xref: /linux/arch/x86/crypto/sha1_avx2_x86_64_asm.S (revision 1ac731c529cd4d6adbce134754b51ff7d822b145)
1/*
2 *	Implement fast SHA-1 with AVX2 instructions. (x86_64)
3 *
4 * This file is provided under a dual BSD/GPLv2 license.  When using or
5 * redistributing this file, you may do so under either license.
6 *
7 * GPL LICENSE SUMMARY
8 *
9 * Copyright(c) 2014 Intel Corporation.
10 *
11 * This program is free software; you can redistribute it and/or modify
12 * it under the terms of version 2 of the GNU General Public License as
13 * published by the Free Software Foundation.
14 *
15 * This program is distributed in the hope that it will be useful, but
16 * WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
18 * General Public License for more details.
19 *
20 * Contact Information:
21 * Ilya Albrekht <ilya.albrekht@intel.com>
22 * Maxim Locktyukhin <maxim.locktyukhin@intel.com>
23 * Ronen Zohar <ronen.zohar@intel.com>
24 * Chandramouli Narayanan <mouli@linux.intel.com>
25 *
26 * BSD LICENSE
27 *
28 * Copyright(c) 2014 Intel Corporation.
29 *
30 * Redistribution and use in source and binary forms, with or without
31 * modification, are permitted provided that the following conditions
32 * are met:
33 *
34 * Redistributions of source code must retain the above copyright
35 * notice, this list of conditions and the following disclaimer.
36 * Redistributions in binary form must reproduce the above copyright
37 * notice, this list of conditions and the following disclaimer in
38 * the documentation and/or other materials provided with the
39 * distribution.
40 * Neither the name of Intel Corporation nor the names of its
41 * contributors may be used to endorse or promote products derived
42 * from this software without specific prior written permission.
43 *
44 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
45 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
46 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
47 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
48 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
49 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
50 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
51 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
52 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
53 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
54 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
55 *
56 */
57
58/*
59 * SHA-1 implementation with Intel(R) AVX2 instruction set extensions.
60 *
61 *This implementation is based on the previous SSSE3 release:
62 *Visit http://software.intel.com/en-us/articles/
63 *and refer to improving-the-performance-of-the-secure-hash-algorithm-1/
64 *
65 *Updates 20-byte SHA-1 record at start of 'state', from 'input', for
66 *even number of 'blocks' consecutive 64-byte blocks.
67 *
68 *extern "C" void sha1_transform_avx2(
69 *	struct sha1_state *state, const u8* input, int blocks );
70 */
71
72#include <linux/linkage.h>
73
74#define	CTX	%rdi	/* arg1 */
75#define BUF	%rsi	/* arg2 */
76#define CNT	%rdx	/* arg3 */
77
78#define	REG_A	%ecx
79#define	REG_B	%esi
80#define	REG_C	%edi
81#define	REG_D	%eax
82#define	REG_E	%edx
83#define	REG_TB	%ebx
