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
|
/*
* (C) 2018 Jack Lloyd
*
* Botan is released under the Simplified BSD License (see license.txt)
*/
#include <botan/sm4.h>
#include <arm_neon.h>
namespace Botan {
namespace {
static const uint8_t qswap_tbl[16] = {
12, 13, 14, 15, 8, 9, 10, 11, 4, 5, 6, 7, 0, 1, 2, 3
};
static const uint8_t bswap_tbl[16] = {
15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0
};
inline uint32x4_t qswap_32(uint32x4_t B)
{
return vreinterpretq_u32_u8(vqtbl1q_u8(vreinterpretq_u8_u32(B), vld1q_u8(qswap_tbl)));
}
inline uint32x4_t bswap_32(uint32x4_t B)
{
return vreinterpretq_u32_u8(vrev32q_u8(vreinterpretq_u8_u32(B)));
}
/*
Swap both the quad-words and bytes within each word
equivalent to return bswap_32(qswap_32(B))
*/
inline uint32x4_t bqswap_32(uint32x4_t B)
{
return vreinterpretq_u32_u8(vqtbl1q_u8(vreinterpretq_u8_u32(B), vld1q_u8(bswap_tbl)));
}
#define SM4_E(B0, B1, B2, B3, K) do { \
B0 = vsm4eq_u32(B0, K); \
B1 = vsm4eq_u32(B1, K); \
B2 = vsm4eq_u32(B2, K); \
B3 = vsm4eq_u32(B3, K); \
} while(0)
}
void BOTAN_FUNC_ISA("arch=armv8.2-a+sm4")
SM4::sm4_armv8_encrypt(const uint8_t input8[], uint8_t output8[], size_t blocks) const
{
const uint32x4_t K0 = vld1q_u32(&m_RK[ 0]);
const uint32x4_t K1 = vld1q_u32(&m_RK[ 4]);
const uint32x4_t K2 = vld1q_u32(&m_RK[ 8]);
const uint32x4_t K3 = vld1q_u32(&m_RK[12]);
const uint32x4_t K4 = vld1q_u32(&m_RK[16]);
const uint32x4_t K5 = vld1q_u32(&m_RK[20]);
const uint32x4_t K6 = vld1q_u32(&m_RK[24]);
const uint32x4_t K7 = vld1q_u32(&m_RK[28]);
const uint32_t* input32 = reinterpret_cast<const uint32_t*>(reinterpret_cast<const void*>(input8));
uint32_t* output32 = reinterpret_cast<uint32_t*>(reinterpret_cast<void*>(output8));
while(blocks >= 4)
{
uint32x4_t B0 = bswap_32(vld1q_u32(input32));
uint32x4_t B1 = bswap_32(vld1q_u32(input32+4));
uint32x4_t B2 = bswap_32(vld1q_u32(input32+8));
uint32x4_t B3 = bswap_32(vld1q_u32(input32+12));
SM4_E(B0, B1, B2, B3, K0);
SM4_E(B0, B1, B2, B3, K1);
SM4_E(B0, B1, B2, B3, K2);
SM4_E(B0, B1, B2, B3, K3);
SM4_E(B0, B1, B2, B3, K4);
SM4_E(B0, B1, B2, B3, K5);
SM4_E(B0, B1, B2, B3, K6);
SM4_E(B0, B1, B2, B3, K7);
vst1q_u32(output32 , bqswap_32(B0));
vst1q_u32(output32+ 4, bqswap_32(B1));
vst1q_u32(output32+ 8, bqswap_32(B2));
vst1q_u32(output32+12, bqswap_32(B3));
input32 += 4*4;
output32 += 4*4;
blocks -= 4;
}
for(size_t i = 0; i != blocks; ++i)
{
uint32x4_t B = bswap_32(vld1q_u32(input32));
B = vsm4eq_u32(B, K0);
B = vsm4eq_u32(B, K1);
B = vsm4eq_u32(B, K2);
B = vsm4eq_u32(B, K3);
B = vsm4eq_u32(B, K4);
B = vsm4eq_u32(B, K5);
B = vsm4eq_u32(B, K6);
B = vsm4eq_u32(B, K7);
vst1q_u32(output32, bqswap_32(B));
input32 += 4;
output32 += 4;
}
}
void BOTAN_FUNC_ISA("arch=armv8.2-a+sm4")
SM4::sm4_armv8_decrypt(const uint8_t input8[], uint8_t output8[], size_t blocks) const
{
const uint32x4_t K0 = qswap_32(vld1q_u32(&m_RK[ 0]));
const uint32x4_t K1 = qswap_32(vld1q_u32(&m_RK[ 4]));
const uint32x4_t K2 = qswap_32(vld1q_u32(&m_RK[ 8]));
const uint32x4_t K3 = qswap_32(vld1q_u32(&m_RK[12]));
const uint32x4_t K4 = qswap_32(vld1q_u32(&m_RK[16]));
const uint32x4_t K5 = qswap_32(vld1q_u32(&m_RK[20]));
const uint32x4_t K6 = qswap_32(vld1q_u32(&m_RK[24]));
const uint32x4_t K7 = qswap_32(vld1q_u32(&m_RK[28]));
const uint32_t* input32 = reinterpret_cast<const uint32_t*>(reinterpret_cast<const void*>(input8));
uint32_t* output32 = reinterpret_cast<uint32_t*>(reinterpret_cast<void*>(output8));
while(blocks >= 4)
{
uint32x4_t B0 = bswap_32(vld1q_u32(input32));
uint32x4_t B1 = bswap_32(vld1q_u32(input32+4));
uint32x4_t B2 = bswap_32(vld1q_u32(input32+8));
uint32x4_t B3 = bswap_32(vld1q_u32(input32+12));
SM4_E(B0, B1, B2, B3, K7);
SM4_E(B0, B1, B2, B3, K6);
SM4_E(B0, B1, B2, B3, K5);
SM4_E(B0, B1, B2, B3, K4);
SM4_E(B0, B1, B2, B3, K3);
SM4_E(B0, B1, B2, B3, K2);
SM4_E(B0, B1, B2, B3, K1);
SM4_E(B0, B1, B2, B3, K0);
vst1q_u32(output32 , bqswap_32(B0));
vst1q_u32(output32+ 4, bqswap_32(B1));
vst1q_u32(output32+ 8, bqswap_32(B2));
vst1q_u32(output32+12, bqswap_32(B3));
input32 += 4*4;
output32 += 4*4;
blocks -= 4;
}
for(size_t i = 0; i != blocks; ++i)
{
uint32x4_t B = bswap_32(vld1q_u32(input32));
B = vsm4eq_u32(B, K7);
B = vsm4eq_u32(B, K6);
B = vsm4eq_u32(B, K5);
B = vsm4eq_u32(B, K4);
B = vsm4eq_u32(B, K3);
B = vsm4eq_u32(B, K2);
B = vsm4eq_u32(B, K1);
B = vsm4eq_u32(B, K0);
vst1q_u32(output32, bqswap_32(B));
input32 += 4;
output32 += 4;
}
}
#undef SM4_E
}
|
