1 /* SPDX-License-Identifier: GPL-2.0-only */ !! 1 /* memcpy.S: Sparc optimized memcpy, bcopy and memmove code
2 /* !! 2 * Hand optimized from GNU libc's memcpy, bcopy and memmove
3 * Copyright (c) 2012-2021, Arm Limited. !! 3 * Copyright (C) 1991,1996 Free Software Foundation
4 * !! 4 * Copyright (C) 1995 Linus Torvalds (Linus.Torvalds@helsinki.fi)
5 * Adapted from the original at: !! 5 * Copyright (C) 1996 David S. Miller (davem@caip.rutgers.edu)
6 * https://github.com/ARM-software/optimized-r !! 6 * Copyright (C) 1996 Eddie C. Dost (ecd@skynet.be)
>> 7 * Copyright (C) 1996 Jakub Jelinek (jj@sunsite.mff.cuni.cz)
7 */ 8 */
8 9
9 #include <linux/linkage.h> !! 10 #ifdef __KERNEL__
10 #include <asm/assembler.h> <<
11 11
12 /* Assumptions: !! 12 #include <asm/cprefix.h>
13 * !! 13
14 * ARMv8-a, AArch64, unaligned accesses. !! 14 #define FUNC(x) \
15 * !! 15 .globl C_LABEL(x); \
>> 16 .type C_LABEL(x),@function; \
>> 17 .align 4; \
>> 18 C_LABEL(x):
>> 19
>> 20 #undef FASTER_REVERSE
>> 21 #undef FASTER_NONALIGNED
>> 22 #define FASTER_ALIGNED
>> 23
>> 24 /* In kernel these functions don't return a value.
>> 25 * One should use macros in asm/string.h for that purpose.
>> 26 * We return 0, so that bugs are more apparent.
16 */ 27 */
>> 28 #define SETUP_RETL
>> 29 #define RETL_INSN clr %o0
>> 30
>> 31 #else
>> 32
>> 33 /* libc */
17 34
18 #define L(label) .L ## label !! 35 #include "DEFS.h"
>> 36
>> 37 #define FASTER_REVERSE
>> 38 #define FASTER_NONALIGNED
>> 39 #define FASTER_ALIGNED
>> 40
>> 41 #define SETUP_RETL mov %o0, %g6
>> 42 #define RETL_INSN mov %g6, %o0
>> 43
>> 44 #endif
>> 45
>> 46 /* Both these macros have to start with exactly the same insn */
>> 47 #define MOVE_BIGCHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, t7) \
>> 48 ldd [%src + offset + 0x00], %t0; \
>> 49 ldd [%src + offset + 0x08], %t2; \
>> 50 ldd [%src + offset + 0x10], %t4; \
>> 51 ldd [%src + offset + 0x18], %t6; \
>> 52 st %t0, [%dst + offset + 0x00]; \
>> 53 st %t1, [%dst + offset + 0x04]; \
>> 54 st %t2, [%dst + offset + 0x08]; \
>> 55 st %t3, [%dst + offset + 0x0c]; \
>> 56 st %t4, [%dst + offset + 0x10]; \
>> 57 st %t5, [%dst + offset + 0x14]; \
>> 58 st %t6, [%dst + offset + 0x18]; \
>> 59 st %t7, [%dst + offset + 0x1c];
>> 60
>> 61 #define MOVE_BIGALIGNCHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, t7) \
>> 62 ldd [%src + offset + 0x00], %t0; \
>> 63 ldd [%src + offset + 0x08], %t2; \
>> 64 ldd [%src + offset + 0x10], %t4; \
>> 65 ldd [%src + offset + 0x18], %t6; \
>> 66 std %t0, [%dst + offset + 0x00]; \
>> 67 std %t2, [%dst + offset + 0x08]; \
>> 68 std %t4, [%dst + offset + 0x10]; \
>> 69 std %t6, [%dst + offset + 0x18];
>> 70
>> 71 #define MOVE_LASTCHUNK(src, dst, offset, t0, t1, t2, t3) \
>> 72 ldd [%src - offset - 0x10], %t0; \
>> 73 ldd [%src - offset - 0x08], %t2; \
>> 74 st %t0, [%dst - offset - 0x10]; \
>> 75 st %t1, [%dst - offset - 0x0c]; \
>> 76 st %t2, [%dst - offset - 0x08]; \
>> 77 st %t3, [%dst - offset - 0x04];
>> 78
>> 79 #define MOVE_LASTALIGNCHUNK(src, dst, offset, t0, t1, t2, t3) \
>> 80 ldd [%src - offset - 0x10], %t0; \
>> 81 ldd [%src - offset - 0x08], %t2; \
>> 82 std %t0, [%dst - offset - 0x10]; \
>> 83 std %t2, [%dst - offset - 0x08];
>> 84
>> 85 #define MOVE_SHORTCHUNK(src, dst, offset, t0, t1) \
>> 86 ldub [%src - offset - 0x02], %t0; \
>> 87 ldub [%src - offset - 0x01], %t1; \
>> 88 stb %t0, [%dst - offset - 0x02]; \
>> 89 stb %t1, [%dst - offset - 0x01];
>> 90
>> 91 /* Both these macros have to start with exactly the same insn */
>> 92 #define RMOVE_BIGCHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, t7) \
>> 93 ldd [%src - offset - 0x20], %t0; \
>> 94 ldd [%src - offset - 0x18], %t2; \
>> 95 ldd [%src - offset - 0x10], %t4; \
>> 96 ldd [%src - offset - 0x08], %t6; \
>> 97 st %t0, [%dst - offset - 0x20]; \
>> 98 st %t1, [%dst - offset - 0x1c]; \
>> 99 st %t2, [%dst - offset - 0x18]; \
>> 100 st %t3, [%dst - offset - 0x14]; \
>> 101 st %t4, [%dst - offset - 0x10]; \
>> 102 st %t5, [%dst - offset - 0x0c]; \
>> 103 st %t6, [%dst - offset - 0x08]; \
>> 104 st %t7, [%dst - offset - 0x04];
>> 105
>> 106 #define RMOVE_BIGALIGNCHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, t7) \
>> 107 ldd [%src - offset - 0x20], %t0; \
>> 108 ldd [%src - offset - 0x18], %t2; \
>> 109 ldd [%src - offset - 0x10], %t4; \
>> 110 ldd [%src - offset - 0x08], %t6; \
>> 111 std %t0, [%dst - offset - 0x20]; \
>> 112 std %t2, [%dst - offset - 0x18]; \
