| lh | 9ed821d | 2023-04-07 01:36:19 -0700 | [diff] [blame^] | 1 | /* |
| 2 | * Copyright (C) 2004-2007 Atmel Corporation |
| 3 | * |
| 4 | * This file is subject to the terms and conditions of the GNU Lesser General |
| 5 | * Public License. See the file "COPYING.LIB" in the main directory of this |
| 6 | * archive for more details. |
| 7 | */ |
| 8 | |
| 9 | /* Don't use r12 as dst since we must return it unmodified */ |
| 10 | #define dst r9 |
| 11 | #define src r11 |
| 12 | #define len r10 |
| 13 | |
| 14 | .text |
| 15 | .global memcpy |
| 16 | .type memcpy, @function |
| 17 | memcpy: |
| 18 | pref src[0] |
| 19 | mov dst, r12 |
| 20 | |
| 21 | /* If we have less than 32 bytes, don't do anything fancy */ |
| 22 | cp.w len, 32 |
| 23 | brge .Lmore_than_31 |
| 24 | |
| 25 | sub len, 1 |
| 26 | retlt r12 |
| 27 | 1: ld.ub r8, src++ |
| 28 | st.b dst++, r8 |
| 29 | sub len, 1 |
| 30 | brge 1b |
| 31 | retal r12 |
| 32 | |
| 33 | .Lmore_than_31: |
| 34 | pushm r0-r7, lr |
| 35 | |
| 36 | /* Check alignment */ |
| 37 | mov r8, src |
| 38 | andl r8, 31, COH |
| 39 | brne .Lunaligned_src |
| 40 | mov r8, dst |
| 41 | andl r8, 3, COH |
| 42 | brne .Lunaligned_dst |
| 43 | |
| 44 | .Laligned_copy: |
| 45 | sub len, 32 |
| 46 | brlt .Lless_than_32 |
| 47 | |
| 48 | 1: /* Copy 32 bytes at a time */ |
| 49 | ldm src, r0-r7 |
| 50 | sub src, -32 |
| 51 | stm dst, r0-r7 |
| 52 | sub dst, -32 |
| 53 | sub len, 32 |
| 54 | brge 1b |
| 55 | |
| 56 | .Lless_than_32: |
| 57 | /* Copy 16 more bytes if possible */ |
| 58 | sub len, -16 |
| 59 | brlt .Lless_than_16 |
| 60 | ldm src, r0-r3 |
| 61 | sub src, -16 |
| 62 | sub len, 16 |
| 63 | stm dst, r0-r3 |
| 64 | sub dst, -16 |
| 65 | |
| 66 | .Lless_than_16: |
| 67 | /* Do the remaining as byte copies */ |
| 68 | neg len |
| 69 | add pc, pc, len << 2 |
| 70 | .rept 15 |
| 71 | ld.ub r0, src++ |
| 72 | st.b dst++, r0 |
| 73 | .endr |
| 74 | |
| 75 | popm r0-r7, pc |
| 76 | |
| 77 | .Lunaligned_src: |
| 78 | /* Make src cacheline-aligned. r8 = (src & 31) */ |
| 79 | rsub r8, r8, 32 |
| 80 | sub len, r8 |
| 81 | 1: ld.ub r0, src++ |
| 82 | st.b dst++, r0 |
| 83 | sub r8, 1 |
| 84 | brne 1b |
| 85 | |
| 86 | /* If dst is word-aligned, we're ready to go */ |
| 87 | pref src[0] |
| 88 | mov r8, 3 |
| 89 | tst dst, r8 |
| 90 | breq .Laligned_copy |
| 91 | |
| 92 | .Lunaligned_dst: |
| 93 | /* src is aligned, but dst is not. Expect bad performance */ |
| 94 | sub len, 4 |
| 95 | brlt 2f |
| 96 | 1: ld.w r0, src++ |
| 97 | st.w dst++, r0 |
| 98 | sub len, 4 |
| 99 | brge 1b |
| 100 | |
| 101 | 2: neg len |
| 102 | add pc, pc, len << 2 |
| 103 | .rept 3 |
| 104 | ld.ub r0, src++ |
| 105 | st.b dst++, r0 |
| 106 | .endr |
| 107 | |
| 108 | popm r0-r7, pc |
| 109 | .size memcpy, . - memcpy |
| 110 | |
| 111 | libc_hidden_def(memcpy) |