reproduce the above copyright
;;       notice, this list of conditions and the following disclaimer in the
;;       documentation and/or other materials provided with the distribution.
;;     * Neither the name of Intel Corporation nor the names of its contributors
;;       may be used to endorse or promote products derived from this software
;;       without specific prior written permission.
;;
;; THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
;; AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
;; IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
;; DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE
;; FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
;; DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
;; SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
;; CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
;; OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
;; OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
;;

;; Stack must be aligned to 32 bytes before call
;;
;; Registers:		RAX RBX RCX RDX RBP RSI RDI R8  R9  R10 R11 R12 R13 R14 R15
;;			-----------------------------------------------------------
;; Windows clobbers:	RAX         RDX             R8  R9  R10 R11 R12 R13 R14 R15
;; Windows preserves:	    RBX RCX     RBP RSI RDI
;;			-----------------------------------------------------------
;; Linux clobbers:	RAX         RDX     RSI         R9  R10 R11 R12 R13 R14 R15
;; Linux preserves:	    RBX RCX     RBP     RDI R8
;;			-----------------------------------------------------------
;; Clobbers ZMM0-31

%include "include/os.asm"
;%define DO_DBGPRINT
%include "include/dbgprint.asm"
%include "mb_mgr_datastruct.asm"
%include "include/transpose_avx512.asm"
%include "include/reg_sizes.asm"

section .data
default rel
align 64
K00_19:	;ddq 0x5A8279995A8279995A8279995A827999
	;ddq 0x5A8279995A8279995A8279995A827999
	;ddq 0x5A8279995A8279995A8279995A827999
	;ddq 0x5A8279995A8279995A8279995A827999
	dq 0x5A8279995A827999, 0x5A8279995A827999
	dq 0x5A8279995A827999, 0x5A8279995A827999
	dq 0x5A8279995A827999, 0x5A8279995A827999
	dq 0x5A8279995A827999, 0x5A8279995A827999
K20_39: ;ddq 0x6ED9EBA16ED9EBA16ED9EBA16ED9EBA1
	;ddq 0x6ED9EBA16ED9EBA16ED9EBA16ED9EBA1
	;ddq 0x6ED9EBA16ED9EBA16ED9EBA16ED9EBA1
	;ddq 0x6ED9EBA16ED9EBA16ED9EBA16ED9EBA1
	dq 0x6ED9EBA16ED9EBA1, 0x6ED9EBA16ED9EBA1
	dq 0x6ED9EBA16ED9EBA1, 0x6ED9EBA16ED9EBA1
	dq 0x6ED9EBA16ED9EBA1, 0x6ED9EBA16ED9EBA1
	dq 0x6ED9EBA16ED9EBA1, 0x6ED9EBA16ED9EBA1
K40_59: ;ddq 0x8F1BBCDC8F1BBCDC8F1BBCDC8F1BBCDC
	;ddq 0x8F1BBCDC8F1BBCDC8F1BBCDC8F1BBCDC
	;ddq 0x8F1BBCDC8F1BBCDC8F1BBCDC8F1BBCDC
	;ddq 0x8F1BBCDC8F1BBCDC8F1BBCDC8F1BBCDC
	dq 0x8F1BBCDC8F1BBCDC, 0x8F1BBCDC8F1BBCDC
	dq 0x8F1BBCDC8F1BBCDC, 0x8F1BBCDC8F1BBCDC
	dq 0x8F1BBCDC8F1BBCDC, 0x8F1BBCDC8F1BBCDC
	dq 0x8F1BBCDC8F1BBCDC, 0x8F1BBCDC8F1BBCDC
K60_79: ;ddq 0xCA62C1D6CA62C1D6CA62C1D6CA62C1D6
	;ddq 0xCA62C1D6CA62C1D6CA62C1D6CA62C1D6
	;ddq 0xCA62C1D6CA62C1D6CA62C1D6CA62C1D6
	;ddq 0xCA62C1D6CA62C1D6CA62C1D6CA62C1D6
	dq 0xCA62C1D6CA62C1D6, 0xCA62C1D6CA62C1D6
	dq 0xCA62C1D6CA62C1D6, 0xCA62C1D6CA62C1D6
	dq 0xCA62C1D6CA62C1D6, 0xCA62C1D6CA62C1D6
	dq 0xCA62C1D6CA62C1D6, 0xCA62C1D6CA62C1D6