>> 113 std %t4, [%dst - offset - 0x10]; \
>> 114 std %t6, [%dst - offset - 0x08];
>> 115
>> 116 #define RMOVE_LASTCHUNK(src, dst, offset, t0, t1, t2, t3) \
>> 117 ldd [%src + offset + 0x00], %t0; \
>> 118 ldd [%src + offset + 0x08], %t2; \
>> 119 st %t0, [%dst + offset + 0x00]; \
>> 120 st %t1, [%dst + offset + 0x04]; \
>> 121 st %t2, [%dst + offset + 0x08]; \
>> 122 st %t3, [%dst + offset + 0x0c];
>> 123
>> 124 #define RMOVE_SHORTCHUNK(src, dst, offset, t0, t1) \
>> 125 ldub [%src + offset + 0x00], %t0; \
>> 126 ldub [%src + offset + 0x01], %t1; \
>> 127 stb %t0, [%dst + offset + 0x00]; \
>> 128 stb %t1, [%dst + offset + 0x01];
>> 129
>> 130 #define SMOVE_CHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, prev, shil, shir, offset2) \
>> 131 ldd [%src + offset + 0x00], %t0; \
>> 132 ldd [%src + offset + 0x08], %t2; \
>> 133 srl %t0, shir, %t5; \
>> 134 srl %t1, shir, %t6; \
>> 135 sll %t0, shil, %t0; \
>> 136 or %t5, %prev, %t5; \
>> 137 sll %t1, shil, %prev; \
>> 138 or %t6, %t0, %t0; \
>> 139 srl %t2, shir, %t1; \
>> 140 srl %t3, shir, %t6; \
>> 141 sll %t2, shil, %t2; \
>> 142 or %t1, %prev, %t1; \
>> 143 std %t4, [%dst + offset + offset2 - 0x04]; \
>> 144 std %t0, [%dst + offset + offset2 + 0x04]; \
>> 145 sll %t3, shil, %prev; \
>> 146 or %t6, %t2, %t4;
>> 147
>> 148 #define SMOVE_ALIGNCHUNK(src, dst, offset, t0, t1, t2, t3, t4, t5, t6, prev, shil, shir, offset2) \
>> 149 ldd [%src + offset + 0x00], %t0; \
>> 150 ldd [%src + offset + 0x08], %t2; \
>> 151 srl %t0, shir, %t4; \
>> 152 srl %t1, shir, %t5; \
>> 153 sll %t0, shil, %t6; \
>> 154 or %t4, %prev, %t0; \
>> 155 sll %t1, shil, %prev; \
>> 156 or %t5, %t6, %t1; \
>> 157 srl %t2, shir, %t4; \
>> 158 srl %t3, shir, %t5; \
>> 159 sll %t2, shil, %t6; \
>> 160 or %t4, %prev, %t2; \
>> 161 sll %t3, shil, %prev; \
>> 162 or %t5, %t6, %t3; \
>> 163 std %t0, [%dst + offset + offset2 + 0x00]; \
>> 164 std %t2, [%dst + offset + offset2 + 0x08];
>> 165
>> 166 .text
>> 167 .align 4
>> 168
>> 169 #ifdef FASTER_REVERSE
>> 170
>> 171 70: /* rdword_align */
>> 172
>> 173 andcc %o1, 1, %g0
>> 174 be 4f
>> 175 andcc %o1, 2, %g0
>> 176
>> 177 ldub [%o1 - 1], %g2
>> 178 sub %o1, 1, %o1
>> 179 stb %g2, [%o0 - 1]
>> 180 sub %o2, 1, %o2
>> 181 be 3f
>> 182 sub %o0, 1, %o0
>> 183 4:
>> 184 lduh [%o1 - 2], %g2
>> 185 sub %o1, 2, %o1
>> 186 sth %g2, [%o0 - 2]
>> 187 sub %o2, 2, %o2
>> 188 b 3f
>> 189 sub %o0, 2, %o0
>> 190
>> 191 #endif /* FASTER_REVERSE */
>> 192
>> 193 0:
>> 194 retl
>> 195 nop ! Only bcopy returns here and it retuns void...
>> 196
>> 197 FUNC(bcopy)
>> 198 mov %o0, %o3
>> 199 mov %o1, %o0
>> 200 mov %o3, %o1
>> 201 tst %o2
>> 202 bcs 0b
>> 203 /* Do the cmp in the delay slot */
>> 204 #ifdef __KERNEL__
>> 205 FUNC(amemmove)
>> 206 FUNC(__memmove)
>> 207 #endif
>> 208 FUNC(memmove)
>> 209 cmp %o0, %o1
>> 210 SETUP_RETL
>> 211 bleu 9f
>> 212 sub %o0, %o1, %o4
>> 213
>> 214 add %o1, %o2, %o3
>> 215 cmp %o3, %o0
>> 216 bleu 0f
>> 217 andcc %o4, 3, %o5
>> 218
>> 219 #ifndef FASTER_REVERSE
>> 220
>> 221 add %o1, %o2, %o1
>> 222 add %o0, %o2, %o0
>> 223 sub %o1, 1, %o1
>> 224 sub %o0, 1, %o0
>> 225
>> 226 1: /* reverse_bytes */
>> 227
>> 228 ldub [%o1], %o4
>> 229 subcc %o2, 1, %o2
>> 230 stb %o4, [%o0]
>> 231 sub %o1, 1, %o1
>> 232 bne 1b
>> 233 sub %o0, 1, %o0
>> 234
>> 235 retl
>> 236 RETL_INSN
>> 237
>> 238 #else /* FASTER_REVERSE */
>> 239
>> 240 add %o1, %o2, %o1
>> 241 add %o0, %o2, %o0
>> 242 bne 77f
>> 243 cmp %o2, 15
>> 244 bleu 91f
>> 245 andcc %o1, 3, %g0
>> 246 bne 70b
>> 247 3:
>> 248 andcc %o1, 4, %g0
>> 249
>> 250 be 2f
>> 251 mov %o2, %g1
>> 252
>> 253 ld [%o1 - 4], %o4
>> 254 sub %g1, 4, %g1
>> 255 st %o4, [%o0 - 4]
>> 256 sub %o1, 4, %o1
>> 257 sub %o0, 4, %o0
>> 258 2:
>> 259 andcc %g1, 0xffffff80, %g7
>> 260 be 3f
>> 261 andcc %o0, 4, %g0
>> 262
>> 263 be 74f + 4
>> 264 5:
>> 265 RMOVE_BIGCHUNK(o1, o0, 0x00, o2, o3, o4, o5, g2, g3, g4, g5)
>> 266 RMOVE_BIGCHUNK(o1, o0, 0x20, o2, o3, o4, o5, g2, g3, g4, g5)
>> 267 RMOVE_BIGCHUNK(o1, o0, 0x40, o2, o3, o4, o5, g2, g3, g4, g5)
>> 268 RMOVE_BIGCHUNK(o1, o0, 0x60, o2, o3, o4, o5, g2, g3, g4, g5)
>> 269 subcc %g7, 128, %g7
>> 270 sub %o1, 128, %o1
>> 271 bne 5b
>> 272 sub %o0, 128, %o0
>> 273 3:
>> 274 andcc %g1, 0x70, %g7
>> 275 be 72f
>> 276 andcc %g1, 8, %g0
>> 277
>> 278 sethi %hi(72f), %o5
>> 279 srl %g7, 1, %o4
>> 280 add %g7, %o4, %o4
>> 281 sub %o1, %g7, %o1
>> 282 sub %o5, %o4, %o5
>> 283 jmpl %o5 + %lo(72f), %g0
>> 284 sub %o0, %g7, %o0
>> 285
