arch/arm/lib/memcpy.S

   1 /*
   2  *  linux/arch/arm/lib/memcpy.S
   3  *
   4  *  Author:     Nicolas Pitre
   5  *  Created:    Sep 28, 2005
   6  *  Copyright:  MontaVista Software, Inc.
   7  *
   8  *  This program is free software; you can redistribute it and/or modify
   9  *  it under the terms of the GNU General Public License version 2 as
  10  *  published by the Free Software Foundation.
  11  */
  12
  13 #include <asm/assembler.h>
  14
  15 #define W(instr)        instr
  16
  17 #define LDR1W_SHIFT     0
  18 #define STR1W_SHIFT     0
  19
  20         .macro ldr1w ptr reg abort
  21         W(ldr) \reg, [\ptr], #4
  22         .endm
  23
  24         .macro ldr4w ptr reg1 reg2 reg3 reg4 abort
  25         ldmia \ptr!, {\reg1, \reg2, \reg3, \reg4}
  26         .endm
  27
  28         .macro ldr8w ptr reg1 reg2 reg3 reg4 reg5 reg6 reg7 reg8 abort
  29         ldmia \ptr!, {\reg1, \reg2, \reg3, \reg4, \reg5, \reg6, \reg7, \reg8}
  30         .endm
  31
  32         .macro ldr1b ptr reg cond=al abort
  33         ldr\cond\()b \reg, [\ptr], #1
  34         .endm
  35
  36         .macro str1w ptr reg abort
  37         W(str) \reg, [\ptr], #4
  38         .endm
  39
  40         .macro str8w ptr reg1 reg2 reg3 reg4 reg5 reg6 reg7 reg8 abort
  41         stmia \ptr!, {\reg1, \reg2, \reg3, \reg4, \reg5, \reg6, \reg7, \reg8}
  42         .endm
  43
  44         .macro str1b ptr reg cond=al abort
  45         str\cond\()b \reg, [\ptr], #1
  46         .endm
  47
  48         .macro enter reg1 reg2
  49         stmdb sp!, {r0, \reg1, \reg2}
  50         .endm
  51
  52         .macro exit reg1 reg2
  53         ldmfd sp!, {r0, \reg1, \reg2}
  54         .endm
  55
  56         .text
  57
  58 /* Prototype: void *memcpy(void *dest, const void *src, size_t n); */
  59
  60 .globl memcpy
  61 memcpy:
  62
  63                 cmp     r0, r1
  64                 moveq   pc, lr
  65
  66                 enter   r4, lr
  67
  68                 subs    r2, r2, #4
  69                 blt     8f
  70                 ands    ip, r0, #3
  71         PLD(    pld     [r1, #0]                )
  72                 bne     9f
  73                 ands    ip, r1, #3
  74                 bne     10f
  75
  76 1:              subs    r2, r2, #(28)
  77                 stmfd   sp!, {r5 - r8}
  78                 blt     5f
  79
  80         CALGN(  ands    ip, r0, #31             )
  81         CALGN(  rsb     r3, ip, #32             )
  82         CALGN(  sbcnes  r4, r3, r2              )  @ C is always set here
  83         CALGN(  bcs     2f                      )
  84         CALGN(  adr     r4, 6f                  )
  85         CALGN(  subs    r2, r2, r3              )  @ C gets set
  86         CALGN(  add     pc, r4, ip              )
  87
  88         PLD(    pld     [r1, #0]                )
  89 2:      PLD(    subs    r2, r2, #96             )
  90         PLD(    pld     [r1, #28]               )
  91         PLD(    blt     4f                      )
  92         PLD(    pld     [r1, #60]               )
  93         PLD(    pld     [r1, #92]               )
  94
  95 3:      PLD(    pld     [r1, #124]              )
  96 4:              ldr8w   r1, r3, r4, r5, r6, r7, r8, ip, lr, abort=20f
  97                 subs    r2, r2, #32
  98                 str8w   r0, r3, r4, r5, r6, r7, r8, ip, lr, abort=20f
  99                 bge     3b
 100         PLD(    cmn     r2, #96                 )
 101         PLD(    bge     4b                      )
 102
 103 5:              ands    ip, r2, #28
 104                 rsb     ip, ip, #32
 105 #if LDR1W_SHIFT > 0
 106                 lsl     ip, ip, #LDR1W_SHIFT
 107 #endif
 108                 addne   pc, pc, ip              @ C is always clear here
 109                 b       7f
 110 6:
 111                 .rept   (1 << LDR1W_SHIFT)
 112                 W(nop)
 113                 .endr
 114                 ldr1w   r1, r3, abort=20f
 115                 ldr1w   r1, r4, abort=20f
 116                 ldr1w   r1, r5, abort=20f
 117                 ldr1w   r1, r6, abort=20f
 118                 ldr1w   r1, r7, abort=20f
 119                 ldr1w   r1, r8, abort=20f
 120                 ldr1w   r1, lr, abort=20f
 121
 122 #if LDR1W_SHIFT < STR1W_SHIFT
 123                 lsl     ip, ip, #STR1W_SHIFT - LDR1W_SHIFT
 124 #elif LDR1W_SHIFT > STR1W_SHIFT
 125                 lsr     ip, ip, #LDR1W_SHIFT - STR1W_SHIFT
 126 #endif
 127                 add     pc, pc, ip
 128                 nop
 129                 .rept   (1 << STR1W_SHIFT)
