/*
* MD5 hash in x86-64 assembly
*
* Copyright (c) 2016 Project Nayuki. (MIT License)
* https://www.nayuki.io/page/fast-md5-hash-implementation-in-x86-assembly
*
* Permission is hereby granted, free of charge, to any person obtaining a copy of
* this software and associated documentation files (the "Software"), to deal in
* the Software without restriction, including without limitation the rights to
* use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of
* the Software, and to permit persons to whom the Software is furnished to do so,
* subject to the following conditions:
* - The above copyright notice and this permission notice shall be included in
* all copies or substantial portions of the Software.
* - The Software is provided "as is", without warranty of any kind, express or
* implied, including but not limited to the warranties of merchantability,
* fitness for a particular purpose and noninfringement. In no event shall the
* authors or copyright holders be liable for any claim, damages or other
* liability, whether in an action of contract, tort or otherwise, arising from,
* out of or in connection with the Software or the use or other dealings in the
* Software.
*/
/* void md5_compress(uint32_t state[4], const uint8_t block[64]) */
#ifdef __APPLE__
.globl _md5_compress
_md5_compress:
#else
.globl md5_compress
md5_compress:
#endif
/*
* Storage usage:
* Bytes Location Description
* 4 eax MD5 state variable A
* 4 ebx MD5 state variable B
* 4 ecx MD5 state variable C
* 4 edx MD5 state variable D
* 4 esi Temporary for calculation per round
* 4 edi Temporary for calculation per round
* 8 rbp Base address of block array argument (read-only)
* 8 r8 Base address of state array argument (read-only)
* 16 xmm0 Caller's value of rbx (only low 64 bits are used)
* 16 xmm1 Caller's value of rbp (only low 64 bits are used)
*/
#define ROUND0(a, b, c, d, k, s, t) \
movl %c, %esi; \
addl (k*4)(%rbp), %a; \
xorl %d, %esi; \
andl %b, %esi; \
xorl %d, %esi; \
leal t(%esi,%a), %a; \
roll $s, %a; \
addl %b, %a;
#define ROUND1(a, b, c, d, k, s, t) \
movl %d, %esi; \
movl %d, %edi; \
addl (k*4)(%rbp), %a; \
notl %esi; \
andl %b, %edi; \
andl %c, %esi; \
orl %edi, %esi; \
leal t(%esi,%a), %a; \
roll $s, %a; \
addl %b, %a;
#define ROUND2(a, b, c, d, k, s, t) \
movl %c, %esi; \
addl (k*4)(%rbp), %a; \
xorl %d, %esi; \
xorl %b, %esi; \
leal t(%esi,%a), %a; \
roll $s, %a; \
addl %b, %a;
#define ROUND3(a, b, c, d, k, s, t) \
movl %d, %esi; \
not %esi; \
addl (k*4)(%rbp), %a; \
orl %b, %esi; \
xorl %c, %esi; \
leal t(%esi,%a), %a; \
roll $s, %a; \
addl %b, %a;
/* Save registers */
movq %rbx, %xmm0
movq %rbp, %xmm1
/* Load arguments */
movq %rsi, %rbp
movl 0(%rdi), %eax /* a */
movl 4(%rdi), %ebx /* b */
movl 8(%rdi), %ecx /* c */
movl 12(%rdi), %edx /* d */
movq %rdi, %r8
/* 64 rounds of hashing */
ROUND0(eax, ebx, ecx, edx, 0, 7, -0x28955B88)
ROUND0(edx, eax, ebx, ecx, 1, 12, -0x173848AA)
ROUND0(ecx, edx, eax, ebx, 2, 17, 0x242070DB)