>> 286 71: /* rmemcpy_table */
>> 287 RMOVE_LASTCHUNK(o1, o0, 0x60, g2, g3, g4, g5)
>> 288 RMOVE_LASTCHUNK(o1, o0, 0x50, g2, g3, g4, g5)
>> 289 RMOVE_LASTCHUNK(o1, o0, 0x40, g2, g3, g4, g5)
>> 290 RMOVE_LASTCHUNK(o1, o0, 0x30, g2, g3, g4, g5)
>> 291 RMOVE_LASTCHUNK(o1, o0, 0x20, g2, g3, g4, g5)
>> 292 RMOVE_LASTCHUNK(o1, o0, 0x10, g2, g3, g4, g5)
>> 293 RMOVE_LASTCHUNK(o1, o0, 0x00, g2, g3, g4, g5)
>> 294
>> 295 72: /* rmemcpy_table_end */
>> 296
>> 297 be 73f
>> 298 andcc %g1, 4, %g0
>> 299
>> 300 ldd [%o1 - 0x08], %g2
>> 301 sub %o0, 8, %o0
>> 302 sub %o1, 8, %o1
>> 303 st %g2, [%o0]
>> 304 st %g3, [%o0 + 0x04]
>> 305
>> 306 73: /* rmemcpy_last7 */
>> 307
>> 308 be 1f
>> 309 andcc %g1, 2, %g0
>> 310
>> 311 ld [%o1 - 4], %g2
>> 312 sub %o1, 4, %o1
>> 313 st %g2, [%o0 - 4]
>> 314 sub %o0, 4, %o0
>> 315 1:
>> 316 be 1f
>> 317 andcc %g1, 1, %g0
>> 318
>> 319 lduh [%o1 - 2], %g2
>> 320 sub %o1, 2, %o1
>> 321 sth %g2, [%o0 - 2]
>> 322 sub %o0, 2, %o0
>> 323 1:
>> 324 be 1f
>> 325 nop
>> 326
>> 327 ldub [%o1 - 1], %g2
>> 328 stb %g2, [%o0 - 1]
>> 329 1:
>> 330 retl
>> 331 RETL_INSN
>> 332
>> 333 74: /* rldd_std */
>> 334 RMOVE_BIGALIGNCHUNK(o1, o0, 0x00, o2, o3, o4, o5, g2, g3, g4, g5)
>> 335 RMOVE_BIGALIGNCHUNK(o1, o0, 0x20, o2, o3, o4, o5, g2, g3, g4, g5)
>> 336 RMOVE_BIGALIGNCHUNK(o1, o0, 0x40, o2, o3, o4, o5, g2, g3, g4, g5)
>> 337 RMOVE_BIGALIGNCHUNK(o1, o0, 0x60, o2, o3, o4, o5, g2, g3, g4, g5)
>> 338 subcc %g7, 128, %g7
>> 339 sub %o1, 128, %o1
>> 340 bne 74b
>> 341 sub %o0, 128, %o0
>> 342
>> 343 andcc %g1, 0x70, %g7
>> 344 be 72b
>> 345 andcc %g1, 8, %g0
>> 346
>> 347 sethi %hi(72b), %o5
>> 348 srl %g7, 1, %o4
>> 349 add %g7, %o4, %o4
>> 350 sub %o1, %g7, %o1
>> 351 sub %o5, %o4, %o5
>> 352 jmpl %o5 + %lo(72b), %g0
>> 353 sub %o0, %g7, %o0
>> 354
>> 355 75: /* rshort_end */
>> 356
>> 357 and %o2, 0xe, %o3
>> 358 2:
>> 359 sethi %hi(76f), %o5
>> 360 sll %o3, 3, %o4
>> 361 sub %o0, %o3, %o0
>> 362 sub %o5, %o4, %o5
>> 363 sub %o1, %o3, %o1
>> 364 jmpl %o5 + %lo(76f), %g0
>> 365 andcc %o2, 1, %g0
>> 366
>> 367 RMOVE_SHORTCHUNK(o1, o0, 0x0c, g2, g3)
>> 368 RMOVE_SHORTCHUNK(o1, o0, 0x0a, g2, g3)
>> 369 RMOVE_SHORTCHUNK(o1, o0, 0x08, g2, g3)
>> 370 RMOVE_SHORTCHUNK(o1, o0, 0x06, g2, g3)
>> 371 RMOVE_SHORTCHUNK(o1, o0, 0x04, g2, g3)
>> 372 RMOVE_SHORTCHUNK(o1, o0, 0x02, g2, g3)
>> 373 RMOVE_SHORTCHUNK(o1, o0, 0x00, g2, g3)
>> 374
>> 375 76: /* rshort_table_end */
>> 376
>> 377 be 1f
>> 378 nop
>> 379 ldub [%o1 - 1], %g2
>> 380 stb %g2, [%o0 - 1]
>> 381 1:
>> 382 retl
>> 383 RETL_INSN
>> 384
>> 385 91: /* rshort_aligned_end */
>> 386
>> 387 bne 75b
>> 388 andcc %o2, 8, %g0
>> 389
>> 390 be 1f
>> 391 andcc %o2, 4, %g0
>> 392
>> 393 ld [%o1 - 0x08], %g2
>> 394 ld [%o1 - 0x04], %g3
>> 395 sub %o1, 8, %o1
>> 396 st %g2, [%o0 - 0x08]
>> 397 st %g3, [%o0 - 0x04]
>> 398 sub %o0, 8, %o0
>> 399 1:
>> 400 b 73b
>> 401 mov %o2, %g1
>> 402
>> 403 77: /* rnon_aligned */
>> 404 cmp %o2, 15
>> 405 bleu 75b
>> 406 andcc %o0, 3, %g0
>> 407 be 64f
>> 408 andcc %o0, 1, %g0
>> 409 be 63f
>> 410 andcc %o0, 2, %g0
>> 411 ldub [%o1 - 1], %g5
>> 412 sub %o1, 1, %o1
>> 413 stb %g5, [%o0 - 1]
>> 414 sub %o0, 1, %o0
>> 415 be 64f
>> 416 sub %o2, 1, %o2
>> 417 63:
>> 418 ldub [%o1 - 1], %g5
>> 419 sub %o1, 2, %o1
>> 420 stb %g5, [%o0 - 1]
>> 421 sub %o0, 2, %o0
>> 422 ldub [%o1], %g5
>> 423 sub %o2, 2, %o2
>> 424 stb %g5, [%o0]
>> 425 64:
>> 426 and %o1, 3, %g2
>> 427 and %o1, -4, %o1
>> 428 and %o2, 0xc, %g3
>> 429 add %o1, 4, %o1
>> 430 cmp %g3, 4
>> 431 sll %g2, 3, %g4
>> 432 mov 32, %g2
>> 433 be 4f
>> 434 sub %g2, %g4, %g7
>> 435
>> 436 blu 3f
>> 437 cmp %g3, 8
>> 438
>> 439 be 2f
>> 440 srl %o2, 2, %g3
>> 441
>> 442 ld [%o1 - 4], %o3
>> 443 add %o0, -8, %o0
>> 444 ld [%o1 - 8], %o4
>> 445 add %o1, -16, %o1
>> 446 b 7f
>> 447 add %g3, 1, %g3
>> 448 2:
>> 449 ld [%o1 - 4], %o4
>> 450 add %o0, -4, %o0
>> 451 ld [%o1 - 8], %g1
>> 452 add %o1, -12, %o1
>> 453 b 8f
>> 454 add %g3, 2, %g3
>> 455 3:
>> 456 ld [%o1 - 4], %o5
>> 457 add %o0, -12, %o0
>> 458 ld [%o1 - 8], %o3
>> 459 add %o1, -20, %o1
>> 460 b 6f
>> 461 srl %o2, 2, %g3
>> 462 4:
>> 463 ld [%o1 - 4], %g1
>> 464 srl %o2, 2, %g3
>> 465 ld [%o1 - 8], %o5
>> 466 add %o1, -24, %o1
>> 467 add %o0, -16, %o0
>> 468 add %g3, -1, %g3
>> 469
>> 470 ld [%o1 + 12], %o3
>> 471 5:
>> 472 sll %o5, %g4, %g2
>> 473 srl %g1, %g7, %g5
>> 474 or %g2, %g5, %g2
>> 475 st %g2, [%o0 + 12]
>> 476 6:
>> 477 ld [%o1 + 8], %o4
>> 478 sll %o3, %g4, %g2
>> 479 srl %o5, %g7, %g5