PSHUFFLE_BYTE_FLIP_MASK: ;ddq 0x0c0d0e0f08090a0b0405060700010203
			 ;ddq 0x0c0d0e0f08090a0b0405060700010203
			 ;ddq 0x0c0d0e0f08090a0b0405060700010203
			 ;ddq 0x0c0d0e0f08090a0b0405060700010203
	dq 0x0405060700010203, 0x0c0d0e0f08090a0b
	dq 0x0405060700010203, 0x0c0d0e0f08090a0b
	dq 0x0405060700010203, 0x0c0d0e0f08090a0b
	dq 0x0405060700010203, 0x0c0d0e0f08090a0b

section .text

%define APPEND(a,b) a %+ b

%ifdef LINUX
%define arg1	rdi
%define arg2	rsi
%define arg3	rdx
%define arg4	rcx
%else
%define arg1	rcx
%define arg2	rdx
%define arg3	r8
%define arg4	r9
%endif

%define state	arg1
%define SIZE	arg2
%define IDX	arg3

%define A	zmm0
%define B	zmm1
%define C	zmm2
%define D	zmm3
%define E	zmm4
%define KT  	zmm5
%define AA	zmm6
%define BB	zmm7
%define CC	zmm8
%define DD	zmm9
%define EE	zmm10
%define TMP0	zmm11
%define TMP1	zmm12
%define TMP2	zmm13
%define TMP3	zmm14
%define TMP4	zmm15

%define W0	zmm16
%define W1	zmm17
%define W2	zmm18
%define W3	zmm19
%define W4	zmm20
%define W5	zmm21
%define W6	zmm22
%define W7	zmm23
%define W8	zmm24
%define W9	zmm25
%define W10	zmm26
%define W11	zmm27
%define W12	zmm28
%define W13	zmm29
%define W14	zmm30
%define W15	zmm31

%define inp0	r9
%define inp1	r10
%define inp2	r11
%define inp3	r12
%define inp4	r13
%define inp5	r14
%define inp6	r15
%define inp7	rax

%macro ROTATE_ARGS 0
%xdefine TMP_ E
%xdefine E D
%xdefine D C
%xdefine C B
%xdefine B A
%xdefine A TMP_
%endm

%macro PROCESS_LOOP 2
%define %%WT		%1
%define %%F_IMMED	%2

	; T = ROTL_5(A) + Ft(B,C,D) + E + Kt + Wt
	; E=D, D=C, C=ROTL_30(B), B=A, A=T

	; Ft
	;  0-19 Ch(B,C,D) = (B&C) ^ (~B&D)
	; 20-39, 60-79 Parity(B,C,D) = B ^ C ^ D
	; 40-59 Maj(B,C,D) = (B&C) ^ (B&D) ^ (C&D)

	vmovdqa32	TMP1, B			; Copy B
	vpaddd		E, E, %%WT		; E = E + Wt
	vpternlogd	TMP1, C, D, %%F_IMMED	; TMP1 = Ft(B,C,D)
	vpaddd		E, E, KT		; E = E + Wt + Kt
	vprold		TMP0, A, 5		; TMP0 = ROTL_5(A)
	vpaddd		E, E, TMP1		; E = Ft(B,C,D) + E + Kt + Wt
	vprold		B, B, 30		; B = ROTL_30(B)
	vpaddd		E, E, TMP0		; E = T

	ROTATE_ARGS
%endmacro

%macro MSG_SCHED_ROUND_16_79 4
%define %%WT	%1
%define %%WTp2	%2
%define %%WTp8	%3
%define %%WTp13	%4
	; Wt = ROTL_1(Wt-3 ^ Wt-8 ^ Wt-14 ^ Wt-16)
	; Wt+16 = ROTL_1(Wt+13 ^ Wt+8 ^ Wt+2 ^ Wt)
	vpternlogd	%%WT, %%WTp2, %%WTp8, 0x96
	vpxord		%%WT, %%WT, %%WTp13
	vprold		%%WT, %%WT, 1
%endmacro


