| 1 | /* Optimized strlen implementation using LoongArch LSX instructions. |
|---|
| 2 | Copyright (C) 2023-2024 Free Software Foundation, Inc. |
|---|
| 3 | |
|---|
| 4 | This file is part of the GNU C Library. |
|---|
| 5 | |
|---|
| 6 | The GNU C Library is free software; you can redistribute it and/or |
|---|
| 7 | modify it under the terms of the GNU Lesser General Public |
|---|
| 8 | License as published by the Free Software Foundation; either |
|---|
| 9 | version 2.1 of the License, or (at your option) any later version. |
|---|
| 10 | |
|---|
| 11 | The GNU C Library is distributed in the hope that it will be useful, |
|---|
| 12 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
|---|
| 13 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|---|
| 14 | Lesser General Public License for more details. |
|---|
| 15 | |
|---|
| 16 | You should have received a copy of the GNU Lesser General Public |
|---|
| 17 | License along with the GNU C Library. If not, see |
|---|
| 18 | <https://www.gnu.org/licenses/>. */ |
|---|
| 19 | |
|---|
| 20 | #include <sysdep.h> |
|---|
| 21 | #include <sys/regdef.h> |
|---|
| 22 | #include <sys/asm.h> |
|---|
| 23 | |
|---|
| 24 | #if IS_IN (libc) && !defined __loongarch_soft_float |
|---|
| 25 | |
|---|
| 26 | #ifndef AS_STRCHRNUL |
|---|
| 27 | # define STRCHR __strchr_lsx |
|---|
| 28 | #endif |
|---|
| 29 | |
|---|
| 30 | LEAF(STRCHR, 6) |
|---|
| 31 | andi t1, a0, 0xf |
|---|
| 32 | bstrins.d a0, zero, 3, 0 |
|---|
| 33 | vld vr0, a0, 0 |
|---|
| 34 | li.d t2, -1 |
|---|
| 35 | |
|---|
| 36 | vreplgr2vr.b vr1, a1 |
|---|
| 37 | sll.d t3, t2, t1 |
|---|
| 38 | vxor.v vr2, vr0, vr1 |
|---|
| 39 | vmin.bu vr0, vr0, vr2 |
|---|
| 40 | |
|---|
| 41 | vmsknz.b vr0, vr0 |
|---|
| 42 | movfr2gr.s t0, fa0 |
|---|
| 43 | ext.w.h t0, t0 |
|---|
| 44 | orn t0, t0, t3 |
|---|
| 45 | |
|---|
| 46 | beq t0, t2, L(loop) |
|---|
| 47 | L(found): |
|---|
| 48 | cto.w t0, t0 |
|---|
| 49 | add.d a0, a0, t0 |
|---|
| 50 | #ifndef AS_STRCHRNUL |
|---|
| 51 | vreplve.b vr2, vr2, t0 |
|---|
| 52 | vpickve2gr.bu t1, vr2, 0 |
|---|
| 53 | masknez a0, a0, t1 |
|---|
| 54 | #endif |
|---|
| 55 | jr ra |
|---|
| 56 | |
|---|
| 57 | |
|---|
| 58 | L(loop): |
|---|
| 59 | vld vr0, a0, 16 |
|---|
| 60 | addi.d a0, a0, 16 |
|---|
| 61 | vxor.v vr2, vr0, vr1 |
|---|
| 62 | vmin.bu vr0, vr0, vr2 |
|---|
| 63 | |
|---|
| 64 | vsetanyeqz.b fcc0, vr0 |
|---|
| 65 | bceqz fcc0, L(loop) |
|---|
| 66 | vmsknz.b vr0, vr0 |
|---|
| 67 | movfr2gr.s t0, fa0 |
|---|
| 68 | |
|---|
| 69 | b L(found) |
|---|
| 70 | END(STRCHR) |
|---|
| 71 | |
|---|
| 72 | libc_hidden_builtin_def (STRCHR) |
|---|
| 73 | #endif |
|---|
| 74 | |
|---|