>> 480 or %g2, %g5, %g2
>> 481 st %g2, [%o0 + 8]
>> 482 7:
>> 483 ld [%o1 + 4], %g1
>> 484 sll %o4, %g4, %g2
>> 485 srl %o3, %g7, %g5
>> 486 or %g2, %g5, %g2
>> 487 st %g2, [%o0 + 4]
>> 488 8:
>> 489 ld [%o1], %o5
>> 490 sll %g1, %g4, %g2
>> 491 srl %o4, %g7, %g5
>> 492 addcc %g3, -4, %g3
>> 493 or %g2, %g5, %g2
>> 494 add %o1, -16, %o1
>> 495 st %g2, [%o0]
>> 496 add %o0, -16, %o0
>> 497 bne,a 5b
>> 498 ld [%o1 + 12], %o3
>> 499 sll %o5, %g4, %g2
>> 500 srl %g1, %g7, %g5
>> 501 srl %g4, 3, %g3
>> 502 or %g2, %g5, %g2
>> 503 add %o1, %g3, %o1
>> 504 andcc %o2, 2, %g0
>> 505 st %g2, [%o0 + 12]
>> 506 be 1f
>> 507 andcc %o2, 1, %g0
>> 508
>> 509 ldub [%o1 + 15], %g5
>> 510 add %o1, -2, %o1
>> 511 stb %g5, [%o0 + 11]
>> 512 add %o0, -2, %o0
>> 513 ldub [%o1 + 16], %g5
>> 514 stb %g5, [%o0 + 12]
>> 515 1:
>> 516 be 1f
>> 517 nop
>> 518 ldub [%o1 + 15], %g5
>> 519 stb %g5, [%o0 + 11]
>> 520 1:
>> 521 retl
>> 522 RETL_INSN
>> 523
>> 524 #endif /* FASTER_REVERSE */
>> 525
>> 526 /* NOTE: This code is executed just for the cases,
>> 527 where %src (=%o1) & 3 is != 0.
>> 528 We need to align it to 4. So, for (%src & 3)
>> 529 1 we need to do ldub,lduh
>> 530 2 lduh
>> 531 3 just ldub
>> 532 so even if it looks weird, the branches
>> 533 are correct here. -jj
>> 534 */
>> 535 78: /* dword_align */
19 536
20 #define dstin x0 !! 537 andcc %o1, 1, %g0
21 #define src x1 !! 538 be 4f
22 #define count x2 !! 539 andcc %o1, 2, %g0
23 #define dst x3 !! 540
24 #define srcend x4 !! 541 ldub [%o1], %g2
25 #define dstend x5 !! 542 add %o1, 1, %o1
26 #define A_l x6 !! 543 stb %g2, [%o0]
27 #define A_lw w6 !! 544 sub %o2, 1, %o2
28 #define A_h x7 !! 545 bne 3f
29 #define B_l x8 !! 546 add %o0, 1, %o0
30 #define B_lw w8 !! 547 4:
31 #define B_h x9 !! 548 lduh [%o1], %g2
32 #define C_l x10 !! 549 add %o1, 2, %o1
33 #define C_lw w10 !! 550 sth %g2, [%o0]
34 #define C_h x11 !! 551 sub %o2, 2, %o2
35 #define D_l x12 !! 552 b 3f
36 #define D_h x13 !! 553 add %o0, 2, %o0
37 #define E_l x14 !! 554
38 #define E_h x15 !! 555 #ifdef __KERNEL__
39 #define F_l x16 !! 556 FUNC(__memcpy)
40 #define F_h x17 !! 557 #endif
41 #define G_l count !! 558 FUNC(memcpy) /* %o0=dst %o1=src %o2=len */
42 #define G_h dst !! 559
43 #define H_l src !! 560 sub %o0, %o1, %o4
44 #define H_h srcend !! 561 SETUP_RETL
45 #define tmp1 x14 !! 562 9:
46 !! 563 andcc %o4, 3, %o5
47 /* This implementation handles overlaps and su !! 564 0:
48 from a single entry point. It uses unalign !! 565 bne 86f
49 sequences to keep the code small, simple an !! 566 cmp %o2, 15
50 !! 567
51 Copies are split into 3 main cases: small c !! 568 bleu 90f
52 copies of up to 128 bytes, and large copies !! 569 andcc %o1, 3, %g0
53 check is negligible since it is only requir !! 570
54 !! 571 bne 78b
55 Large copies use a software pipelined loop !! 572 3:
56 The destination pointer is 16-byte aligned !! 573 andcc %o1, 4, %g0
57 The loop tail is handled by always copying !! 574
58 */ !! 575 be 2f
59 !! 576 mov %o2, %g1
60 SYM_FUNC_START(__pi_memcpy) !! 577
61 add srcend, src, count !! 578 ld [%o1], %o4
62 add dstend, dstin, count !! 579 sub %g1, 4, %g1
63 cmp count, 128 !! 580 st %o4, [%o0]
64 b.hi L(copy_long) !! 581 add %o1, 4, %o1
65 cmp count, 32 !! 582 add %o0, 4, %o0
66 b.hi L(copy32_128) !! 583 2:
67 !! 584 andcc %g1, 0xffffff80, %g7
68 /* Small copies: 0..32 bytes. */ !! 585 be 3f
69 cmp count, 16 !! 586 andcc %o0, 4, %g0
70 b.lo L(copy16) !! 587
71 ldp A_l, A_h, [src] !! 588 be 82f + 4
72 ldp D_l, D_h, [srcend, -16] !! 589 5:
73 stp A_l, A_h, [dstin] !! 590 MOVE_BIGCHUNK(o1, o0, 0x00, o2, o3, o4, o5, g2, g3, g4, g5)
74 stp D_l, D_h, [dstend, -16] !! 591 MOVE_BIGCHUNK(o1, o0, 0x20, o2, o3, o4, o5, g2, g3, g4, g5)
75 ret !! 592 MOVE_BIGCHUNK(o1, o0, 0x40, o2, o3, o4, o5, g2, g3, g4, g5)
76 !! 593 MOVE_BIGCHUNK(o1, o0, 0x60, o2, o3, o4, o5, g2, g3, g4, g5)
77 /* Copy 8-15 bytes. */ !! 594 subcc %g7, 128, %g7
78 L(copy16): !! 595 add %o1, 128, %o1
79 tbz count, 3, L(copy8) !! 596 bne 5b
80 ldr A_l, [src] !! 597 add %o0, 128, %o0
81 ldr A_h, [srcend, -8] !! 598 3:
82 str A_l, [dstin] !! 599 andcc %g1, 0x70, %g7
83 str A_h, [dstend, -8] !! 600 be 80f
84 ret !! 601 andcc %g1, 8, %g0
85 !! 602
86 .p2align 3 !! 603 sethi %hi(80f), %o5
87 /* Copy 4-7 bytes. */ !! 604 srl %g7, 1, %o4
88 L(copy8): !! 605 add %g7, %o4, %o4
89 tbz count, 2, L(copy4) !! 606 add %o1, %g7, %o1
90 ldr A_lw, [src] !! 607 sub %o5, %o4, %o5
91 ldr B_lw, [srcend, -4] !! 608 jmpl %o5 + %lo(80f), %g0