; Note this is reading in two blocks of data from each lane,
; in preparation for the upcoming needed transpose to build msg schedule.
; Each register will contain 32 bytes from one lane plus 32 bytes
; from another lane.
; The first 8 registers will contain the first 32 bytes of all lanes,
; where register X (0 <= X <= 7) will contain bytes 0-31 from lane X in the first half
; and 0-31 bytes from lane X+8 in the second half.
; The last 8 registers will contain the last 32 bytes of all lanes,
; where register Y (8 <= Y <= 15) wil contain bytes 32-63 from lane Y-8 in the first half
; and 32-63 bytes from lane Y in the second half.
; This method helps reducing the number of shuffles required to transpose the data.
%macro MSG_SCHED_ROUND_00_15 6
%define %%Wt         %1 ; [out] zmm register to load the next block
%define %%LANE_IDX   %2 ; [in] lane index (0-15)
%define %%BASE_PTR   %3 ; [in] base address of the input data
%define %%OFFSET_PTR %4 ; [in] offset to get next block of data from the lane
%define %%TMP1       %5 ; [clobbered] temporary gp register
%define %%TMP2       %6 ; [clobbered] temporary gp register
%if (%%LANE_IDX < 8)
	mov	      %%TMP1,	   [%%BASE_PTR + %%LANE_IDX*PTR_SZ]
	mov	      %%TMP2,      [%%BASE_PTR + (%%LANE_IDX+8)*PTR_SZ]
	vmovups       YWORD(%%Wt), [%%TMP1+%%OFFSET_PTR]
	vinserti64x4  %%Wt, %%Wt,  [%%TMP2+%%OFFSET_PTR], 0x01
%else
	mov	     %%TMP1,      [%%BASE_PTR + (%%LANE_IDX-8)*PTR_SZ]
	mov	     %%TMP2,      [%%BASE_PTR + %%LANE_IDX*PTR_SZ]
	vmovups      YWORD(%%Wt), [%%TMP1+%%OFFSET_PTR+32]
	vinserti64x4 %%Wt, %%Wt,  [%%TMP2+%%OFFSET_PTR+32], 0x01
%endif
%endmacro

align 64
; void sha1_mult_x16_avx3(void **input_data, UINT128 *digest, UINT32 size)
; arg 1 : pointer to SHA1 args structure
; arg 2 : size (in blocks) ;; assumed to be >= 1
MKGLOBAL(sha1_x16_avx512,function,internal)
sha1_x16_avx512:
	;; Initialize digests
	vmovdqu32	A, [state + 0*SHA1_DIGEST_ROW_SIZE]
	vmovdqu32	B, [state + 1*SHA1_DIGEST_ROW_SIZE]
	vmovdqu32	C, [state + 2*SHA1_DIGEST_ROW_SIZE]
	vmovdqu32	D, [state + 3*SHA1_DIGEST_ROW_SIZE]
	vmovdqu32	E, [state + 4*SHA1_DIGEST_ROW_SIZE]
	DBGPRINTL_ZMM "Sha1-AVX512 incoming transposed digest", A, B, C, D, E
	DBGPRINTL64   "SIZE", SIZE

	xor IDX, IDX

	;; Load first blocks of data into ZMM registers before
	;; performing a 16x16 32-bit transpose.
	;; To speed up the transpose, data is loaded in chunks of 32 bytes,
	;; interleaving data between lane X and lane X+8.
	;; This way, final shuffles between top half and bottom half
	;; of the matrix are avoided.
	mov	inp0, [state + _data_ptr_sha1 + 0*PTR_SZ]
	mov	inp1, [state + _data_ptr_sha1 + 1*PTR_SZ]
	mov	inp2, [state + _data_ptr_sha1 + 2*PTR_SZ]
	mov	inp3, [state + _data_ptr_sha1 + 3*PTR_SZ]
	mov	inp4, [state + _data_ptr_sha1 + 4*PTR_SZ]
	mov	inp5, [state + _data_ptr_sha1 + 5*PTR_SZ]
	mov	inp6, [state + _data_ptr_sha1 + 6*PTR_SZ]
	mov	inp7, [state + _data_ptr_sha1 + 7*PTR_SZ]

