Text file src/pkg/crypto/md5/md5block_amd64p32.s
1 // Original source:
2 // http://www.zorinaq.com/papers/md5-amd64.html
3 // http://www.zorinaq.com/papers/md5-amd64.tar.bz2
4 //
5 // Translated from Perl generating GNU assembly into
6 // #defines generating 6a assembly by the Go Authors.
7 //
8 // Restrictions to make code safe for Native Client:
9 // replace BP with R11, reloaded before use at return.
10 // replace R15 with R11.
11
12 #include "textflag.h"
13
14 // MD5 optimized for AMD64.
15 //
16 // Author: Marc Bevand <bevand_m (at) epita.fr>
17 // Licence: I hereby disclaim the copyright on this code and place it
18 // in the public domain.
19
20 TEXT ·block(SB),NOSPLIT,$0-16
21 MOVL dig+0(FP), R11
22 MOVL p+4(FP), SI
23 MOVL p_len+8(FP), DX
24 SHRQ $6, DX
25 SHLQ $6, DX
26
27 LEAQ (SI)(DX*1), DI
28 MOVL (0*4)(R11), AX
29 MOVL (1*4)(R11), BX
30 MOVL (2*4)(R11), CX
31 MOVL (3*4)(R11), DX
32
33 CMPQ SI, DI
34 JEQ end
35
36 loop:
37 MOVL AX, R12
38 MOVL BX, R13
39 MOVL CX, R14
40 MOVL DX, R11
41
42 MOVL (0*4)(SI), R8
43 MOVL DX, R9
44
45 #define ROUND1(a, b, c, d, index, const, shift) \
46 XORL c, R9; \
47 LEAL const(a)(R8*1), a; \
48 ANDL b, R9; \
49 XORL d, R9; \
50 MOVL (index*4)(SI), R8; \
51 ADDL R9, a; \
52 ROLL $shift, a; \
53 MOVL c, R9; \
54 ADDL b, a
55
56 ROUND1(AX,BX,CX,DX, 1,0xd76aa478, 7);
57 ROUND1(DX,AX,BX,CX, 2,0xe8c7b756,12);
58 ROUND1(CX,DX,AX,BX, 3,0x242070db,17);
59 ROUND1(BX,CX,DX,AX, 4,0xc1bdceee,22);
60 ROUND1(AX,BX,CX,DX, 5,0xf57c0faf, 7);
61 ROUND1(DX,AX,BX,CX, 6,0x4787c62a,12);
62 ROUND1(CX,DX,AX,BX, 7,0xa8304613,17);
63 ROUND1(BX,CX,DX,AX, 8,0xfd469501,22);
64 ROUND1(AX,BX,CX,DX, 9,0x698098d8, 7);
65 ROUND1(DX,AX,BX,CX,10,0x8b44f7af,12);
66 ROUND1(CX,DX,AX,BX,11,0xffff5bb1,17);
67 ROUND1(BX,CX,DX,AX,12,0x895cd7be,22);
68 ROUND1(AX,BX,CX,DX,13,0x6b901122, 7);
69 ROUND1(DX,AX,BX,CX,14,0xfd987193,12);
70 ROUND1(CX,DX,AX,BX,15,0xa679438e,17);
71 ROUND1(BX,CX,DX,AX, 0,0x49b40821,22);
72
73 MOVL (1*4)(SI), R8
74 MOVL DX, R9
75 MOVL DX, R10
76
77 #define ROUND2(a, b, c, d, index, const, shift) \
78 NOTL R9; \
79 LEAL const(a)(R8*1),a; \
80 ANDL b, R10; \
81 ANDL c, R9; \
82 MOVL (index*4)(SI),R8; \
83 ORL R9, R10; \
84 MOVL c, R9; \
85 ADDL R10, a; \
86 MOVL c, R10; \
87 ROLL $shift, a; \
88 ADDL b, a
89
90 ROUND2(AX,BX,CX,DX, 6,0xf61e2562, 5);
91 ROUND2(DX,AX,BX,CX,11,0xc040b340, 9);
92 ROUND2(CX,DX,AX,BX, 0,0x265e5a51,14);
93 ROUND2(BX,CX,DX,AX, 5,0xe9b6c7aa,20);
94 ROUND2(AX,BX,CX,DX,10,0xd62f105d, 5);
95 ROUND2(DX,AX,BX,CX,15, 0x2441453, 9);
96 ROUND2(CX,DX,AX,BX, 4,0xd8a1e681,14);