92 str A_lw, [dstin] !! 609 add %o0, %g7, %o0
93 str B_lw, [dstend, -4] !! 610
94 ret !! 611 79: /* memcpy_table */
95 !! 612
96 /* Copy 0..3 bytes using a branchless !! 613 MOVE_LASTCHUNK(o1, o0, 0x60, g2, g3, g4, g5)
97 L(copy4): !! 614 MOVE_LASTCHUNK(o1, o0, 0x50, g2, g3, g4, g5)
98 cbz count, L(copy0) !! 615 MOVE_LASTCHUNK(o1, o0, 0x40, g2, g3, g4, g5)
99 lsr tmp1, count, 1 !! 616 MOVE_LASTCHUNK(o1, o0, 0x30, g2, g3, g4, g5)
100 ldrb A_lw, [src] !! 617 MOVE_LASTCHUNK(o1, o0, 0x20, g2, g3, g4, g5)
101 ldrb C_lw, [srcend, -1] !! 618 MOVE_LASTCHUNK(o1, o0, 0x10, g2, g3, g4, g5)
102 ldrb B_lw, [src, tmp1] !! 619 MOVE_LASTCHUNK(o1, o0, 0x00, g2, g3, g4, g5)
103 strb A_lw, [dstin] !! 620
104 strb B_lw, [dstin, tmp1] !! 621 80: /* memcpy_table_end */
105 strb C_lw, [dstend, -1] !! 622 be 81f
106 L(copy0): !! 623 andcc %g1, 4, %g0
107 ret !! 624
108 !! 625 ldd [%o1], %g2
109 .p2align 4 !! 626 add %o0, 8, %o0
110 /* Medium copies: 33..128 bytes. */ !! 627 st %g2, [%o0 - 0x08]
111 L(copy32_128): !! 628 add %o1, 8, %o1
112 ldp A_l, A_h, [src] !! 629 st %g3, [%o0 - 0x04]
113 ldp B_l, B_h, [src, 16] !! 630
114 ldp C_l, C_h, [srcend, -32] !! 631 81: /* memcpy_last7 */
115 ldp D_l, D_h, [srcend, -16] !! 632
116 cmp count, 64 !! 633 be 1f
117 b.hi L(copy128) !! 634 andcc %g1, 2, %g0
118 stp A_l, A_h, [dstin] !! 635
119 stp B_l, B_h, [dstin, 16] !! 636 ld [%o1], %g2
120 stp C_l, C_h, [dstend, -32] !! 637 add %o1, 4, %o1
121 stp D_l, D_h, [dstend, -16] !! 638 st %g2, [%o0]
122 ret !! 639 add %o0, 4, %o0
123 !! 640 1:
124 .p2align 4 !! 641 be 1f
125 /* Copy 65..128 bytes. */ !! 642 andcc %g1, 1, %g0
126 L(copy128): !! 643
127 ldp E_l, E_h, [src, 32] !! 644 lduh [%o1], %g2
128 ldp F_l, F_h, [src, 48] !! 645 add %o1, 2, %o1
129 cmp count, 96 !! 646 sth %g2, [%o0]
130 b.ls L(copy96) !! 647 add %o0, 2, %o0
131 ldp G_l, G_h, [srcend, -64] !! 648 1:
132 ldp H_l, H_h, [srcend, -48] !! 649 be 1f
133 stp G_l, G_h, [dstend, -64] !! 650 nop
134 stp H_l, H_h, [dstend, -48] !! 651
135 L(copy96): !! 652 ldub [%o1], %g2
136 stp A_l, A_h, [dstin] !! 653 stb %g2, [%o0]
137 stp B_l, B_h, [dstin, 16] !! 654 1:
138 stp E_l, E_h, [dstin, 32] !! 655 retl
139 stp F_l, F_h, [dstin, 48] !! 656 RETL_INSN
140 stp C_l, C_h, [dstend, -32] !! 657
141 stp D_l, D_h, [dstend, -16] !! 658 82: /* ldd_std */
142 ret !! 659 MOVE_BIGALIGNCHUNK(o1, o0, 0x00, o2, o3, o4, o5, g2, g3, g4, g5)
143 !! 660 MOVE_BIGALIGNCHUNK(o1, o0, 0x20, o2, o3, o4, o5, g2, g3, g4, g5)
144 .p2align 4 !! 661 MOVE_BIGALIGNCHUNK(o1, o0, 0x40, o2, o3, o4, o5, g2, g3, g4, g5)
145 /* Copy more than 128 bytes. */ !! 662 MOVE_BIGALIGNCHUNK(o1, o0, 0x60, o2, o3, o4, o5, g2, g3, g4, g5)
146 L(copy_long): !! 663 subcc %g7, 128, %g7
147 /* Use backwards copy if there is an o !! 664 add %o1, 128, %o1
148 sub tmp1, dstin, src !! 665 bne 82b
149 cbz tmp1, L(copy0) !! 666 add %o0, 128, %o0
150 cmp tmp1, count !! 667
151 b.lo L(copy_long_backwards) !! 668 #ifndef FASTER_ALIGNED
152 !! 669
153 /* Copy 16 bytes and then align dst to !! 670 andcc %g1, 0x70, %g7
154 !! 671 be 80b
155 ldp D_l, D_h, [src] !! 672 andcc %g1, 8, %g0
156 and tmp1, dstin, 15 !! 673
157 bic dst, dstin, 15 !! 674 sethi %hi(80b), %o5
158 sub src, src, tmp1 !! 675 srl %g7, 1, %o4
159 add count, count, tmp1 /* Cou !! 676 add %g7, %o4, %o4
160 ldp A_l, A_h, [src, 16] !! 677 add %o1, %g7, %o1
161 stp D_l, D_h, [dstin] !! 678 sub %o5, %o4, %o5
162 ldp B_l, B_h, [src, 32] !! 679 jmpl %o5 + %lo(80b), %g0
163 ldp C_l, C_h, [src, 48] !! 680 add %o0, %g7, %o0
164 ldp D_l, D_h, [src, 64]! !! 681
165 subs count, count, 128 + 16 /* Tes !! 682 #else /* FASTER_ALIGNED */
166 b.ls L(copy64_from_end) !! 683
167 !! 684 andcc %g1, 0x70, %g7
168 L(loop64): !! 685 be 84f
169 stp A_l, A_h, [dst, 16] !! 686 andcc %g1, 8, %g0
170 ldp A_l, A_h, [src, 16] !! 687
171 stp B_l, B_h, [dst, 32] !! 688 sethi %hi(84f), %o5
172 ldp B_l, B_h, [src, 32] !! 689 add %o1, %g7, %o1
173 stp C_l, C_h, [dst, 48] !! 690 sub %o5, %g7, %o5
174 ldp C_l, C_h, [src, 48] !! 691 jmpl %o5 + %lo(84f), %g0
175 stp D_l, D_h, [dst, 64]! !! 692 add %o0, %g7, %o0
176 ldp D_l, D_h, [src, 64]! !! 693
177 subs count, count, 64 !! 694 83: /* amemcpy_table */
178 b.hi L(loop64) !! 695
179 !! 696 MOVE_LASTALIGNCHUNK(o1, o0, 0x60, g2, g3, g4, g5)
180 /* Write the last iteration and copy 6 !! 697 MOVE_LASTALIGNCHUNK(o1, o0, 0x50, g2, g3, g4, g5)
181 L(copy64_from_end): !! 698 MOVE_LASTALIGNCHUNK(o1, o0, 0x40, g2, g3, g4, g5)
