-
Notifications
You must be signed in to change notification settings - Fork 0
/
blake2b.riscv64.S
242 lines (209 loc) · 5.24 KB
/
blake2b.riscv64.S
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
/*
riscv64-linux-gnu-as blake2b.riscv64.S -o blake2b.riscv64.o
riscv64-linux-gnu-objdump -D blake2b.riscv64.o
*/
.globl blake2b_compress
.macro rotr reg,imm,tmpreg
srli \tmpreg, \reg, \imm
slli \reg, \reg, 64-\imm
or \reg, \reg, \tmpreg
.endm
.macro mix va vb vc vd x y tmpreg
add \va, \va, \x
add \va, \va, \vb
xor \vd, \vd, \va
rotr \vd, 32, \tmpreg
add \vc, \vc, \vd
xor \vb, \vb, \vc
rotr \vb, 24, \tmpreg
add \va, \va, \y
add \va, \va, \vb
xor \vd, \vd, \va
rotr \vd, 16, \tmpreg
add \vc, \vc, \vd
xor \vb, \vb, \vc
rotr \vb, 63, \tmpreg
.endm
.macro call_mix mptr vptr va vb vc vd sa sb
ld t0, 8*\sa(\mptr)
ld t1, 8*\sb(\mptr)
ld t2, \va*8(\vptr)
ld t3, \vb*8(\vptr)
ld t4, \vc*8(\vptr)
ld t5, \vd*8(\vptr)
mix t2, t3, t4, t5, t0, t1, t6
sd t2, \va*8(\vptr)
sd t3, \vb*8(\vptr)
sd t4, \vc*8(\vptr)
sd t5, \vd*8(\vptr)
.endm
.macro compress_loop_iter mptr vptr s0 s1 s2 s3 s4 s5 s6 s7 s8 s9 s10 s11 s12 s13 s14 s15
call_mix \mptr \vptr 0 4 8 12 \s0 \s1
call_mix \mptr \vptr 1 5 9 13 \s2 \s3
call_mix \mptr \vptr 2 6 10 14 \s4 \s5
call_mix \mptr \vptr 3 7 11 15 \s6 \s7
call_mix \mptr \vptr 0 5 10 15 \s8 \s9
call_mix \mptr \vptr 1 6 11 12 \s10 \s11
call_mix \mptr \vptr 2 7 8 13 \s12 \s13
call_mix \mptr \vptr 3 4 9 14 \s14 \s15
.endm
.macro compress_loop mptr vptr
compress_loop_iter \mptr \vptr 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
compress_loop_iter \mptr \vptr 14 10 4 8 9 15 13 6 1 12 0 2 11 7 5 3
compress_loop_iter \mptr \vptr 11 8 12 0 5 2 15 13 10 14 3 6 7 1 9 4
compress_loop_iter \mptr \vptr 7 9 3 1 13 12 11 14 2 6 5 10 4 0 15 8
compress_loop_iter \mptr \vptr 9 0 5 7 2 4 10 15 14 1 11 12 6 8 3 13
compress_loop_iter \mptr \vptr 2 12 6 10 0 11 8 3 4 13 7 5 15 14 1 9
compress_loop_iter \mptr \vptr 12 5 1 15 14 13 4 10 0 7 6 3 9 2 8 11
compress_loop_iter \mptr \vptr 13 11 7 14 12 1 3 9 5 0 15 4 8 6 2 10
compress_loop_iter \mptr \vptr 6 15 14 9 11 3 0 8 12 2 13 7 1 4 10 5
compress_loop_iter \mptr \vptr 10 2 8 4 7 6 1 5 15 11 9 14 3 12 13 0
compress_loop_iter \mptr \vptr 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
compress_loop_iter \mptr \vptr 14 10 4 8 9 15 13 6 1 12 0 2 11 7 5 3
.endm
.macro mix_v_h i reg0 reg1 reg2 hptr vptr
ld \reg0, \i*8(\hptr)
ld \reg1, \i*8(\vptr)
ld \reg2, (\i+8)*8(\vptr)
xor \reg0, \reg0, \reg1
xor \reg0, \reg0, \reg2
sd \reg0, \i*8(\hptr)
.endm
.macro compress hptr chunkptr tptr isLastBlock ivptr vptr
# set up local work vector V
ld t0, 0*8(\hptr)