97 ROUND2(BX,CX,DX,AX, 9,0xe7d3fbc8,20);
98 ROUND2(AX,BX,CX,DX,14,0x21e1cde6, 5);
99 ROUND2(DX,AX,BX,CX, 3,0xc33707d6, 9);
100 ROUND2(CX,DX,AX,BX, 8,0xf4d50d87,14);
101 ROUND2(BX,CX,DX,AX,13,0x455a14ed,20);
102 ROUND2(AX,BX,CX,DX, 2,0xa9e3e905, 5);
103 ROUND2(DX,AX,BX,CX, 7,0xfcefa3f8, 9);
104 ROUND2(CX,DX,AX,BX,12,0x676f02d9,14);
105 ROUND2(BX,CX,DX,AX, 0,0x8d2a4c8a,20);
106
107 MOVL (5*4)(SI), R8
108 MOVL CX, R9
109
110 #define ROUND3(a, b, c, d, index, const, shift) \
111 LEAL const(a)(R8*1),a; \
112 MOVL (index*4)(SI),R8; \
113 XORL d, R9; \
114 XORL b, R9; \
115 ADDL R9, a; \
116 ROLL $shift, a; \
117 MOVL b, R9; \
118 ADDL b, a
119
120 ROUND3(AX,BX,CX,DX, 8,0xfffa3942, 4);
121 ROUND3(DX,AX,BX,CX,11,0x8771f681,11);
122 ROUND3(CX,DX,AX,BX,14,0x6d9d6122,16);
123 ROUND3(BX,CX,DX,AX, 1,0xfde5380c,23);
124 ROUND3(AX,BX,CX,DX, 4,0xa4beea44, 4);
125 ROUND3(DX,AX,BX,CX, 7,0x4bdecfa9,11);
126 ROUND3(CX,DX,AX,BX,10,0xf6bb4b60,16);
127 ROUND3(BX,CX,DX,AX,13,0xbebfbc70,23);
128 ROUND3(AX,BX,CX,DX, 0,0x289b7ec6, 4);
129 ROUND3(DX,AX,BX,CX, 3,0xeaa127fa,11);
130 ROUND3(CX,DX,AX,BX, 6,0xd4ef3085,16);
131 ROUND3(BX,CX,DX,AX, 9, 0x4881d05,23);
132 ROUND3(AX,BX,CX,DX,12,0xd9d4d039, 4);
133 ROUND3(DX,AX,BX,CX,15,0xe6db99e5,11);
134 ROUND3(CX,DX,AX,BX, 2,0x1fa27cf8,16);
135 ROUND3(BX,CX,DX,AX, 0,0xc4ac5665,23);
136
137 MOVL (0*4)(SI), R8
138 MOVL $0xffffffff, R9
139 XORL DX, R9
140
141 #define ROUND4(a, b, c, d, index, const, shift) \
142 LEAL const(a)(R8*1),a; \
143 ORL b, R9; \
144 XORL c, R9; \
145 ADDL R9, a; \
146 MOVL (index*4)(SI),R8; \
147 MOVL $0xffffffff, R9; \
148 ROLL $shift, a; \
149 XORL c, R9; \
150 ADDL b, a
151
152 ROUND4(AX,BX,CX,DX, 7,0xf4292244, 6);
153 ROUND4(DX,AX,BX,CX,14,0x432aff97,10);
154 ROUND4(CX,DX,AX,BX, 5,0xab9423a7,15);
155 ROUND4(BX,CX,DX,AX,12,0xfc93a039,21);
156 ROUND4(AX,BX,CX,DX, 3,0x655b59c3, 6);
157 ROUND4(DX,AX,BX,CX,10,0x8f0ccc92,10);
158 ROUND4(CX,DX,AX,BX, 1,0xffeff47d,15);
159 ROUND4(BX,CX,DX,AX, 8,0x85845dd1,21);
160 ROUND4(AX,BX,CX,DX,15,0x6fa87e4f, 6);
161 ROUND4(DX,AX,BX,CX, 6,0xfe2ce6e0,10);
162 ROUND4(CX,DX,AX,BX,13,0xa3014314,15);
163 ROUND4(BX,CX,DX,AX, 4,0x4e0811a1,21);
164 ROUND4(AX,BX,CX,DX,11,0xf7537e82, 6);
165 ROUND4(DX,AX,BX,CX, 2,0xbd3af235,10);
166 ROUND4(CX,DX,AX,BX, 9,0x2ad7d2bb,15);
167 ROUND4(BX,CX,DX,AX, 0,0xeb86d391,21);
168
169 ADDL R12, AX
170 ADDL R13, BX
171 ADDL R14, CX
172 ADDL R11, DX
173
174 ADDQ $64, SI
175 CMPQ SI, DI
176 JB loop
177
178 end:
179 MOVL dig+0(FP), R11
180 MOVL AX, (0*4)(R11)
181 MOVL BX, (1*4)(R11)
182 MOVL CX, (2*4)(R11)
183 MOVL DX, (3*4)(R11)
184 RET
View as plain text