ceph/src/crypto/isa-l/isa-l_crypto/aes/cbc_enc_192_x8_sb.asm

   1 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
   2 ;  Copyright(c) 2011-2016 Intel Corporation All rights reserved.
   3 ;
   4 ;  Redistribution and use in source and binary forms, with or without
   5 ;  modification, are permitted provided that the following conditions
   6 ;  are met:
   7 ;    * Redistributions of source code must retain the above copyright
   8 ;      notice, this list of conditions and the following disclaimer.
   9 ;    * Redistributions in binary form must reproduce the above copyright
  10 ;      notice, this list of conditions and the following disclaimer in
  11 ;      the documentation and/or other materials provided with the
  12 ;      distribution.
  13 ;    * Neither the name of Intel Corporation nor the names of its
  14 ;      contributors may be used to endorse or promote products derived
  15 ;      from this software without specific prior written permission.
  16 ;
  17 ;  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
  18 ;  "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
  19 ;  LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
  20 ;  A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
  21 ;  OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
  22 ;  SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
  23 ;  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
  24 ;  DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
  25 ;  THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  26 ;  (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
  27 ;  OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  28 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
  29
  30 ;;; routine to do a 192 bit CBC AES encrypt
  31 ;; clobbers all registers except for ARG1 and rbp
  32 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
  33 ;;; Updates In and Out pointers at end
  34 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
  35 ;;void aes_cbc_enc_192_x8(void      *in,
  36 ;;                        uint8_t   *IV,
  37 ;;                        uint8_t    keys,
  38 ;;                        void      *out,
  39 ;;                        uint64_t   len_bytes);
  40 ; arg 1: IN:   pointer to input (cipher text)
  41 ; arg 2: IV:   pointer to IV
  42 ; arg 3: KEYS: pointer to keys
  43 ; arg 4: OUT:  pointer to output (plain text)
  44 ; arg 5: LEN:  length in bytes (multiple of 16)
  45 ;; clobbers all registers except for ARG1 and rbp
  46
  47 %include "reg_sizes.asm"
  48
  49 %ifidn __OUTPUT_FORMAT__, elf64
  50 %define IN0             rdi
  51 %define IN              rdi
  52 %define IV              rsi
  53 %define KEYS    rdx
  54 %define OUT     rcx
  55 %define LEN             r8
  56 %define KEYS0   rdx
  57 %define OUT0    rcx
  58 %define func(x) x:
  59 %define FUNC_SAVE
  60 %define FUNC_RESTORE
  61 %endif
  62
  63 %ifidn __OUTPUT_FORMAT__, win64
  64 %define IN0             rcx
  65 %define IN              rcx
  66 %define IV              rdx
  67 %define KEYS0   r8
  68 %define OUT0    r9
  69 %define KEYS    r8
  70 %define OUT             r9
  71 %define LEN             r10
  72 %define PS              8
  73 %define stack_size      10*16 + 1*8     ; must be an odd multiple of 8
  74 %define arg(x)          [rsp + stack_size + PS + PS*x]
  75
  76 %define func(x) proc_frame x
  77 %macro FUNC_SAVE 0
  78         alloc_stack     stack_size
  79         vmovdqa [rsp + 0*16], xmm6
  80         vmovdqa [rsp + 1*16], xmm7
  81         vmovdqa [rsp + 2*16], xmm8
  82         vmovdqa [rsp + 3*16], xmm9
  83         vmovdqa [rsp + 4*16], xmm10
  84         vmovdqa [rsp + 5*16], xmm11
  85         vmovdqa [rsp + 6*16], xmm12
  86         vmovdqa [rsp + 7*16], xmm13
  87         vmovdqa [rsp + 8*16], xmm14
  88         vmovdqa [rsp + 9*16], xmm15
  89         end_prolog
  90         mov     LEN, arg(4)
  91 %endmacro
  92
  93 %macro FUNC_RESTORE 0
  94         vmovdqa xmm6, [rsp + 0*16]
  95         vmovdqa xmm7, [rsp + 1*16]
  96         vmovdqa xmm8, [rsp + 2*16]
  97         vmovdqa xmm9, [rsp + 3*16]
  98         vmovdqa xmm10, [rsp + 4*16]
  99         vmovdqa xmm11, [rsp + 5*16]
 100         vmovdqa xmm12, [rsp + 6*16]
 101         vmovdqa xmm13, [rsp + 7*16]
 102         vmovdqa xmm14, [rsp + 8*16]
 103         vmovdqa xmm15, [rsp + 9*16]
 104         add     rsp, stack_size
 105 %endmacro
 106 %endif
 107
 108 %define KEY_ROUNDS 13
 109 %define XMM_USAGE    (16)
 110 %DEFINE UNROLLED_LOOPS (3)
 111 %define PARALLEL_BLOCKS (UNROLLED_LOOPS)
 112
 113 ; instruction set specific operation definitions
 114 %define MOVDQ         vmovdqu
 115 %macro PXOR 2
 116    vpxor %1, %1, %2
 117 %endm
 118
 119 %macro AES_ENC 2
 120   vaesenc %1, %1, %2
 121 %endm
 122
 123 %macro AES_ENC_LAST 2
 124   vaesenclast %1, %1, %2
 125 %endm
 126
 127 %include "cbc_common.asm"
 128
 129 mk_global aes_cbc_enc_192_x8, function
 130 func(aes_cbc_enc_192_x8)
 131         endbranch
 132         FUNC_SAVE
 133
 134         mov     IDX, 0
 135
 136         FILL_KEY_CACHE   CKEY_CNT, FIRST_CKEY, KEYS, MOVDQ
 137         CBC_ENC_INIT     FIRST_XDATA, TMP, MOVDQ, PXOR, IV, IN, IDX
 138
 139 main_loop:
 140         CBC_ENC_SUBLOOP  KEY_ROUNDS, UNROLLED_LOOPS, FIRST_XDATA, MOVDQ, PXOR, AES_ENC, AES_ENC_LAST, TMP, TMP_CNT,     FIRST_CKEY, CKEY_CNT, KEYS, CACHED_KEYS, IN, OUT, IDX, LEN
 141         jne     main_loop
 142
 143 done:
 144         FUNC_RESTORE
 145         ret
 146
 147 endproc_frame