84#define	REG_TA	%r12d
85#define	REG_RA	%rcx
86#define	REG_RB	%rsi
87#define	REG_RC	%rdi
88#define	REG_RD	%rax
89#define	REG_RE	%rdx
90#define	REG_RTA	%r12
91#define	REG_RTB	%rbx
92#define	REG_T1	%r11d
93#define	xmm_mov	vmovups
94#define	avx2_zeroupper	vzeroupper
95#define	RND_F1	1
96#define	RND_F2	2
97#define	RND_F3	3
98
99.macro REGALLOC
100	.set A, REG_A
101	.set B, REG_B
102	.set C, REG_C
103	.set D, REG_D
104	.set E, REG_E
105	.set TB, REG_TB
106	.set TA, REG_TA
107
108	.set RA, REG_RA
109	.set RB, REG_RB
110	.set RC, REG_RC
111	.set RD, REG_RD
112	.set RE, REG_RE
113
114	.set RTA, REG_RTA
115	.set RTB, REG_RTB
116
117	.set T1, REG_T1
118.endm
119
120#define HASH_PTR	%r9
121#define BLOCKS_CTR	%r8
122#define BUFFER_PTR	%r10
123#define BUFFER_PTR2	%r13
124
125#define PRECALC_BUF	%r14
126#define WK_BUF		%r15
127
128#define W_TMP		%xmm0
129#define WY_TMP		%ymm0
130#define WY_TMP2		%ymm9
131
132# AVX2 variables
133#define WY0		%ymm3
134#define WY4		%ymm5
135#define WY08		%ymm7
136#define WY12		%ymm8
137#define WY16		%ymm12
138#define WY20		%ymm13
139#define WY24		%ymm14
140#define WY28		%ymm15
141
142#define YMM_SHUFB_BSWAP	%ymm10
143
144/*
145 * Keep 2 iterations precalculated at a time:
146 *    - 80 DWORDs per iteration * 2
147 */
148#define W_SIZE		(80*2*2 +16)
149
150#define WK(t)	((((t) % 80) / 4)*32 + ( (t) % 4)*4 + ((t)/80)*16 )(WK_BUF)
151#define PRECALC_WK(t)	((t)*2*2)(PRECALC_BUF)
152
153
154.macro UPDATE_HASH  hash, val
155	add	\hash, \val
156	mov	\val, \hash
157.endm
158
159.macro PRECALC_RESET_WY
160	.set WY_00, WY0
161	.set WY_04, WY4
162	.set WY_08, WY08
163	.set WY_12, WY12
164	.set WY_16, WY16
165	.set WY_20, WY20
166	.set WY_24, WY24
167	.set WY_28, WY28
168	.set WY_32, WY_00
169.endm
170
171.macro PRECALC_ROTATE_WY
172	/* Rotate macros */
173	.set WY_32, WY_28
174	.set WY_28, WY_24
175	.set WY_24, WY_20
176	.set WY_20, WY_16
177	.set WY_16, WY_12
178	.set WY_12, WY_08
179	.set WY_08, WY_04
180	.set WY_04, WY_00
181	.set WY_00, WY_32
182
183	/* Define register aliases */
184	.set WY, WY_00
185	.set WY_minus_04, WY_04
186	.set WY_minus_08, WY_08
187	.set WY_minus_12, WY_12
188	.set WY_minus_16, WY_16
189	.set WY_minus_20, WY_20
190	.set WY_minus_24, WY_24
191	.set WY_minus_28, WY_28
192	.set WY_minus_32, WY
193.endm
194
195.macro PRECALC_00_15
196	.if (i == 0) # Initialize and rotate registers
197		PRECALC_RESET_WY
198		PRECALC_ROTATE_WY
199	.endif
200
201	/* message scheduling pre-compute for rounds 0-15 */
202	.if   ((i & 7) == 0)
203		/*
204		 * blended AVX2 and ALU instruction scheduling
205		 * 1 vector iteration per 8 rounds
206		 */
207		vmovdqu (i * 2)(BUFFER_PTR), W_TMP
208	.elseif ((i & 7) == 1)
209		vinsertf128 $1, ((i-1) * 2)(BUFFER_PTR2),\
210			 WY_TMP, WY_TMP
211	.elseif ((i & 7) == 2)
212		vpshufb YMM_SHUFB_BSWAP, WY_TMP, WY
213	.elseif ((i & 7) == 4)
214		vpaddd  K_XMM + K_XMM_AR(%rip), WY, WY_TMP
215	.elseif ((i & 7) == 7)
216		vmovdqu  WY_TMP, PRECALC_WK(i&~7)
217
218		PRECALC_ROTATE_WY
219	.endif
220.endm
221
222.macro PRECALC_16_31
223	/*
224	 * message scheduling pre-compute for rounds 16-31
225	 * calculating last 32 w[i] values in 8 XMM registers
226	 * pre-calculate K+w[i] values and store to mem