182 ldp E_l, E_h, [srcend, -64] !! 699 MOVE_LASTALIGNCHUNK(o1, o0, 0x30, g2, g3, g4, g5)
183 stp A_l, A_h, [dst, 16] !! 700 MOVE_LASTALIGNCHUNK(o1, o0, 0x20, g2, g3, g4, g5)
184 ldp A_l, A_h, [srcend, -48] !! 701 MOVE_LASTALIGNCHUNK(o1, o0, 0x10, g2, g3, g4, g5)
185 stp B_l, B_h, [dst, 32] !! 702 MOVE_LASTALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5)
186 ldp B_l, B_h, [srcend, -32] !! 703
187 stp C_l, C_h, [dst, 48] !! 704 84: /* amemcpy_table_end */
188 ldp C_l, C_h, [srcend, -16] !! 705 be 85f
189 stp D_l, D_h, [dst, 64] !! 706 andcc %g1, 4, %g0
190 stp E_l, E_h, [dstend, -64] !! 707
191 stp A_l, A_h, [dstend, -48] !! 708 ldd [%o1], %g2
192 stp B_l, B_h, [dstend, -32] !! 709 add %o0, 8, %o0
193 stp C_l, C_h, [dstend, -16] !! 710 std %g2, [%o0 - 0x08]
194 ret !! 711 add %o1, 8, %o1
195 !! 712 85: /* amemcpy_last7 */
196 .p2align 4 !! 713 be 1f
197 !! 714 andcc %g1, 2, %g0
198 /* Large backwards copy for overlappin !! 715
199 Copy 16 bytes and then align dst to !! 716 ld [%o1], %g2
200 L(copy_long_backwards): !! 717 add %o1, 4, %o1
201 ldp D_l, D_h, [srcend, -16] !! 718 st %g2, [%o0]
202 and tmp1, dstend, 15 !! 719 add %o0, 4, %o0
203 sub srcend, srcend, tmp1 !! 720 1:
204 sub count, count, tmp1 !! 721 be 1f
205 ldp A_l, A_h, [srcend, -16] !! 722 andcc %g1, 1, %g0
206 stp D_l, D_h, [dstend, -16] !! 723
207 ldp B_l, B_h, [srcend, -32] !! 724 lduh [%o1], %g2
208 ldp C_l, C_h, [srcend, -48] !! 725 add %o1, 2, %o1
209 ldp D_l, D_h, [srcend, -64]! !! 726 sth %g2, [%o0]
210 sub dstend, dstend, tmp1 !! 727 add %o0, 2, %o0
211 subs count, count, 128 !! 728 1:
212 b.ls L(copy64_from_start) !! 729 be 1f
213 !! 730 nop
214 L(loop64_backwards): !! 731
215 stp A_l, A_h, [dstend, -16] !! 732 ldub [%o1], %g2
216 ldp A_l, A_h, [srcend, -16] !! 733 stb %g2, [%o0]
217 stp B_l, B_h, [dstend, -32] !! 734 1:
218 ldp B_l, B_h, [srcend, -32] !! 735 retl
219 stp C_l, C_h, [dstend, -48] !! 736 RETL_INSN
220 ldp C_l, C_h, [srcend, -48] !! 737
221 stp D_l, D_h, [dstend, -64]! !! 738 #endif /* FASTER_ALIGNED */
222 ldp D_l, D_h, [srcend, -64]! !! 739
223 subs count, count, 64 !! 740 86: /* non_aligned */
224 b.hi L(loop64_backwards) !! 741 cmp %o2, 6
225 !! 742 bleu 88f
226 /* Write the last iteration and copy 6 !! 743
227 L(copy64_from_start): !! 744 #ifdef FASTER_NONALIGNED
228 ldp G_l, G_h, [src, 48] !! 745
229 stp A_l, A_h, [dstend, -16] !! 746 cmp %o2, 256
230 ldp A_l, A_h, [src, 32] !! 747 bcc 87f
231 stp B_l, B_h, [dstend, -32] !! 748
232 ldp B_l, B_h, [src, 16] !! 749 #endif /* FASTER_NONALIGNED */
233 stp C_l, C_h, [dstend, -48] !! 750
234 ldp C_l, C_h, [src] !! 751 andcc %o0, 3, %g0
235 stp D_l, D_h, [dstend, -64] !! 752 be 61f
236 stp G_l, G_h, [dstin, 48] !! 753 andcc %o0, 1, %g0
237 stp A_l, A_h, [dstin, 32] !! 754 be 60f
238 stp B_l, B_h, [dstin, 16] !! 755 andcc %o0, 2, %g0
239 stp C_l, C_h, [dstin] !! 756
240 ret !! 757 ldub [%o1], %g5
241 SYM_FUNC_END(__pi_memcpy) !! 758 add %o1, 1, %o1
242 !! 759 stb %g5, [%o0]
243 SYM_FUNC_ALIAS(__memcpy, __pi_memcpy) !! 760 sub %o2, 1, %o2
244 EXPORT_SYMBOL(__memcpy) !! 761 bne 61f
245 SYM_FUNC_ALIAS_WEAK(memcpy, __memcpy) !! 762 add %o0, 1, %o0
246 EXPORT_SYMBOL(memcpy) !! 763 60:
247 !! 764 ldub [%o1], %g3
248 SYM_FUNC_ALIAS(__pi_memmove, __pi_memcpy) !! 765 add %o1, 2, %o1
249 !! 766 stb %g3, [%o0]
250 SYM_FUNC_ALIAS(__memmove, __pi_memmove) !! 767 sub %o2, 2, %o2
251 EXPORT_SYMBOL(__memmove) !! 768 ldub [%o1 - 1], %g3
252 SYM_FUNC_ALIAS_WEAK(memmove, __memmove) !! 769 add %o0, 2, %o0
253 EXPORT_SYMBOL(memmove) !! 770 stb %g3, [%o0 - 1]
>> 771 61:
>> 772 and %o1, 3, %g2
>> 773 and %o2, 0xc, %g3
>> 774 and %o1, -4, %o1
>> 775 cmp %g3, 4
>> 776 sll %g2, 3, %g4
>> 777 mov 32, %g2
>> 778 be 4f
>> 779 sub %g2, %g4, %g7
>> 780
>> 781 blu 3f
>> 782 cmp %g3, 0x8
>> 783
>> 784 be 2f
>> 785 srl %o2, 2, %g3
>> 786
>> 787 ld [%o1], %o3
>> 788 add %o0, -8, %o0
>> 789 ld [%o1 + 4], %o4
>> 790 b 8f
>> 791 add %g3, 1, %g3
>> 792 2:
>> 793 ld [%o1], %o4
>> 794 add %o0, -12, %o0
>> 795 ld [%o1 + 4], %o5
>> 796 add %g3, 2, %g3
>> 797 b 9f
>> 798 add %o1, -4, %o1
>> 799 3:
>> 800 ld [%o1], %g1
>> 801 add %o0, -4, %o0
>> 802 ld [%o1 + 4], %o3
>> 803 srl %o2, 2, %g3
>> 804 b 7f
>> 805 add %o1, 4, %o1
>> 806 4:
>> 807 ld [%o1], %o5
>> 808 cmp %o2, 7
>> 809 ld [%o1 + 4], %g1
>> 810 srl %o2, 2, %g3
>> 811 bleu 10f
>> 812 add %o1, 8, %o1
>> 813
>> 814 ld [%o1], %o3
>> 815 add %g3, -1, %g3
>> 816 5:
>> 817 sll %o5, %g4, %g2
>> 818 srl %g1, %g7, %g5
>> 819 or %g2, %g5, %g2