	TRANSPOSE16_U32_LOAD_FIRST8 W0, W1, W2,  W3,  W4,  W5,  W6,  W7, \
				    W8, W9, W10, W11, W12, W13, W14, W15, \
				    inp0, inp1, inp2, inp3, inp4, inp5, \
				    inp6, inp7, IDX

	mov	inp0, [state + _data_ptr_sha1 + 8*PTR_SZ]
	mov	inp1, [state + _data_ptr_sha1 + 9*PTR_SZ]
	mov	inp2, [state + _data_ptr_sha1 +10*PTR_SZ]
	mov	inp3, [state + _data_ptr_sha1 +11*PTR_SZ]
	mov	inp4, [state + _data_ptr_sha1 +12*PTR_SZ]
	mov	inp5, [state + _data_ptr_sha1 +13*PTR_SZ]
	mov	inp6, [state + _data_ptr_sha1 +14*PTR_SZ]
	mov	inp7, [state + _data_ptr_sha1 +15*PTR_SZ]

	TRANSPOSE16_U32_LOAD_LAST8 W0, W1, W2,  W3,  W4,  W5,  W6,  W7, \
				   W8, W9, W10, W11, W12, W13, W14, W15, \
				   inp0, inp1, inp2, inp3, inp4, inp5, \
				   inp6, inp7, IDX
lloop:
	vmovdqa32	TMP2, [rel PSHUFFLE_BYTE_FLIP_MASK]

	add	IDX, 64

	TRANSPOSE16_U32 W0, W1, W2, W3, W4, W5, W6, W7, W8, W9, W10, W11, W12, W13, W14, W15, TMP0, TMP1, TMP3, TMP4
	DBGPRINTL_ZMM "Sha1-AVX512 incoming transposed input", W0, W1, W2, W3, W4, W6, W7, W8, W9, W10, W11, W12, W13, W14, W15

%assign I 0
%rep 16
       	vpshufb	APPEND(W,I), APPEND(W,I), TMP2
%assign I (I+1)
%endrep

	; Save digests for later addition
	vmovdqa32	AA, A
	vmovdqa32	BB, B
	vmovdqa32	CC, C
	vmovdqa32	DD, D
	vmovdqa32	EE, E

	vmovdqa32	KT, [rel K00_19]
%assign I 0xCA
%assign J 0
%assign K 2
%assign L 8
%assign M 13
%assign N 0
%rep 64
	PROCESS_LOOP  APPEND(W,J),  I
	MSG_SCHED_ROUND_16_79  APPEND(W,J), APPEND(W,K), APPEND(W,L), APPEND(W,M)
	%if N = 19
		vmovdqa32	KT, [rel K20_39]
		%assign I 0x96
	%elif N = 39
		vmovdqa32	KT, [rel K40_59]
		%assign I 0xE8
	%elif N = 59
		vmovdqa32	KT, [rel K60_79]
		%assign I 0x96
	%endif
%assign J ((J+1)% 16)
%assign K ((K+1)% 16)
%assign L ((L+1)% 16)
%assign M ((M+1)% 16)
%assign N (N+1)
%endrep

	; Check if this is the last block
	sub 	SIZE, 1
	je	lastLoop

%assign I 0x96
%assign J 0
%rep 16
	PROCESS_LOOP  APPEND(W,J),  I
	MSG_SCHED_ROUND_00_15 APPEND(W,J), J, state + _data_ptr_sha1, IDX, inp0, inp1
%assign J (J+1)
%endrep

	; Add old digest
	vpaddd		A,A,AA
	vpaddd		B,B,BB
	vpaddd		C,C,CC
	vpaddd		D,D,DD
	vpaddd		E,E,EE

	jmp lloop

lastLoop:
; Need to reset argument rotation values to Round 64 values
%xdefine TMP_ A
%xdefine A B
%xdefine B C
%xdefine C D
%xdefine D E
%xdefine E TMP_

	; Process last 16 rounds
%assign I 0x96
%assign J 0
%rep 16
	PROCESS_LOOP  APPEND(W,J), I
%assign J (J+1)
%endrep

