| Palmer Dabbelt | 5d8544e | 2017-07-10 18:03:19 -0700 | [diff] [blame] | 1 | /* |
| 2 | * Copyright (C) 2013 Regents of the University of California |
| 3 | * |
| 4 | * This program is free software; you can redistribute it and/or |
| 5 | * modify it under the terms of the GNU General Public License |
| 6 | * as published by the Free Software Foundation, version 2. |
| 7 | * |
| 8 | * This program is distributed in the hope that it will be useful, |
| 9 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 10 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 11 | * GNU General Public License for more details. |
| 12 | */ |
| 13 | |
| 14 | #include <linux/linkage.h> |
| 15 | #include <asm/asm.h> |
| 16 | |
| 17 | /* void *memcpy(void *, const void *, size_t) */ |
| 18 | ENTRY(memcpy) |
| 19 | move t6, a0 /* Preserve return value */ |
| 20 | |
| 21 | /* Defer to byte-oriented copy for small sizes */ |
| 22 | sltiu a3, a2, 128 |
| 23 | bnez a3, 4f |
| 24 | /* Use word-oriented copy only if low-order bits match */ |
| 25 | andi a3, t6, SZREG-1 |
| 26 | andi a4, a1, SZREG-1 |
| 27 | bne a3, a4, 4f |
| 28 | |
| 29 | beqz a3, 2f /* Skip if already aligned */ |
| 30 | /* |
| 31 | * Round to nearest double word-aligned address |
| 32 | * greater than or equal to start address |
| 33 | */ |
| 34 | andi a3, a1, ~(SZREG-1) |
| 35 | addi a3, a3, SZREG |
| 36 | /* Handle initial misalignment */ |
| 37 | sub a4, a3, a1 |
| 38 | 1: |
| 39 | lb a5, 0(a1) |
| 40 | addi a1, a1, 1 |
| 41 | sb a5, 0(t6) |
| 42 | addi t6, t6, 1 |
| 43 | bltu a1, a3, 1b |
| 44 | sub a2, a2, a4 /* Update count */ |
| 45 | |
| 46 | 2: |
| 47 | andi a4, a2, ~((16*SZREG)-1) |
| 48 | beqz a4, 4f |
| 49 | add a3, a1, a4 |
| 50 | 3: |
| 51 | REG_L a4, 0(a1) |
| 52 | REG_L a5, SZREG(a1) |
| 53 | REG_L a6, 2*SZREG(a1) |
| 54 | REG_L a7, 3*SZREG(a1) |
| 55 | REG_L t0, 4*SZREG(a1) |
| 56 | REG_L t1, 5*SZREG(a1) |
| 57 | REG_L t2, 6*SZREG(a1) |
| 58 | REG_L t3, 7*SZREG(a1) |
| 59 | REG_L t4, 8*SZREG(a1) |
| 60 | REG_L t5, 9*SZREG(a1) |
| 61 | REG_S a4, 0(t6) |
| 62 | REG_S a5, SZREG(t6) |
| 63 | REG_S a6, 2*SZREG(t6) |
| 64 | REG_S a7, 3*SZREG(t6) |
| 65 | REG_S t0, 4*SZREG(t6) |
| 66 | REG_S t1, 5*SZREG(t6) |
| 67 | REG_S t2, 6*SZREG(t6) |
| 68 | REG_S t3, 7*SZREG(t6) |
| 69 | REG_S t4, 8*SZREG(t6) |
| 70 | REG_S t5, 9*SZREG(t6) |
| 71 | REG_L a4, 10*SZREG(a1) |
| 72 | REG_L a5, 11*SZREG(a1) |
| 73 | REG_L a6, 12*SZREG(a1) |
| 74 | REG_L a7, 13*SZREG(a1) |
| 75 | REG_L t0, 14*SZREG(a1) |
| 76 | REG_L t1, 15*SZREG(a1) |
| 77 | addi a1, a1, 16*SZREG |
| 78 | REG_S a4, 10*SZREG(t6) |
| 79 | REG_S a5, 11*SZREG(t6) |
| 80 | REG_S a6, 12*SZREG(t6) |
| 81 | REG_S a7, 13*SZREG(t6) |
| 82 | REG_S t0, 14*SZREG(t6) |
| 83 | REG_S t1, 15*SZREG(t6) |
| 84 | addi t6, t6, 16*SZREG |
| 85 | bltu a1, a3, 3b |
| 86 | andi a2, a2, (16*SZREG)-1 /* Update count */ |
| 87 | |
| 88 | 4: |
| 89 | /* Handle trailing misalignment */ |
| 90 | beqz a2, 6f |
| 91 | add a3, a1, a2 |
| 92 | |
| 93 | /* Use word-oriented copy if co-aligned to word boundary */ |
| 94 | or a5, a1, t6 |
| 95 | or a5, a5, a3 |
| 96 | andi a5, a5, 3 |
| 97 | bnez a5, 5f |
| 98 | 7: |
| 99 | lw a4, 0(a1) |
| 100 | addi a1, a1, 4 |
| 101 | sw a4, 0(t6) |
| 102 | addi t6, t6, 4 |
| 103 | bltu a1, a3, 7b |
| 104 | |
| 105 | ret |
| 106 | |
| 107 | 5: |
| 108 | lb a4, 0(a1) |
| 109 | addi a1, a1, 1 |
| 110 | sb a4, 0(t6) |
| 111 | addi t6, t6, 1 |
| 112 | bltu a1, a3, 5b |
| 113 | 6: |
| 114 | ret |
| 115 | END(memcpy) |