>> 820 st %g2, [%o0]
>> 821 7:
>> 822 ld [%o1 + 4], %o4
>> 823 sll %g1, %g4, %g2
>> 824 srl %o3, %g7, %g5
>> 825 or %g2, %g5, %g2
>> 826 st %g2, [%o0 + 4]
>> 827 8:
>> 828 ld [%o1 + 8], %o5
>> 829 sll %o3, %g4, %g2
>> 830 srl %o4, %g7, %g5
>> 831 or %g2, %g5, %g2
>> 832 st %g2, [%o0 + 8]
>> 833 9:
>> 834 ld [%o1 + 12], %g1
>> 835 sll %o4, %g4, %g2
>> 836 srl %o5, %g7, %g5
>> 837 addcc %g3, -4, %g3
>> 838 or %g2, %g5, %g2
>> 839 add %o1, 16, %o1
>> 840 st %g2, [%o0 + 12]
>> 841 add %o0, 16, %o0
>> 842 bne,a 5b
>> 843 ld [%o1], %o3
>> 844 10:
>> 845 sll %o5, %g4, %g2
>> 846 srl %g1, %g7, %g5
>> 847 srl %g7, 3, %g3
>> 848 or %g2, %g5, %g2
>> 849 sub %o1, %g3, %o1
>> 850 andcc %o2, 2, %g0
>> 851 st %g2, [%o0]
>> 852 be 1f
>> 853 andcc %o2, 1, %g0
>> 854
>> 855 ldub [%o1], %g2
>> 856 add %o1, 2, %o1
>> 857 stb %g2, [%o0 + 4]
>> 858 add %o0, 2, %o0
>> 859 ldub [%o1 - 1], %g2
>> 860 stb %g2, [%o0 + 3]
>> 861 1:
>> 862 be 1f
>> 863 nop
>> 864 ldub [%o1], %g2
>> 865 stb %g2, [%o0 + 4]
>> 866 1:
>> 867 retl
>> 868 RETL_INSN
>> 869
>> 870 #ifdef FASTER_NONALIGNED
>> 871
>> 872 87: /* faster_nonaligned */
>> 873
>> 874 andcc %o1, 3, %g0
>> 875 be 3f
>> 876 andcc %o1, 1, %g0
>> 877
>> 878 be 4f
>> 879 andcc %o1, 2, %g0
>> 880
>> 881 ldub [%o1], %g2
>> 882 add %o1, 1, %o1
>> 883 stb %g2, [%o0]
>> 884 sub %o2, 1, %o2
>> 885 bne 3f
>> 886 add %o0, 1, %o0
>> 887 4:
>> 888 lduh [%o1], %g2
>> 889 add %o1, 2, %o1
>> 890 srl %g2, 8, %g3
>> 891 sub %o2, 2, %o2
>> 892 stb %g3, [%o0]
>> 893 add %o0, 2, %o0
>> 894 stb %g2, [%o0 - 1]
>> 895 3:
>> 896 andcc %o1, 4, %g0
>> 897
>> 898 bne 2f
>> 899 cmp %o5, 1
>> 900
>> 901 ld [%o1], %o4
>> 902 srl %o4, 24, %g2
>> 903 stb %g2, [%o0]
>> 904 srl %o4, 16, %g3
>> 905 stb %g3, [%o0 + 1]
>> 906 srl %o4, 8, %g2
>> 907 stb %g2, [%o0 + 2]
>> 908 sub %o2, 4, %o2
>> 909 stb %o4, [%o0 + 3]
>> 910 add %o1, 4, %o1
>> 911 add %o0, 4, %o0
>> 912 2:
>> 913 be 33f
>> 914 cmp %o5, 2
>> 915 be 32f
>> 916 sub %o2, 4, %o2
>> 917 31:
>> 918 ld [%o1], %g2
>> 919 add %o1, 4, %o1
>> 920 srl %g2, 24, %g3
>> 921 and %o0, 7, %g5
>> 922 stb %g3, [%o0]
>> 923 cmp %g5, 7
>> 924 sll %g2, 8, %g1
>> 925 add %o0, 4, %o0
>> 926 be 41f
>> 927 and %o2, 0xffffffc0, %o3
>> 928 ld [%o0 - 7], %o4
>> 929 4:
>> 930 SMOVE_CHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 931 SMOVE_CHUNK(o1, o0, 0x10, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 932 SMOVE_CHUNK(o1, o0, 0x20, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 933 SMOVE_CHUNK(o1, o0, 0x30, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 934 subcc %o3, 64, %o3
>> 935 add %o1, 64, %o1
>> 936 bne 4b
>> 937 add %o0, 64, %o0
>> 938
>> 939 andcc %o2, 0x30, %o3
>> 940 be,a 1f
>> 941 srl %g1, 16, %g2
>> 942 4:
>> 943 SMOVE_CHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 944 subcc %o3, 16, %o3
>> 945 add %o1, 16, %o1
>> 946 bne 4b
>> 947 add %o0, 16, %o0
>> 948
>> 949 srl %g1, 16, %g2
>> 950 1:
>> 951 st %o4, [%o0 - 7]
>> 952 sth %g2, [%o0 - 3]
>> 953 srl %g1, 8, %g4
>> 954 b 88f
>> 955 stb %g4, [%o0 - 1]
>> 956 32:
>> 957 ld [%o1], %g2
>> 958 add %o1, 4, %o1
>> 959 srl %g2, 16, %g3
>> 960 and %o0, 7, %g5
>> 961 sth %g3, [%o0]
>> 962 cmp %g5, 6
>> 963 sll %g2, 16, %g1
>> 964 add %o0, 4, %o0
>> 965 be 42f
>> 966 and %o2, 0xffffffc0, %o3
>> 967 ld [%o0 - 6], %o4
>> 968 4:
>> 969 SMOVE_CHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 970 SMOVE_CHUNK(o1, o0, 0x10, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 971 SMOVE_CHUNK(o1, o0, 0x20, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 972 SMOVE_CHUNK(o1, o0, 0x30, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 973 subcc %o3, 64, %o3
>> 974 add %o1, 64, %o1
>> 975 bne 4b
>> 976 add %o0, 64, %o0
>> 977
>> 978 andcc %o2, 0x30, %o3
>> 979 be,a 1f
>> 980 srl %g1, 16, %g2
>> 981 4:
>> 982 SMOVE_CHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 983 subcc %o3, 16, %o3
>> 984 add %o1, 16, %o1
>> 985 bne 4b
>> 986 add %o0, 16, %o0
>> 987
>> 988 srl %g1, 16, %g2
>> 989 1:
>> 990 st %o4, [%o0 - 6]
>> 991 b 88f
>> 992 sth %g2, [%o0 - 2]
>> 993 33:
>> 994 ld [%o1], %g2
>> 995 sub %o2, 4, %o2
>> 996 srl %g2, 24, %g3
>> 997 and %o0, 7, %g5
>> 998 stb %g3, [%o0]
>> 999 cmp %g5, 5
>> 1000 srl %g2, 8, %g4
>> 1001 sll %g2, 24, %g1
>> 1002 sth %g4, [%o0 + 1]
>> 1003 add %o1, 4, %o1
>> 1004 be 43f
>> 1005 and %o2, 0xffffffc0, %o3
>> 1006
>> 1007 ld [%o0 - 1], %o4
>> 1008 add %o0, 4, %o0
>> 1009 4:
>> 1010 SMOVE_CHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, -1)
