Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1 | /* $Id: memmove.S,v 1.2 2001/07/27 11:51:09 gniibe Exp $ |
| 2 | * |
| 3 | * "memmove" implementation of SuperH |
| 4 | * |
| 5 | * Copyright (C) 1999 Niibe Yutaka |
| 6 | * |
| 7 | */ |
| 8 | |
| 9 | /* |
| 10 | * void *memmove(void *dst, const void *src, size_t n); |
| 11 | * The memory areas may overlap. |
| 12 | */ |
| 13 | |
| 14 | #include <linux/linkage.h> |
| 15 | ENTRY(memmove) |
| 16 | ! if dest > src, call memcpy (it copies in decreasing order) |
| 17 | cmp/hi r5,r4 |
| 18 | bf 1f |
| 19 | mov.l 2f,r0 |
| 20 | jmp @r0 |
| 21 | nop |
| 22 | .balign 4 |
| 23 | 2: .long memcpy |
| 24 | 1: |
| 25 | sub r5,r4 ! From here, r4 has the distance to r0 |
| 26 | tst r6,r6 |
| 27 | bt/s 9f ! if n=0, do nothing |
| 28 | mov r5,r0 |
| 29 | add r6,r5 |
| 30 | mov #12,r1 |
| 31 | cmp/gt r6,r1 |
| 32 | bt/s 8f ! if it's too small, copy a byte at once |
| 33 | add #-1,r4 |
| 34 | add #1,r4 |
| 35 | ! |
| 36 | ! [ ... ] DST [ ... ] SRC |
| 37 | ! [ ... ] [ ... ] |
| 38 | ! : : |
| 39 | ! r0+r4--> [ ... ] r0 --> [ ... ] |
| 40 | ! : : |
| 41 | ! [ ... ] [ ... ] |
| 42 | ! r5 --> |
| 43 | ! |
| 44 | mov r4,r1 |
| 45 | mov #3,r2 |
| 46 | and r2,r1 |
| 47 | shll2 r1 |
| 48 | mov r0,r3 ! Save the value on R0 to R3 |
| 49 | mova jmptable,r0 |
| 50 | add r1,r0 |
| 51 | mov.l @r0,r1 |
| 52 | jmp @r1 |
| 53 | mov r3,r0 ! and back to R0 |
| 54 | .balign 4 |
| 55 | jmptable: |
| 56 | .long case0 |
| 57 | .long case1 |
| 58 | .long case2 |
| 59 | .long case3 |
| 60 | |
| 61 | ! copy a byte at once |
| 62 | 8: mov.b @r0+,r1 |
| 63 | cmp/hs r5,r0 |
| 64 | bf/s 8b ! while (r0<r5) |
| 65 | mov.b r1,@(r0,r4) |
| 66 | add #1,r4 |
| 67 | 9: |
| 68 | add r4,r0 |
| 69 | rts |
| 70 | sub r6,r0 |
| 71 | |
| 72 | case_none: |
| 73 | bra 8b |
| 74 | add #-1,r4 |
| 75 | |
| 76 | case0: |
| 77 | ! |
| 78 | ! GHIJ KLMN OPQR --> GHIJ KLMN OPQR |
| 79 | ! |
| 80 | ! First, align to long word boundary |
| 81 | mov r0,r3 |
| 82 | and r2,r3 |
| 83 | tst r3,r3 |
| 84 | bt/s 2f |
| 85 | add #-1,r4 |
| 86 | mov #4,r2 |
| 87 | sub r3,r2 |
| 88 | 1: dt r2 |
| 89 | mov.b @r0+,r1 |
| 90 | bf/s 1b |
| 91 | mov.b r1,@(r0,r4) |
| 92 | ! |
| 93 | 2: ! Second, copy a long word at once |
| 94 | add #-3,r4 |
| 95 | add #-3,r5 |
| 96 | 3: mov.l @r0+,r1 |
| 97 | cmp/hs r5,r0 |
| 98 | bf/s 3b |
| 99 | mov.l r1,@(r0,r4) |
| 100 | add #3,r5 |
| 101 | ! |
| 102 | ! Third, copy a byte at once, if necessary |
| 103 | cmp/eq r5,r0 |
| 104 | bt/s 9b |
| 105 | add #4,r4 |
| 106 | bra 8b |
| 107 | add #-1,r4 |
| 108 | |
| 109 | case3: |
| 110 | ! |
| 111 | ! GHIJ KLMN OPQR --> ...G HIJK LMNO PQR. |
| 112 | ! |
| 113 | ! First, align to long word boundary |
| 114 | mov r0,r3 |
| 115 | and r2,r3 |
| 116 | tst r3,r3 |
| 117 | bt/s 2f |
| 118 | add #-1,r4 |
| 119 | mov #4,r2 |
| 120 | sub r3,r2 |
| 121 | 1: dt r2 |
| 122 | mov.b @r0+,r1 |
| 123 | bf/s 1b |
| 124 | mov.b r1,@(r0,r4) |
| 125 | ! |
| 126 | 2: ! Second, read a long word and write a long word at once |
