| David Howells | b920de1 | 2008-02-08 04:19:31 -0800 | [diff] [blame] | 1 | /* MN10300 Optimised simple memory to memory copy, with support for overlapping |
| 2 | * regions |
| 3 | * |
| 4 | * Copyright (C) 2007 Red Hat, Inc. All Rights Reserved. |
| 5 | * Written by David Howells (dhowells@redhat.com) |
| 6 | * |
| 7 | * This program is free software; you can redistribute it and/or |
| 8 | * modify it under the terms of the GNU General Public Licence |
| 9 | * as published by the Free Software Foundation; either version |
| 10 | * 2 of the Licence, or (at your option) any later version. |
| 11 | */ |
| 12 | #include <asm/cache.h> |
| 13 | |
| 14 | .section .text |
| 15 | .balign L1_CACHE_BYTES |
| 16 | |
| 17 | ############################################################################### |
| 18 | # |
| 19 | # void *memmove(void *dst, const void *src, size_t n) |
| 20 | # |
| 21 | ############################################################################### |
| 22 | .globl memmove |
| 23 | .type memmove,@function |
| 24 | memmove: |
| 25 | # fall back to memcpy if dst < src to work bottom up |
| 26 | cmp d1,d0 |
| 27 | bcs memmove_memcpy |
| 28 | |
| 29 | # work top down |
| 30 | movm [d2,d3],(sp) |
| 31 | mov d0,(12,sp) |
| 32 | mov d1,(16,sp) |
| 33 | mov (20,sp),d2 # count |
| 34 | add d0,d2,a0 # dst end |
| 35 | add d1,d2,a1 # src end |
| 36 | mov d0,e3 # the return value |
| 37 | |
| 38 | cmp +0,d2 |
| 39 | beq memmove_done # return if zero-length copy |
| 40 | |
| 41 | # see if the three parameters are all four-byte aligned |
| 42 | or d0,d1,d3 |
| 43 | or d2,d3 |
| 44 | and +3,d3 |
| 45 | bne memmove_1 # jump if not |
| 46 | |
| 47 | # we want to transfer as much as we can in chunks of 32 bytes |
| 48 | add -4,a1 |
| 49 | cmp +31,d2 |
| 50 | bls memmove_4_remainder # 4-byte aligned remainder |
| 51 | |
| 52 | add -32,d2 |
| 53 | mov +32,d3 |
| 54 | |
| 55 | memmove_4_loop: |
| 56 | mov (a1),d0 |
| 57 | sub_sub +4,a1,+4,a0 |
| 58 | mov d0,(a0) |
| 59 | mov (a1),d1 |
| 60 | sub_sub +4,a1,+4,a0 |
| 61 | mov d1,(a0) |
| 62 | |
| 63 | mov (a1),d0 |
| 64 | sub_sub +4,a1,+4,a0 |
| 65 | mov d0,(a0) |
| 66 | mov (a1),d1 |
| 67 | sub_sub +4,a1,+4,a0 |
| 68 | mov d1,(a0) |
| 69 | |
| 70 | mov (a1),d0 |
| 71 | sub_sub +4,a1,+4,a0 |
| 72 | mov d0,(a0) |
| 73 | mov (a1),d1 |
| 74 | sub_sub +4,a1,+4,a0 |
| 75 | mov d1,(a0) |
| 76 | |
| 77 | mov (a1),d0 |
| 78 | sub_sub +4,a1,+4,a0 |
| 79 | mov d0,(a0) |
| 80 | mov (a1),d1 |
| 81 | sub_sub +4,a1,+4,a0 |
| 82 | mov d1,(a0) |
| 83 | |
| 84 | sub d3,d2 |
| 85 | bcc memmove_4_loop |
| 86 | |
| 87 | add d3,d2 |
| 88 | beq memmove_4_no_remainder |
| 89 | |
| 90 | memmove_4_remainder: |
| 91 | # cut 4-7 words down to 0-3 |
| 92 | cmp +16,d2 |
| 93 | bcs memmove_4_three_or_fewer_words |
| 94 | mov (a1),d0 |
| 95 | sub_sub +4,a1,+4,a0 |
| 96 | mov d0,(a0) |
| 97 | mov (a1),d1 |
| 98 | sub_sub +4,a1,+4,a0 |
| 99 | mov d1,(a0) |
| 100 | mov (a1),e0 |
| 101 | sub_sub +4,a1,+4,a0 |
| 102 | mov e0,(a0) |
| 103 | mov (a1),e1 |
| 104 | sub_sub +4,a1,+4,a0 |
| 105 | mov e1,(a0) |
| 106 | add -16,d2 |
| 107 | beq memmove_4_no_remainder |
| 108 | |
| 109 | # copy the remaining 1, 2 or 3 words |
| 110 | memmove_4_three_or_fewer_words: |
| 111 | cmp +8,d2 |
| 112 | bcs memmove_4_one_word |
| 113 | beq memmove_4_two_words |
| 114 | mov (a1),d0 |
| 115 | sub_sub +4,a1,+4,a0 |
| 116 | mov d0,(a0) |
| 117 | memmove_4_two_words: |
| 118 | mov (a1),d0 |
| 119 | sub_sub +4,a1,+4,a0 |
| 120 | mov d0,(a0) |
| 121 | memmove_4_one_word: |
| 122 | mov (a1),d0 |
| 123 | sub_sub +4,a1,+4,a0 |
| 124 | mov d0,(a0) |
| 125 | |
| 126 | memmove_4_no_remainder: |
| 127 | # check we copied the correct amount |
| 128 | # TODO: REMOVE CHECK |
| 129 | sub e3,a0,d2 |
| 130 | beq memmove_done |
| 131 | break |
| 132 | break |
| 133 | break |
| 134 | |
| 135 | memmove_done: |
| 136 | mov e3,a0 |
| 137 | ret [d2,d3],8 |
| 138 | |
| 139 | # handle misaligned copying |
| 140 | memmove_1: |
| 141 | add -1,a1 |
| 142 | add -1,d2 |
| 143 | mov +1,d3 |
| 144 | setlb # setlb requires the next insns |
| 145 | # to occupy exactly 4 bytes |
| 146 | |
| 147 | sub d3,d2 |
| 148 | movbu (a1),d0 |
| 149 | sub_sub d3,a1,d3,a0 |
| 150 | movbu d0,(a0) |
| 151 | lcc |
| 152 | |
| 153 | mov e3,a0 |
| 154 | ret [d2,d3],8 |
| 155 | |
| 156 | memmove_memcpy: |
| 157 | jmp memcpy |
| 158 | |
| 159 | memmove_end: |
| 160 | .size memmove, memmove_end-memmove |