227	 * for later load by ALU add instruction
228	 *
229	 * "brute force" vectorization for rounds 16-31 only
230	 * due to w[i]->w[i-3] dependency
231	 */
232	.if   ((i & 7) == 0)
233		/*
234		 * blended AVX2 and ALU instruction scheduling
235		 * 1 vector iteration per 8 rounds
236		 */
237		/* w[i-14] */
238		vpalignr	$8, WY_minus_16, WY_minus_12, WY
239		vpsrldq	$4, WY_minus_04, WY_TMP               /* w[i-3] */
240	.elseif ((i & 7) == 1)
241		vpxor	WY_minus_08, WY, WY
242		vpxor	WY_minus_16, WY_TMP, WY_TMP
243	.elseif ((i & 7) == 2)
244		vpxor	WY_TMP, WY, WY
245		vpslldq	$12, WY, WY_TMP2
246	.elseif ((i & 7) == 3)
247		vpslld	$1, WY, WY_TMP
248		vpsrld	$31, WY, WY
249	.elseif ((i & 7) == 4)
250		vpor	WY, WY_TMP, WY_TMP
251		vpslld	$2, WY_TMP2, WY
252	.elseif ((i & 7) == 5)
253		vpsrld	$30, WY_TMP2, WY_TMP2
254		vpxor	WY, WY_TMP, WY_TMP
255	.elseif ((i & 7) == 7)
256		vpxor	WY_TMP2, WY_TMP, WY
257		vpaddd  K_XMM + K_XMM_AR(%rip), WY, WY_TMP
258		vmovdqu	WY_TMP, PRECALC_WK(i&~7)
259
260		PRECALC_ROTATE_WY
261	.endif
262.endm
263
264.macro PRECALC_32_79
265	/*
266	 * in SHA-1 specification:
267	 * w[i] = (w[i-3] ^ w[i-8]  ^ w[i-14] ^ w[i-16]) rol 1
268	 * instead we do equal:
269	 * w[i] = (w[i-6] ^ w[i-16] ^ w[i-28] ^ w[i-32]) rol 2
270	 * allows more efficient vectorization
271	 * since w[i]=>w[i-3] dependency is broken
272	 */
273
274	.if   ((i & 7) == 0)
275	/*
276	 * blended AVX2 and ALU instruction scheduling
277	 * 1 vector iteration per 8 rounds
278	 */
279		vpalignr	$8, WY_minus_08, WY_minus_04, WY_TMP
280	.elseif ((i & 7) == 1)
281		/* W is W_minus_32 before xor */
282		vpxor	WY_minus_28, WY, WY
283	.elseif ((i & 7) == 2)
284		vpxor	WY_minus_16, WY_TMP, WY_TMP
285	.elseif ((i & 7) == 3)
286		vpxor	WY_TMP, WY, WY
287	.elseif ((i & 7) == 4)
288		vpslld	$2, WY, WY_TMP
289	.elseif ((i & 7) == 5)
290		vpsrld	$30, WY, WY
291		vpor	WY, WY_TMP, WY
292	.elseif ((i & 7) == 7)
293		vpaddd  K_XMM + K_XMM_AR(%rip), WY, WY_TMP
294		vmovdqu	WY_TMP, PRECALC_WK(i&~7)
295
296		PRECALC_ROTATE_WY
297	.endif
298.endm
299
300.macro PRECALC r, s
301	.set i, \r
302
303	.if (i < 40)
304		.set K_XMM, 32*0
305	.elseif (i < 80)
306		.set K_XMM, 32*1
307	.elseif (i < 120)
308		.set K_XMM, 32*2
309	.else
310		.set K_XMM, 32*3
311	.endif
312
313	.if (i<32)
314		PRECALC_00_15	\s
315	.elseif (i<64)
316		PRECALC_16_31	\s
317	.elseif (i < 160)
318		PRECALC_32_79	\s
319	.endif
320.endm
321
322.macro ROTATE_STATE
323	.set T_REG, E
324	.set E, D
325	.set D, C
326	.set C, B
327	.set B, TB
328	.set TB, A
329	.set A, T_REG
330
331	.set T_REG, RE
332	.set RE, RD
333	.set RD, RC
334	.set RC, RB
335	.set RB, RTB
336	.set RTB, RA
337	.set RA, T_REG
338.endm
339
340/* Macro relies on saved ROUND_Fx */
341
342.macro RND_FUN f, r
343	.if (\f == RND_F1)
344		ROUND_F1	\r
345	.elseif (\f == RND_F2)
346		ROUND_F2	\r
347	.elseif (\f == RND_F3)
348		ROUND_F3	\r
349	.endif
350.endm
351
352.macro RR r
353	.set round_id, (\r % 80)
354
355	.if (round_id == 0)        /* Precalculate F for first round */
356		.set ROUND_FUNC, RND_F1
357		mov	B, TB
358
359		rorx	$(32-30), B, B    /* b>>>2 */
360		andn	D, TB, T1
361		and	C, TB
362		xor	T1, TB
363	.endif
364
365	RND_FUN ROUND_FUNC, \r
366	ROTATE_STATE
367
368	.if   (round_id == 18)
