sys/crypto/aesni/aesencdec_amd64.S

   1 /*-
   2  * Copyright (c) 2010 Konstantin Belousov <kib@FreeBSD.org>
   3  * All rights reserved.
   4  *
   5  * Redistribution and use in source and binary forms, with or without
   6  * modification, are permitted provided that the following conditions
   7  * are met:
   8  * 1. Redistributions of source code must retain the above copyright
   9  *    notice, this list of conditions and the following disclaimer.
  10  * 2. Redistributions in binary form must reproduce the above copyright
  11  *    notice, this list of conditions and the following disclaimer in the
  12  *    documentation and/or other materials provided with the distribution.
  13  *
  14  * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND
  15  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  16  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  17  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE
  18  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  19  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  20  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  21  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  22  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  23  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  24  * SUCH DAMAGE.
  25  *
  26  * FreeBSD: src/sys/crypto/aesni/aesencdec_amd64.S,v 1.1 2010/07/23 11:00:46 kib Exp
  27  */
  28
  29 #include <machine/asmacros.h>
  30
  31         .text
  32
  33 ENTRY(aesni_enc)
  34         .cfi_startproc
  35         movdqu  (%rdx),%xmm0
  36         cmpq    $0,%r8
  37         je      1f
  38         movdqu  (%r8),%xmm1     /* unaligned load into reg */
  39         pxor    %xmm1,%xmm0     /* pxor otherwise can fault on iv */
  40 1:
  41         pxor    (%rsi),%xmm0
  42 2:
  43         addq    $0x10,%rsi
  44 //      aesenc  (%rsi),%xmm0
  45         .byte   0x66,0x0f,0x38,0xdc,0x06
  46         decl    %edi
  47         jne     2b
  48         addq    $0x10,%rsi
  49 //      aesenclast (%rsi),%xmm0
  50         .byte   0x66,0x0f,0x38,0xdd,0x06
  51         movdqu  %xmm0,(%rcx)
  52         retq
  53         .cfi_endproc
  54 END(aesni_enc)
  55
  56 ENTRY(aesni_dec)
  57         .cfi_startproc
  58         movdqu  (%rdx),%xmm0
  59         pxor    (%rsi),%xmm0
  60 1:
  61         addq    $0x10,%rsi
  62 //      aesdec  (%rsi),%xmm0
  63         .byte   0x66,0x0f,0x38,0xde,0x06
  64         decl    %edi
  65         jne     1b
  66         addq    $0x10,%rsi
  67 //      aesdeclast (%rsi),%xmm0
  68         .byte   0x66,0x0f,0x38,0xdf,0x06
  69         cmpq    $0,%r8
  70         je      2f
  71         movdqu  (%r8),%xmm1
  72         pxor    %xmm1,%xmm0
  73 2:
  74         movdqu  %xmm0,(%rcx)
  75         retq
  76         .cfi_endproc
  77 END(aesni_dec)
  78
  79 ENTRY(aesni_decrypt_cbc)
  80         .cfi_startproc
  81         shrq    $4,%rdx
  82         movdqu  (%r8),%xmm1
  83 1:
  84         movdqu  (%rcx),%xmm0
  85         movdqa  %xmm0,%xmm2
  86         pxor    (%rsi),%xmm0
  87         cmpl    $12,%edi
  88 //      aesdec  0x10(%rsi),%xmm0
  89         .byte   0x66,0x0f,0x38,0xde,0x46,0x10
  90 //      aesdec  0x20(%rsi),%xmm0
  91         .byte   0x66,0x0f,0x38,0xde,0x46,0x20
  92 //      aesdec  0x30(%rsi),%xmm0
  93         .byte   0x66,0x0f,0x38,0xde,0x46,0x30
  94 //      aesdec  0x40(%rsi),%xmm0
  95         .byte   0x66,0x0f,0x38,0xde,0x46,0x40
  96 //      aesdec  0x50(%rsi),%xmm0
  97         .byte   0x66,0x0f,0x38,0xde,0x46,0x50
  98 //      aesdec  0x60(%rsi),%xmm0
  99         .byte   0x66,0x0f,0x38,0xde,0x46,0x60
 100 //      aesdec  0x70(%rsi),%xmm0
 101         .byte   0x66,0x0f,0x38,0xde,0x46,0x70
 102 //      aesdec  0x80(%rsi),%xmm0
 103         .byte   0x66,0x0f,0x38,0xde,0x86,0x80,0x00,0x00,0x00
 104 //      aesdec  0x90(%rsi),%xmm0
 105         .byte   0x66,0x0f,0x38,0xde,0x86,0x90,0x00,0x00,0x00
 106         jge     2f
 107 //      aesdeclast 0xa0(%rsi),%xmm0
 108         .byte   0x66,0x0f,0x38,0xdf,0x86,0xa0,0x00,0x00,0x00
 109         jmp     4f
 110 2:
 111 //      aesdec  0xa0(%rsi),%xmm0
 112         .byte   0x66,0x0f,0x38,0xde,0x86,0xa0,0x00,0x00,0x00
 113 //      aesdec  0xb0(%rsi),%xmm0
 114         .byte   0x66,0x0f,0x38,0xde,0x86,0xb0,0x00,0x00,0x00
 115         jg      3f
 116 //      aesdeclast 0xc0(%rsi),%xmm0
 117         .byte   0x66,0x0f,0x38,0xdf,0x86,0xc0,0x00,0x00,0x00
 118         jmp     4f
 119 3:
 120 //      aesdec  0xc0(%rsi),%xmm0
 121         .byte   0x66,0x0f,0x38,0xde,0x86,0xc0,0x00,0x00,0x00
 122 //      aesdec  0xd0(%rsi),%xmm0
 123         .byte   0x66,0x0f,0x38,0xde,0x86,0xd0,0x00,0x00,0x00
 124 //      aesdeclast 0xe0(%rsi),%xmm0
 125         .byte   0x66,0x0f,0x38,0xdf,0x86,0xe0,0x00,0x00,0x00
 126 4:
 127         pxor    %xmm1,%xmm0
 128         movdqu  %xmm0,(%rcx)
 129         movdqa  %xmm2,%xmm1     // iv
 130         addq    $0x10,%rcx
 131         decq    %rdx
 132         jne     1b
 133         retq
 134         .cfi_endproc
 135 END(aesni_decrypt_cbc)