>> 1011 SMOVE_CHUNK(o1, o0, 0x10, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, -1)
>> 1012 SMOVE_CHUNK(o1, o0, 0x20, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, -1)
>> 1013 SMOVE_CHUNK(o1, o0, 0x30, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, -1)
>> 1014 subcc %o3, 64, %o3
>> 1015 add %o1, 64, %o1
>> 1016 bne 4b
>> 1017 add %o0, 64, %o0
>> 1018
>> 1019 andcc %o2, 0x30, %o3
>> 1020 be,a 1f
>> 1021 srl %g1, 24, %g2
>> 1022 4:
>> 1023 SMOVE_CHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, -1)
>> 1024 subcc %o3, 16, %o3
>> 1025 add %o1, 16, %o1
>> 1026 bne 4b
>> 1027 add %o0, 16, %o0
>> 1028
>> 1029 srl %g1, 24, %g2
>> 1030 1:
>> 1031 st %o4, [%o0 - 5]
>> 1032 b 88f
>> 1033 stb %g2, [%o0 - 1]
>> 1034 41:
>> 1035 SMOVE_ALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 1036 SMOVE_ALIGNCHUNK(o1, o0, 0x10, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 1037 SMOVE_ALIGNCHUNK(o1, o0, 0x20, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 1038 SMOVE_ALIGNCHUNK(o1, o0, 0x30, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 1039 subcc %o3, 64, %o3
>> 1040 add %o1, 64, %o1
>> 1041 bne 41b
>> 1042 add %o0, 64, %o0
>> 1043
>> 1044 andcc %o2, 0x30, %o3
>> 1045 be,a 1f
>> 1046 srl %g1, 16, %g2
>> 1047 4:
>> 1048 SMOVE_ALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 8, 24, -3)
>> 1049 subcc %o3, 16, %o3
>> 1050 add %o1, 16, %o1
>> 1051 bne 4b
>> 1052 add %o0, 16, %o0
>> 1053
>> 1054 srl %g1, 16, %g2
>> 1055 1:
>> 1056 sth %g2, [%o0 - 3]
>> 1057 srl %g1, 8, %g4
>> 1058 b 88f
>> 1059 stb %g4, [%o0 - 1]
>> 1060 43:
>> 1061 SMOVE_ALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, 3)
>> 1062 SMOVE_ALIGNCHUNK(o1, o0, 0x10, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, 3)
>> 1063 SMOVE_ALIGNCHUNK(o1, o0, 0x20, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, 3)
>> 1064 SMOVE_ALIGNCHUNK(o1, o0, 0x30, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, 3)
>> 1065 subcc %o3, 64, %o3
>> 1066 add %o1, 64, %o1
>> 1067 bne 43b
>> 1068 add %o0, 64, %o0
>> 1069
>> 1070 andcc %o2, 0x30, %o3
>> 1071 be,a 1f
>> 1072 srl %g1, 24, %g2
>> 1073 4:
>> 1074 SMOVE_ALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 24, 8, 3)
>> 1075 subcc %o3, 16, %o3
>> 1076 add %o1, 16, %o1
>> 1077 bne 4b
>> 1078 add %o0, 16, %o0
>> 1079
>> 1080 srl %g1, 24, %g2
>> 1081 1:
>> 1082 stb %g2, [%o0 + 3]
>> 1083 b 88f
>> 1084 add %o0, 4, %o0
>> 1085 42:
>> 1086 SMOVE_ALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 1087 SMOVE_ALIGNCHUNK(o1, o0, 0x10, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 1088 SMOVE_ALIGNCHUNK(o1, o0, 0x20, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 1089 SMOVE_ALIGNCHUNK(o1, o0, 0x30, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 1090 subcc %o3, 64, %o3
>> 1091 add %o1, 64, %o1
>> 1092 bne 42b
>> 1093 add %o0, 64, %o0
>> 1094
>> 1095 andcc %o2, 0x30, %o3
>> 1096 be,a 1f
>> 1097 srl %g1, 16, %g2
>> 1098 4:
>> 1099 SMOVE_ALIGNCHUNK(o1, o0, 0x00, g2, g3, g4, g5, o4, o5, g7, g1, 16, 16, -2)
>> 1100 subcc %o3, 16, %o3
>> 1101 add %o1, 16, %o1
>> 1102 bne 4b
>> 1103 add %o0, 16, %o0
>> 1104
>> 1105 srl %g1, 16, %g2
>> 1106 1:
>> 1107 sth %g2, [%o0 - 2]
>> 1108
>> 1109 /* Fall through */
>> 1110
>> 1111 #endif /* FASTER_NONALIGNED */
>> 1112
>> 1113 88: /* short_end */
>> 1114
>> 1115 and %o2, 0xe, %o3
>> 1116 20:
>> 1117 sethi %hi(89f), %o5
>> 1118 sll %o3, 3, %o4
>> 1119 add %o0, %o3, %o0
>> 1120 sub %o5, %o4, %o5
>> 1121 add %o1, %o3, %o1
>> 1122 jmpl %o5 + %lo(89f), %g0
>> 1123 andcc %o2, 1, %g0
>> 1124
>> 1125 MOVE_SHORTCHUNK(o1, o0, 0x0c, g2, g3)
>> 1126 MOVE_SHORTCHUNK(o1, o0, 0x0a, g2, g3)
>> 1127 MOVE_SHORTCHUNK(o1, o0, 0x08, g2, g3)
>> 1128 MOVE_SHORTCHUNK(o1, o0, 0x06, g2, g3)
>> 1129 MOVE_SHORTCHUNK(o1, o0, 0x04, g2, g3)
>> 1130 MOVE_SHORTCHUNK(o1, o0, 0x02, g2, g3)
>> 1131 MOVE_SHORTCHUNK(o1, o0, 0x00, g2, g3)
>> 1132
>> 1133 89: /* short_table_end */
>> 1134
>> 1135 be 1f
>> 1136 nop
>> 1137
>> 1138 ldub [%o1], %g2
>> 1139 stb %g2, [%o0]
>> 1140 1:
>> 1141 retl
>> 1142 RETL_INSN
>> 1143
>> 1144 90: /* short_aligned_end */
>> 1145 bne 88b
>> 1146 andcc %o2, 8, %g0
>> 1147
>> 1148 be 1f
>> 1149 andcc %o2, 4, %g0
>> 1150
>> 1151 ld [%o1 + 0x00], %g2
>> 1152 ld [%o1 + 0x04], %g3
>> 1153 add %o1, 8, %o1
>> 1154 st %g2, [%o0 + 0x00]
>> 1155 st %g3, [%o0 + 0x04]
>> 1156 add %o0, 8, %o0
>> 1157 1:
>> 1158 b 81b
>> 1159 mov %o2, %g1
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.