	; Add old digest
	vpaddd		A,A,AA
	vpaddd		B,B,BB
	vpaddd		C,C,CC
	vpaddd		D,D,DD
	vpaddd		E,E,EE

	; Write out digest
	; Do we need to untranspose digests???
	vmovdqu32	[state + 0*SHA1_DIGEST_ROW_SIZE], A
	vmovdqu32	[state + 1*SHA1_DIGEST_ROW_SIZE], B
	vmovdqu32	[state + 2*SHA1_DIGEST_ROW_SIZE], C
	vmovdqu32	[state + 3*SHA1_DIGEST_ROW_SIZE], D
	vmovdqu32	[state + 4*SHA1_DIGEST_ROW_SIZE], E
	DBGPRINTL_ZMM "Sha1-AVX512 outgoing transposed digest", A, B, C, D, E

	;; update input pointers
	mov	inp0, [state + _data_ptr_sha1 + 0*PTR_SZ]
	mov	inp1, [state + _data_ptr_sha1 + 1*PTR_SZ]
	mov	inp2, [state + _data_ptr_sha1 + 2*PTR_SZ]
	mov	inp3, [state + _data_ptr_sha1 + 3*PTR_SZ]
	mov	inp4, [state + _data_ptr_sha1 + 4*PTR_SZ]
	mov	inp5, [state + _data_ptr_sha1 + 5*PTR_SZ]
	mov	inp6, [state + _data_ptr_sha1 + 6*PTR_SZ]
	mov	inp7, [state + _data_ptr_sha1 + 7*PTR_SZ]
	add	inp0, IDX
	add	inp1, IDX
	add	inp2, IDX
	add	inp3, IDX
	add	inp4, IDX
	add	inp5, IDX
	add	inp6, IDX
	add	inp7, IDX
	mov	[state + _data_ptr_sha1 + 0*PTR_SZ], inp0
	mov	[state + _data_ptr_sha1 + 1*PTR_SZ], inp1
	mov	[state + _data_ptr_sha1 + 2*PTR_SZ], inp2
	mov	[state + _data_ptr_sha1 + 3*PTR_SZ], inp3
	mov	[state + _data_ptr_sha1 + 4*PTR_SZ], inp4
	mov	[state + _data_ptr_sha1 + 5*PTR_SZ], inp5
	mov	[state + _data_ptr_sha1 + 6*PTR_SZ], inp6
	mov	[state + _data_ptr_sha1 + 7*PTR_SZ], inp7

	mov	inp0, [state + _data_ptr_sha1 + 8*PTR_SZ]
	mov	inp1, [state + _data_ptr_sha1 + 9*PTR_SZ]
	mov	inp2, [state + _data_ptr_sha1 + 10*PTR_SZ]
	mov	inp3, [state + _data_ptr_sha1 + 11*PTR_SZ]
	mov	inp4, [state + _data_ptr_sha1 + 12*PTR_SZ]
	mov	inp5, [state + _data_ptr_sha1 + 13*PTR_SZ]
	mov	inp6, [state + _data_ptr_sha1 + 14*PTR_SZ]
	mov	inp7, [state + _data_ptr_sha1 + 15*PTR_SZ]
	add	inp0, IDX
	add	inp1, IDX
	add	inp2, IDX
	add	inp3, IDX
	add	inp4, IDX
	add	inp5, IDX
	add	inp6, IDX
	add	inp7, IDX
	mov	[state + _data_ptr_sha1 + 8*PTR_SZ], inp0
	mov	[state + _data_ptr_sha1 + 9*PTR_SZ], inp1
	mov	[state + _data_ptr_sha1 + 10*PTR_SZ], inp2
	mov	[state + _data_ptr_sha1 + 11*PTR_SZ], inp3
	mov	[state + _data_ptr_sha1 + 12*PTR_SZ], inp4
	mov	[state + _data_ptr_sha1 + 13*PTR_SZ], inp5
	mov	[state + _data_ptr_sha1 + 14*PTR_SZ], inp6
	mov	[state + _data_ptr_sha1 + 15*PTR_SZ], inp7

	ret

%ifdef LINUX
section .note.GNU-stack noalloc noexec nowrite progbits
%endif