sd t0, 0*8(\vptr)
ld t1, 1*8(\hptr)
sd t1, 1*8(\vptr)
ld t2, 2*8(\hptr)
sd t2, 2*8(\vptr)
ld t3, 3*8(\hptr)
sd t3, 3*8(\vptr)
ld t4, 4*8(\hptr)
sd t4, 4*8(\vptr)
ld t5, 5*8(\hptr)
sd t5, 5*8(\vptr)
ld t6, 6*8(\hptr)
sd t6, 6*8(\vptr)
ld t0, 7*8(\hptr)
sd t0, 7*8(\vptr)
ld t1, 0*8(\ivptr)
sd t1, 8*8(\vptr)
ld t2, 1*8(\ivptr)
sd t2, 9*8(\vptr)
ld t3, 2*8(\ivptr)
sd t3, 10*8(\vptr)
ld t4, 3*8(\ivptr)
sd t4, 11*8(\vptr)
ld t5, 4*8(\ivptr)
#sd t5, 12*8(\vptr)
ld t6, 5*8(\ivptr)
#sd t6, 13*8(\vptr)
ld t0, 6*8(\ivptr)
#sd t0, 14*8(\vptr)
ld t1, 7*8(\ivptr)
sd t1, 15*8(\vptr)
# mix the 128-bit counter t into v12:v13
ld t2, 0*8(\tptr)
ld t3, 1*8(\tptr)
xor t5, t5, t2
xor t6, t6, t3
sd t5, 12*8(\vptr)
sd t6, 13*8(\vptr)
# last block requries inverting all the bits in v14
la t2, FFptr
ld t4, 0(t2)
mv t1, t0
xor t1, t1, t4
sd t1, 8(t2)
snez t3, \isLastBlock
slli t3, t3, 3
add t2, t2, t3
ld t4, 0(t2)
#xor t0, t0, t4 #oops, just need one xor, flipped seqz to snez above
#sd t0, 14*8(\vptr)
sd t4, 14*8(\vptr)
# twelve rounds of cryptographic mixing
compress_loop \chunkptr \vptr
# testing
#la a0, V
# mix the upper and lower halves of V into ongoing state vector h
mix_v_h 0 t0 t1 t2 \hptr \vptr
mix_v_h 1 t3 t4 t5 \hptr \vptr
mix_v_h 2 t0 t1 t2 \hptr \vptr
mix_v_h 3 t3 t4 t5 \hptr \vptr
mix_v_h 4 t0 t1 t2 \hptr \vptr
mix_v_h 5 t3 t4 t5 \hptr \vptr
mix_v_h 6 t0 t1 t2 \hptr \vptr
mix_v_h 7 t3 t4 t5 \hptr \vptr
/*
*/
# testing junk
#ld t2, 8(\ivptr)
#sd t2, 8*8(\vptr)
#mv a0, \hptr
#ret
.endm
.text
blake2b_compress:
# we don't use caller-saved registers, so no need to save them
/*
mul a0, a0, a0
mul a0, a0, a0
mul a0, a0, a0
mul a0, a0, a0
mul a0, a0, a0
mul a0, a0, a0
*/
# args are pointers to a0=hptr, a1=chunkptr, a2=numBytesCompressedptr, a3=isLastBlock
# where hptr is uint64_t[8]
# chunkptr is uint8_t* and chunk is 128 bytes
# numBytesCompressedptr is uint64_t[2]
# isLastblock is uint64_t
# also, we set a4=vptr, a5=ivptr
# where vptr is uint64_t[16], will be overwritten
# ivptr is uint64_t[8] to const values
#la a4, IV
la a5, V
compress a0 a1 a2 a3 a4 a5
/*
mul a0, a0, a0
mul a0, a0, a0
mul a0, a0, a0
mul a0, a0, a0
mul a0, a0, a0
mul a0, a0, a0
*/
ret
.data
.align 3
IV:
.quad 0x6a09e667f3bcc908
.quad 0xbb67ae8584caa73b
.quad 0x3c6ef372fe94f82b
.quad 0xa54ff53a5f1d36f1
.quad 0x510e527fade682d1
.quad 0x9b05688c2b3e6c1f
.quad 0x1f83d9abfb41bd6b
.quad 0x5be0cd19137e2179
V:
.quad 0x00
.quad 0x00
.quad 0x00
.quad 0x00
.quad 0x00
.quad 0x00
.quad 0x00
.quad 0x00
.quad 0x00
.quad 0x00
.quad 0x00
.quad 0x00
.quad 0x00
.quad 0x00
.quad 0x00
.quad 0x00
FFptr:
.quad 0xFFFFFFFFFFFFFFFF
FFptrPlus8:
.quad 0x00