ROUND0(ebx, ecx, edx, eax, 3, 22, -0x3E423112)
ROUND0(eax, ebx, ecx, edx, 4, 7, -0x0A83F051)
ROUND0(edx, eax, ebx, ecx, 5, 12, 0x4787C62A)
ROUND0(ecx, edx, eax, ebx, 6, 17, -0x57CFB9ED)
ROUND0(ebx, ecx, edx, eax, 7, 22, -0x02B96AFF)
ROUND0(eax, ebx, ecx, edx, 8, 7, 0x698098D8)
ROUND0(edx, eax, ebx, ecx, 9, 12, -0x74BB0851)
ROUND0(ecx, edx, eax, ebx, 10, 17, -0x0000A44F)
ROUND0(ebx, ecx, edx, eax, 11, 22, -0x76A32842)
ROUND0(eax, ebx, ecx, edx, 12, 7, 0x6B901122)
ROUND0(edx, eax, ebx, ecx, 13, 12, -0x02678E6D)
ROUND0(ecx, edx, eax, ebx, 14, 17, -0x5986BC72)
ROUND0(ebx, ecx, edx, eax, 15, 22, 0x49B40821)
ROUND1(eax, ebx, ecx, edx, 1, 5, -0x09E1DA9E)
ROUND1(edx, eax, ebx, ecx, 6, 9, -0x3FBF4CC0)
ROUND1(ecx, edx, eax, ebx, 11, 14, 0x265E5A51)
ROUND1(ebx, ecx, edx, eax, 0, 20, -0x16493856)
ROUND1(eax, ebx, ecx, edx, 5, 5, -0x29D0EFA3)
ROUND1(edx, eax, ebx, ecx, 10, 9, 0x02441453)
ROUND1(ecx, edx, eax, ebx, 15, 14, -0x275E197F)
ROUND1(ebx, ecx, edx, eax, 4, 20, -0x182C0438)
ROUND1(eax, ebx, ecx, edx, 9, 5, 0x21E1CDE6)
ROUND1(edx, eax, ebx, ecx, 14, 9, -0x3CC8F82A)
ROUND1(ecx, edx, eax, ebx, 3, 14, -0x0B2AF279)
ROUND1(ebx, ecx, edx, eax, 8, 20, 0x455A14ED)
ROUND1(eax, ebx, ecx, edx, 13, 5, -0x561C16FB)
ROUND1(edx, eax, ebx, ecx, 2, 9, -0x03105C08)
ROUND1(ecx, edx, eax, ebx, 7, 14, 0x676F02D9)
ROUND1(ebx, ecx, edx, eax, 12, 20, -0x72D5B376)
ROUND2(eax, ebx, ecx, edx, 5, 4, -0x0005C6BE)
ROUND2(edx, eax, ebx, ecx, 8, 11, -0x788E097F)
ROUND2(ecx, edx, eax, ebx, 11, 16, 0x6D9D6122)
ROUND2(ebx, ecx, edx, eax, 14, 23, -0x021AC7F4)
ROUND2(eax, ebx, ecx, edx, 1, 4, -0x5B4115BC)
ROUND2(edx, eax, ebx, ecx, 4, 11, 0x4BDECFA9)
ROUND2(ecx, edx, eax, ebx, 7, 16, -0x0944B4A0)
ROUND2(ebx, ecx, edx, eax, 10, 23, -0x41404390)
ROUND2(eax, ebx, ecx, edx, 13, 4, 0x289B7EC6)
ROUND2(edx, eax, ebx, ecx, 0, 11, -0x155ED806)
ROUND2(ecx, edx, eax, ebx, 3, 16, -0x2B10CF7B)
ROUND2(ebx, ecx, edx, eax, 6, 23, 0x04881D05)
ROUND2(eax, ebx, ecx, edx, 9, 4, -0x262B2FC7)
ROUND2(edx, eax, ebx, ecx, 12, 11, -0x1924661B)
ROUND2(ecx, edx, eax, ebx, 15, 16, 0x1FA27CF8)
ROUND2(ebx, ecx, edx, eax, 2, 23, -0x3B53A99B)
ROUND3(eax, ebx, ecx, edx, 0, 6, -0x0BD6DDBC)
ROUND3(edx, eax, ebx, ecx, 7, 10, 0x432AFF97)
ROUND3(ecx, edx, eax, ebx, 14, 15, -0x546BDC59)
ROUND3(ebx, ecx, edx, eax, 5, 21, -0x036C5FC7)
ROUND3(eax, ebx, ecx, edx, 12, 6, 0x655B59C3)
ROUND3(edx, eax, ebx, ecx, 3, 10, -0x70F3336E)
ROUND3(ecx, edx, eax, ebx, 10, 15, -0x00100B83)
ROUND3(ebx, ecx, edx, eax, 1, 21, -0x7A7BA22F)
ROUND3(eax, ebx, ecx, edx, 8, 6, 0x6FA87E4F)
ROUND3(edx, eax, ebx, ecx, 15, 10, -0x01D31920)
ROUND3(ecx, edx, eax, ebx, 6, 15, -0x5CFEBCEC)
ROUND3(ebx, ecx, edx, eax, 13, 21, 0x4E0811A1)
ROUND3(eax, ebx, ecx, edx, 4, 6, -0x08AC817E)
ROUND3(edx, eax, ebx, ecx, 11, 10, -0x42C50DCB)
ROUND3(ecx, edx, eax, ebx, 2, 15, 0x2AD7D2BB)
ROUND3(ebx, ecx, edx, eax, 9, 21, -0x14792C6F)
/* Save updated state */
addl %eax, 0(%r8)
addl %ebx, 4(%r8)
addl %ecx, 8(%r8)
addl %edx, 12(%r8)
/* Restore registers */
movq %xmm0, %rbx
movq %xmm1, %rbp
retq