xref: /openbmc/linux/arch/arm64/crypto/sha1-ce-core.S (revision 3a35093a)
1/* SPDX-License-Identifier: GPL-2.0-only */
2/*
3 * sha1-ce-core.S - SHA-1 secure hash using ARMv8 Crypto Extensions
4 *
5 * Copyright (C) 2014 Linaro Ltd <ard.biesheuvel@linaro.org>
6 */
7
8#include <linux/linkage.h>
9#include <asm/assembler.h>
10
11	.text
12	.arch		armv8-a+crypto
13
14	k0		.req	v0
15	k1		.req	v1
16	k2		.req	v2
17	k3		.req	v3
18
19	t0		.req	v4
20	t1		.req	v5
21
22	dga		.req	q6
23	dgav		.req	v6
24	dgb		.req	s7
25	dgbv		.req	v7
26
27	dg0q		.req	q12
28	dg0s		.req	s12
29	dg0v		.req	v12
30	dg1s		.req	s13
31	dg1v		.req	v13
32	dg2s		.req	s14
33
34	.macro		add_only, op, ev, rc, s0, dg1
35	.ifc		\ev, ev
36	add		t1.4s, v\s0\().4s, \rc\().4s
37	sha1h		dg2s, dg0s
38	.ifnb		\dg1
39	sha1\op		dg0q, \dg1, t0.4s
40	.else
41	sha1\op		dg0q, dg1s, t0.4s
42	.endif
43	.else
44	.ifnb		\s0
45	add		t0.4s, v\s0\().4s, \rc\().4s
46	.endif
47	sha1h		dg1s, dg0s
48	sha1\op		dg0q, dg2s, t1.4s
49	.endif
50	.endm
51
52	.macro		add_update, op, ev, rc, s0, s1, s2, s3, dg1
53	sha1su0		v\s0\().4s, v\s1\().4s, v\s2\().4s
54	add_only	\op, \ev, \rc, \s1, \dg1
55	sha1su1		v\s0\().4s, v\s3\().4s
56	.endm
57
58	.macro		loadrc, k, val, tmp
59	movz		\tmp, :abs_g0_nc:\val
60	movk		\tmp, :abs_g1:\val
61	dup		\k, \tmp
62	.endm
63
64	/*
65	 * void sha1_ce_transform(struct sha1_ce_state *sst, u8 const *src,
66	 *			  int blocks)
67	 */
68SYM_FUNC_START(sha1_ce_transform)
69	frame_push	3
70
71	mov		x19, x0
72	mov		x20, x1
73	mov		x21, x2
74
75	/* load round constants */
760:	loadrc		k0.4s, 0x5a827999, w6
77	loadrc		k1.4s, 0x6ed9eba1, w6
78	loadrc		k2.4s, 0x8f1bbcdc, w6
79	loadrc		k3.4s, 0xca62c1d6, w6
80
81	/* load state */
82	ld1		{dgav.4s}, [x19]
83	ldr		dgb, [x19, #16]
84
85	/* load sha1_ce_state::finalize */
86	ldr_l		w4, sha1_ce_offsetof_finalize, x4
87	ldr		w4, [x19, x4]
88
89	/* load input */
901:	ld1		{v8.4s-v11.4s}, [x20], #64
91	sub		w21, w21, #1
92
93CPU_LE(	rev32		v8.16b, v8.16b		)
94CPU_LE(	rev32		v9.16b, v9.16b		)
95CPU_LE(	rev32		v10.16b, v10.16b	)
96CPU_LE(	rev32		v11.16b, v11.16b	)
97
982:	add		t0.4s, v8.4s, k0.4s
99	mov		dg0v.16b, dgav.16b
100
101	add_update	c, ev, k0,  8,  9, 10, 11, dgb
102	add_update	c, od, k0,  9, 10, 11,  8
103	add_update	c, ev, k0, 10, 11,  8,  9
104	add_update	c, od, k0, 11,  8,  9, 10
105	add_update	c, ev, k1,  8,  9, 10, 11
106
107	add_update	p, od, k1,  9, 10, 11,  8
108	add_update	p, ev, k1, 10, 11,  8,  9
109	add_update	p, od, k1, 11,  8,  9, 10
110	add_update	p, ev, k1,  8,  9, 10, 11
111	add_update	p, od, k2,  9, 10, 11,  8
112
113	add_update	m, ev, k2, 10, 11,  8,  9
114	add_update	m, od, k2, 11,  8,  9, 10
115	add_update	m, ev, k2,  8,  9, 10, 11
116	add_update	m, od, k2,  9, 10, 11,  8
117	add_update	m, ev, k3, 10, 11,  8,  9
118
119	add_update	p, od, k3, 11,  8,  9, 10
120	add_only	p, ev, k3,  9
121	add_only	p, od, k3, 10
122	add_only	p, ev, k3, 11
123	add_only	p, od
124
125	/* update state */
126	add		dgbv.2s, dgbv.2s, dg1v.2s
127	add		dgav.4s, dgav.4s, dg0v.4s
128
129	cbz		w21, 3f
130
131	if_will_cond_yield_neon
132	st1		{dgav.4s}, [x19]
133	str		dgb, [x19, #16]
134	do_cond_yield_neon
135	b		0b
136	endif_yield_neon
137
138	b		1b
139
140	/*
141	 * Final block: add padding and total bit count.
142	 * Skip if the input size was not a round multiple of the block size,
143	 * the padding is handled by the C code in that case.
144	 */
1453:	cbz		x4, 4f
146	ldr_l		w4, sha1_ce_offsetof_count, x4
147	ldr		x4, [x19, x4]
148	movi		v9.2d, #0
149	mov		x8, #0x80000000
150	movi		v10.2d, #0
151	ror		x7, x4, #29		// ror(lsl(x4, 3), 32)
152	fmov		d8, x8
153	mov		x4, #0
154	mov		v11.d[0], xzr
155	mov		v11.d[1], x7
156	b		2b
157
158	/* store new state */
1594:	st1		{dgav.4s}, [x19]
160	str		dgb, [x19, #16]
161	frame_pop
162	ret
163SYM_FUNC_END(sha1_ce_transform)
164