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