 130                 W(nop)
 131                 .endr
 132                 str1w   r0, r3, abort=20f
 133                 str1w   r0, r4, abort=20f
 134                 str1w   r0, r5, abort=20f
 135                 str1w   r0, r6, abort=20f
 136                 str1w   r0, r7, abort=20f
 137                 str1w   r0, r8, abort=20f
 138                 str1w   r0, lr, abort=20f
 139
 140         CALGN(  bcs     2b                      )
 141
 142 7:              ldmfd   sp!, {r5 - r8}
 143
 144 8:              movs    r2, r2, lsl #31
 145                 ldr1b   r1, r3, ne, abort=21f
 146                 ldr1b   r1, r4, cs, abort=21f
 147                 ldr1b   r1, ip, cs, abort=21f
 148                 str1b   r0, r3, ne, abort=21f
 149                 str1b   r0, r4, cs, abort=21f
 150                 str1b   r0, ip, cs, abort=21f
 151
 152                 exit    r4, pc
 153
 154 9:              rsb     ip, ip, #4
 155                 cmp     ip, #2
 156                 ldr1b   r1, r3, gt, abort=21f
 157                 ldr1b   r1, r4, ge, abort=21f
 158                 ldr1b   r1, lr, abort=21f
 159                 str1b   r0, r3, gt, abort=21f
 160                 str1b   r0, r4, ge, abort=21f
 161                 subs    r2, r2, ip
 162                 str1b   r0, lr, abort=21f
 163                 blt     8b
 164                 ands    ip, r1, #3
 165                 beq     1b
 166
 167 10:             bic     r1, r1, #3
 168                 cmp     ip, #2
 169                 ldr1w   r1, lr, abort=21f
 170                 beq     17f
 171                 bgt     18f
 172
 173
 174                 .macro  forward_copy_shift pull push
 175
 176                 subs    r2, r2, #28
 177                 blt     14f
 178
 179         CALGN(  ands    ip, r0, #31             )
 180         CALGN(  rsb     ip, ip, #32             )
 181         CALGN(  sbcnes  r4, ip, r2              )  @ C is always set here
 182         CALGN(  subcc   r2, r2, ip              )
 183         CALGN(  bcc     15f                     )
 184
 185 11:             stmfd   sp!, {r5 - r9}
 186
 187         PLD(    pld     [r1, #0]                )
 188         PLD(    subs    r2, r2, #96             )
 189         PLD(    pld     [r1, #28]               )
 190         PLD(    blt     13f                     )
 191         PLD(    pld     [r1, #60]               )
 192         PLD(    pld     [r1, #92]               )
 193
 194 12:     PLD(    pld     [r1, #124]              )
 195 13:             ldr4w   r1, r4, r5, r6, r7, abort=19f
 196                 mov     r3, lr, pull #\pull
 197                 subs    r2, r2, #32
 198                 ldr4w   r1, r8, r9, ip, lr, abort=19f
 199                 orr     r3, r3, r4, push #\push
 200                 mov     r4, r4, pull #\pull
 201                 orr     r4, r4, r5, push #\push
 202                 mov     r5, r5, pull #\pull
 203                 orr     r5, r5, r6, push #\push
 204                 mov     r6, r6, pull #\pull
 205                 orr     r6, r6, r7, push #\push
 206                 mov     r7, r7, pull #\pull
 207                 orr     r7, r7, r8, push #\push
 208                 mov     r8, r8, pull #\pull
 209                 orr     r8, r8, r9, push #\push
 210                 mov     r9, r9, pull #\pull
 211                 orr     r9, r9, ip, push #\push
 212                 mov     ip, ip, pull #\pull
 213                 orr     ip, ip, lr, push #\push
 214                 str8w   r0, r3, r4, r5, r6, r7, r8, r9, ip, , abort=19f
 215                 bge     12b
 216         PLD(    cmn     r2, #96                 )
 217         PLD(    bge     13b                     )
 218
 219                 ldmfd   sp!, {r5 - r9}
 220
 221 14:             ands    ip, r2, #28
 222                 beq     16f
 223
 224 15:             mov     r3, lr, pull #\pull
 225                 ldr1w   r1, lr, abort=21f
 226                 subs    ip, ip, #4
 227                 orr     r3, r3, lr, push #\push
 228                 str1w   r0, r3, abort=21f
 229                 bgt     15b
 230         CALGN(  cmp     r2, #0                  )
 231         CALGN(  bge     11b                     )
 232
 233 16:             sub     r1, r1, #(\push / 8)
 234                 b       8b
 235
 236                 .endm
 237
 238
 239                 forward_copy_shift      pull=8  push=24
 240
 241 17:             forward_copy_shift      pull=16 push=16
 242
 243 18:             forward_copy_shift      pull=24 push=8