xref: /freebsd/sys/crypto/armv8/armv8_crypto_wrap.c (revision 193d9e768ba63fcfb187cfd17f461f7d41345048)
1 /*-
2  * Copyright (c) 2016 The FreeBSD Foundation
3  * All rights reserved.
4  *
5  * This software was developed by Andrew Turner under
6  * sponsorship from the FreeBSD Foundation.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  */
29 
30 /*
31  * This code is built with floating-point enabled. Make sure to have entered
32  * into floating-point context before calling any of these functions.
33  */
34 
35 #include <sys/cdefs.h>
36 __FBSDID("$FreeBSD$");
37 
38 #include <sys/param.h>
39 #include <sys/systm.h>
40 #include <sys/malloc.h>
41 #include <sys/queue.h>
42 
43 #include <opencrypto/cryptodev.h>
44 #include <crypto/armv8/armv8_crypto.h>
45 
46 #include <arm_neon.h>
47 
48 static uint8x16_t
49 armv8_aes_enc(int rounds, const uint8x16_t *keysched, const uint8x16_t from)
50 {
51 	uint8x16_t tmp;
52 	int i;
53 
54 	tmp = from;
55 	for (i = 0; i < rounds - 1; i += 2) {
56 		tmp = vaeseq_u8(tmp, keysched[i]);
57 		tmp = vaesmcq_u8(tmp);
58 		tmp = vaeseq_u8(tmp, keysched[i + 1]);
59 		tmp = vaesmcq_u8(tmp);
60 	}
61 
62 	tmp = vaeseq_u8(tmp, keysched[rounds - 1]);
63 	tmp = vaesmcq_u8(tmp);
64 	tmp = vaeseq_u8(tmp, keysched[rounds]);
65 	tmp = veorq_u8(tmp, keysched[rounds + 1]);
66 
67 	return (tmp);
68 }
69 
70 static uint8x16_t
71 armv8_aes_dec(int rounds, const uint8x16_t *keysched, const uint8x16_t from)
72 {
73 	uint8x16_t tmp;
74 	int i;
75 
76 	tmp = from;
77 	for (i = 0; i < rounds - 1; i += 2) {
78 		tmp = vaesdq_u8(tmp, keysched[i]);
79 		tmp = vaesimcq_u8(tmp);
80 		tmp = vaesdq_u8(tmp, keysched[i+1]);
81 		tmp = vaesimcq_u8(tmp);
82 	}
83 
84 	tmp = vaesdq_u8(tmp, keysched[rounds - 1]);
85 	tmp = vaesimcq_u8(tmp);
86 	tmp = vaesdq_u8(tmp, keysched[rounds]);
87 	tmp = veorq_u8(tmp, keysched[rounds + 1]);
88 
89 	return (tmp);
90 }
91 
92 void
93 armv8_aes_encrypt_cbc(int rounds, const void *key_schedule, size_t len,
94     const uint8_t *from, uint8_t *to, const uint8_t iv[static AES_BLOCK_LEN])
95 {
96 	uint8x16_t tot, ivreg, tmp;
97 	size_t i;
98 
99 	len /= AES_BLOCK_LEN;
100 	ivreg = vld1q_u8(iv);
101 	for (i = 0; i < len; i++) {
102 		tmp = vld1q_u8(from);
103 		tot = armv8_aes_enc(rounds - 1, key_schedule,
104 		    veorq_u8(tmp, ivreg));
105 		ivreg = tot;
106 		vst1q_u8(to, tot);
107 		from += AES_BLOCK_LEN;
108 		to += AES_BLOCK_LEN;
109 	}
110 }
111 
112 void
113 armv8_aes_decrypt_cbc(int rounds, const void *key_schedule, size_t len,
114     uint8_t *buf, const uint8_t iv[static AES_BLOCK_LEN])
115 {
116 	uint8x16_t ivreg, nextiv, tmp;
117 	size_t i;
118 
119 	len /= AES_BLOCK_LEN;
120 	ivreg = vld1q_u8(iv);
121 	for (i = 0; i < len; i++) {
122 		nextiv = vld1q_u8(buf);
123 		tmp = armv8_aes_dec(rounds - 1, key_schedule, nextiv);
124 		vst1q_u8(buf, veorq_u8(tmp, ivreg));
125 		ivreg = nextiv;
126 		buf += AES_BLOCK_LEN;
127 	}
128 }
129