369		.set ROUND_FUNC, RND_F2
370	.elseif (round_id == 38)
371		.set ROUND_FUNC, RND_F3
372	.elseif (round_id == 58)
373		.set ROUND_FUNC, RND_F2
374	.endif
375
376	.set round_id, ( (\r+1) % 80)
377
378	RND_FUN ROUND_FUNC, (\r+1)
379	ROTATE_STATE
380.endm
381
382.macro ROUND_F1 r
383	add	WK(\r), E
384
385	andn	C, A, T1			/* ~b&d */
386	lea	(RE,RTB), E		/* Add F from the previous round */
387
388	rorx	$(32-5), A, TA		/* T2 = A >>> 5 */
389	rorx	$(32-30),A, TB		/* b>>>2 for next round */
390
391	PRECALC	(\r)			/* msg scheduling for next 2 blocks */
392
393	/*
394	 * Calculate F for the next round
395	 * (b & c) ^ andn[b, d]
396	 */
397	and	B, A			/* b&c */
398	xor	T1, A			/* F1 = (b&c) ^ (~b&d) */
399
400	lea	(RE,RTA), E		/* E += A >>> 5 */
401.endm
402
403.macro ROUND_F2 r
404	add	WK(\r), E
405	lea	(RE,RTB), E		/* Add F from the previous round */
406
407	/* Calculate F for the next round */
408	rorx	$(32-5), A, TA		/* T2 = A >>> 5 */
409	.if ((round_id) < 79)
410		rorx	$(32-30), A, TB	/* b>>>2 for next round */
411	.endif
412	PRECALC	(\r)			/* msg scheduling for next 2 blocks */
413
414	.if ((round_id) < 79)
415		xor	B, A
416	.endif
417
418	add	TA, E			/* E += A >>> 5 */
419
420	.if ((round_id) < 79)
421		xor	C, A
422	.endif
423.endm
424
425.macro ROUND_F3 r
426	add	WK(\r), E
427	PRECALC	(\r)			/* msg scheduling for next 2 blocks */
428
429	lea	(RE,RTB), E		/* Add F from the previous round */
430
431	mov	B, T1
432	or	A, T1
433
434	rorx	$(32-5), A, TA		/* T2 = A >>> 5 */
435	rorx	$(32-30), A, TB		/* b>>>2 for next round */
436
437	/* Calculate F for the next round
438	 * (b and c) or (d and (b or c))
439	 */
440	and	C, T1
441	and	B, A
442	or	T1, A
443
444	add	TA, E			/* E += A >>> 5 */
445
446.endm
447
448/* Add constant only if (%2 > %3) condition met (uses RTA as temp)
449 * %1 + %2 >= %3 ? %4 : 0
450 */
451.macro ADD_IF_GE a, b, c, d
452	mov     \a, RTA
453	add     $\d, RTA
454	cmp     $\c, \b
455	cmovge  RTA, \a
456.endm
457
458/*
459 * macro implements 80 rounds of SHA-1, for multiple blocks with s/w pipelining
460 */
461.macro SHA1_PIPELINED_MAIN_BODY
462
463	REGALLOC
464
465	mov	(HASH_PTR), A
466	mov	4(HASH_PTR), B
467	mov	8(HASH_PTR), C
468	mov	12(HASH_PTR), D
469	mov	16(HASH_PTR), E
470
471	mov	%rsp, PRECALC_BUF
472	lea	(2*4*80+32)(%rsp), WK_BUF
473
474	# Precalc WK for first 2 blocks
475	ADD_IF_GE BUFFER_PTR2, BLOCKS_CTR, 2, 64
476	.set i, 0
477	.rept    160
478		PRECALC i
479		.set i, i + 1
480	.endr
481
482	/* Go to next block if needed */
483	ADD_IF_GE BUFFER_PTR, BLOCKS_CTR, 3, 128
484	ADD_IF_GE BUFFER_PTR2, BLOCKS_CTR, 4, 128
485	xchg	WK_BUF, PRECALC_BUF
486
487	.align 32
488.L_loop:
489	/*
490	 * code loops through more than one block
491	 * we use K_BASE value as a signal of a last block,
492	 * it is set below by: cmovae BUFFER_PTR, K_BASE
493	 */
494	test BLOCKS_CTR, BLOCKS_CTR
495	jnz .L_begin
496	.align 32
497	jmp	.L_end
498	.align 32
499.L_begin:
500
501	/*
502	 * Do first block
503	 * rounds: 0,2,4,6,8
504	 */
505	.set j, 0
506	.rept 5
507		RR	j
508		.set j, j+2
509	.endr
510
511	/*
512	 * rounds:
513	 * 10,12,14,16,18
514	 * 20,22,24,26,28
515	 * 30,32,34,36,38
516	 * 40,42,44,46,48
517	 * 50,52,54,56,58
518	 */
519	.rept 25