| 127 | add #-2,r4 |
| 128 | mov.l @(r0,r4),r1 |
| 129 | add #-7,r5 |
| 130 | add #-4,r4 |
| 131 | ! |
| 132 | #ifdef __LITTLE_ENDIAN__ |
| 133 | shll8 r1 |
| 134 | 3: mov r1,r3 ! JIHG |
| 135 | shlr8 r3 ! xJIH |
| 136 | mov.l @r0+,r1 ! NMLK |
| 137 | mov r1,r2 |
| 138 | shll16 r2 |
| 139 | shll8 r2 ! Kxxx |
| 140 | or r2,r3 ! KJIH |
| 141 | cmp/hs r5,r0 |
| 142 | bf/s 3b |
| 143 | mov.l r3,@(r0,r4) |
| 144 | #else |
| 145 | shlr8 r1 |
| 146 | 3: mov r1,r3 ! GHIJ |
| 147 | shll8 r3 ! HIJx |
| 148 | mov.l @r0+,r1 ! KLMN |
| 149 | mov r1,r2 |
| 150 | shlr16 r2 |
| 151 | shlr8 r2 ! xxxK |
| 152 | or r2,r3 ! HIJK |
| 153 | cmp/hs r5,r0 |
| 154 | bf/s 3b |
| 155 | mov.l r3,@(r0,r4) |
| 156 | #endif |
| 157 | add #7,r5 |
| 158 | ! |
| 159 | ! Third, copy a byte at once, if necessary |
| 160 | cmp/eq r5,r0 |
| 161 | bt/s 9b |
| 162 | add #7,r4 |
| 163 | add #-3,r0 |
| 164 | bra 8b |
| 165 | add #-1,r4 |
| 166 | |
| 167 | case2: |
| 168 | ! |
| 169 | ! GHIJ KLMN OPQR --> ..GH IJKL MNOP QR.. |
| 170 | ! |
| 171 | ! First, align to word boundary |
| 172 | tst #1,r0 |
| 173 | bt/s 2f |
| 174 | add #-1,r4 |
| 175 | mov.b @r0+,r1 |
| 176 | mov.b r1,@(r0,r4) |
| 177 | ! |
| 178 | 2: ! Second, read a word and write a word at once |
| 179 | add #-1,r4 |
| 180 | add #-1,r5 |
| 181 | ! |
| 182 | 3: mov.w @r0+,r1 |
| 183 | cmp/hs r5,r0 |
| 184 | bf/s 3b |
| 185 | mov.w r1,@(r0,r4) |
| 186 | add #1,r5 |
| 187 | ! |
| 188 | ! Third, copy a byte at once, if necessary |
| 189 | cmp/eq r5,r0 |
| 190 | bt/s 9b |
| 191 | add #2,r4 |
| 192 | mov.b @r0,r1 |
| 193 | mov.b r1,@(r0,r4) |
| 194 | bra 9b |
| 195 | add #1,r0 |
| 196 | |
| 197 | case1: |
| 198 | ! |
| 199 | ! GHIJ KLMN OPQR --> .GHI JKLM NOPQ R... |
| 200 | ! |
| 201 | ! First, align to long word boundary |
| 202 | mov r0,r3 |
| 203 | and r2,r3 |
| 204 | tst r3,r3 |
| 205 | bt/s 2f |
| 206 | add #-1,r4 |
| 207 | mov #4,r2 |
| 208 | sub r3,r2 |
| 209 | 1: dt r2 |
| 210 | mov.b @r0+,r1 |
| 211 | bf/s 1b |
| 212 | mov.b r1,@(r0,r4) |
| 213 | ! |
| 214 | 2: ! Second, read a long word and write a long word at once |
| 215 | mov.l @(r0,r4),r1 |
| 216 | add #-7,r5 |
| 217 | add #-4,r4 |
| 218 | ! |
| 219 | #ifdef __LITTLE_ENDIAN__ |
| 220 | shll16 r1 |
| 221 | shll8 r1 |
| 222 | 3: mov r1,r3 ! JIHG |
| 223 | shlr16 r3 |
| 224 | shlr8 r3 ! xxxJ |
| 225 | mov.l @r0+,r1 ! NMLK |
| 226 | mov r1,r2 |
| 227 | shll8 r2 ! MLKx |
| 228 | or r2,r3 ! MLKJ |
| 229 | cmp/hs r5,r0 |
| 230 | bf/s 3b |
| 231 | mov.l r3,@(r0,r4) |
| 232 | #else |
| 233 | shlr16 r1 |
| 234 | shlr8 r1 |
| 235 | 3: mov r1,r3 ! GHIJ |
| 236 | shll16 r3 |
| 237 | shll8 r3 ! Jxxx |
| 238 | mov.l @r0+,r1 ! KLMN |
| 239 | mov r1,r2 |
| 240 | shlr8 r2 ! xKLM |
| 241 | or r2,r3 ! JKLM |
| 242 | cmp/hs r5,r0 |
| 243 | bf/s 3b ! while(r0<r5) |
| 244 | mov.l r3,@(r0,r4) |
| 245 | #endif |
| 246 | add #7,r5 |
| 247 | ! |
| 248 | ! Third, copy a byte at once, if necessary |
| 249 | cmp/eq r5,r0 |
| 250 | bt/s 9b |
| 251 | add #5,r4 |
| 252 | add #-3,r0 |
| 253 | bra 8b |
| 254 | add #-1,r4 |