520		RR	j
521		.set j, j+2
522	.endr
523
524	/* Update Counter */
525	sub $1, BLOCKS_CTR
526	/* Move to the next block only if needed*/
527	ADD_IF_GE BUFFER_PTR, BLOCKS_CTR, 4, 128
528	/*
529	 * rounds
530	 * 60,62,64,66,68
531	 * 70,72,74,76,78
532	 */
533	.rept 10
534		RR	j
535		.set j, j+2
536	.endr
537
538	UPDATE_HASH	(HASH_PTR), A
539	UPDATE_HASH	4(HASH_PTR), TB
540	UPDATE_HASH	8(HASH_PTR), C
541	UPDATE_HASH	12(HASH_PTR), D
542	UPDATE_HASH	16(HASH_PTR), E
543
544	test	BLOCKS_CTR, BLOCKS_CTR
545	jz	.L_loop
546
547	mov	TB, B
548
549	/* Process second block */
550	/*
551	 * rounds
552	 *  0+80, 2+80, 4+80, 6+80, 8+80
553	 * 10+80,12+80,14+80,16+80,18+80
554	 */
555
556	.set j, 0
557	.rept 10
558		RR	j+80
559		.set j, j+2
560	.endr
561
562	/*
563	 * rounds
564	 * 20+80,22+80,24+80,26+80,28+80
565	 * 30+80,32+80,34+80,36+80,38+80
566	 */
567	.rept 10
568		RR	j+80
569		.set j, j+2
570	.endr
571
572	/*
573	 * rounds
574	 * 40+80,42+80,44+80,46+80,48+80
575	 * 50+80,52+80,54+80,56+80,58+80
576	 */
577	.rept 10
578		RR	j+80
579		.set j, j+2
580	.endr
581
582	/* update counter */
583	sub     $1, BLOCKS_CTR
584	/* Move to the next block only if needed*/
585	ADD_IF_GE BUFFER_PTR2, BLOCKS_CTR, 4, 128
586
587	/*
588	 * rounds
589	 * 60+80,62+80,64+80,66+80,68+80
590	 * 70+80,72+80,74+80,76+80,78+80
591	 */
592	.rept 10
593		RR	j+80
594		.set j, j+2
595	.endr
596
597	UPDATE_HASH	(HASH_PTR), A
598	UPDATE_HASH	4(HASH_PTR), TB
599	UPDATE_HASH	8(HASH_PTR), C
600	UPDATE_HASH	12(HASH_PTR), D
601	UPDATE_HASH	16(HASH_PTR), E
602
603	/* Reset state for AVX2 reg permutation */
604	mov	A, TA
605	mov	TB, A
606	mov	C, TB
607	mov	E, C
608	mov	D, B
609	mov	TA, D
610
611	REGALLOC
612
613	xchg	WK_BUF, PRECALC_BUF
614
615	jmp	.L_loop
616
617	.align 32
618.L_end:
619
620.endm
621/*
622 * macro implements SHA-1 function's body for several 64-byte blocks
623 * param: function's name
624 */
625.macro SHA1_VECTOR_ASM  name
626	SYM_FUNC_START(\name)
627
628	push	%rbx
629	push	%r12
630	push	%r13
631	push	%r14
632	push	%r15
633
634	RESERVE_STACK  = (W_SIZE*4 + 8+24)
635
636	/* Align stack */
637	push	%rbp
638	mov	%rsp, %rbp
639	and	$~(0x20-1), %rsp
640	sub	$RESERVE_STACK, %rsp
641
642	avx2_zeroupper
643
644	/* Setup initial values */
645	mov	CTX, HASH_PTR
646	mov	BUF, BUFFER_PTR
647
648	mov	BUF, BUFFER_PTR2
649	mov	CNT, BLOCKS_CTR
650
651	xmm_mov	BSWAP_SHUFB_CTL(%rip), YMM_SHUFB_BSWAP
652
653	SHA1_PIPELINED_MAIN_BODY
654
655	avx2_zeroupper
656
657	mov	%rbp, %rsp
658	pop	%rbp
659
660	pop	%r15
661	pop	%r14
662	pop	%r13
663	pop	%r12
664	pop	%rbx
665
666	RET
667
668	SYM_FUNC_END(\name)
669.endm
670
671.section .rodata
672
673#define K1 0x5a827999
674#define K2 0x6ed9eba1
675#define K3 0x8f1bbcdc
676#define K4 0xca62c1d6
677
678.align 128
679K_XMM_AR:
680	.long K1, K1, K1, K1
681	.long K1, K1, K1, K1
682	.long K2, K2, K2, K2
683	.long K2, K2, K2, K2
684	.long K3, K3, K3, K3
685	.long K3, K3, K3, K3
686	.long K4, K4, K4, K4
687	.long K4, K4, K4, K4
688
689BSWAP_SHUFB_CTL:
690	.long 0x00010203
691	.long 0x04050607
692	.long 0x08090a0b
693	.long 0x0c0d0e0f
694	.long 0x00010203
695	.long 0x04050607
696	.long 0x08090a0b
697	.long 0x0c0d0e0f
698.text
699
700SHA1_VECTOR_ASM     sha1_transform_avx2
701