| 1 | /* strnlen - calculate the length of a string with limit. |
| 2 | |
| 3 | Copyright (C) 2013-2024 Free Software Foundation, Inc. |
| 4 | |
| 5 | This file is part of the GNU C Library. |
| 6 | |
| 7 | The GNU C Library is free software; you can redistribute it and/or |
| 8 | modify it under the terms of the GNU Lesser General Public |
| 9 | License as published by the Free Software Foundation; either |
| 10 | version 2.1 of the License, or (at your option) any later version. |
| 11 | |
| 12 | The GNU C Library is distributed in the hope that it will be useful, |
| 13 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 14 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 15 | Lesser General Public License for more details. |
| 16 | |
| 17 | You should have received a copy of the GNU Lesser General Public |
| 18 | License along with the GNU C Library. If not, see |
| 19 | <https://www.gnu.org/licenses/>. */ |
| 20 | |
| 21 | #include <sysdep.h> |
| 22 | |
| 23 | /* Assumptions: |
| 24 | * |
| 25 | * ARMv8-a, AArch64, Advanced SIMD. |
| 26 | * MTE compatible. |
| 27 | */ |
| 28 | |
| 29 | #define srcin x0 |
| 30 | #define cntin x1 |
| 31 | #define result x0 |
| 32 | |
| 33 | #define src x2 |
| 34 | #define synd x3 |
| 35 | #define shift x4 |
| 36 | #define tmp x4 |
| 37 | #define cntrem x5 |
| 38 | |
| 39 | #define qdata q0 |
| 40 | #define vdata v0 |
| 41 | #define vhas_chr v1 |
| 42 | #define vend v2 |
| 43 | #define dend d2 |
| 44 | |
| 45 | /* |
| 46 | Core algorithm: |
| 47 | Process the string in 16-byte aligned chunks. Compute a 64-bit mask with |
| 48 | four bits per byte using the shrn instruction. A count trailing zeros then |
| 49 | identifies the first zero byte. */ |
| 50 | |
| 51 | ENTRY (__strnlen) |
| 52 | PTR_ARG (0) |
| 53 | SIZE_ARG (1) |
| 54 | bic src, srcin, 15 |
| 55 | cbz cntin, L(nomatch) |
| 56 | ld1 {vdata.16b}, [src] |
| 57 | cmeq vhas_chr.16b, vdata.16b, 0 |
| 58 | lsl shift, srcin, 2 |
| 59 | shrn vend.8b, vhas_chr.8h, 4 /* 128->64 */ |
| 60 | fmov synd, dend |
| 61 | lsr synd, synd, shift |
| 62 | cbz synd, L(start_loop) |
| 63 | L(finish): |
| 64 | rbit synd, synd |
| 65 | clz synd, synd |
| 66 | lsr result, synd, 2 |
| 67 | cmp cntin, result |
| 68 | csel result, cntin, result, ls |
| 69 | ret |
| 70 | |
| 71 | L(nomatch): |
| 72 | mov result, cntin |
| 73 | ret |
| 74 | |
| 75 | L(start_loop): |
| 76 | sub tmp, src, srcin |
| 77 | add tmp, tmp, 17 |
| 78 | subs cntrem, cntin, tmp |
| 79 | b.lo L(nomatch) |
| 80 | |
| 81 | /* Make sure that it won't overread by a 16-byte chunk */ |
| 82 | tbz cntrem, 4, L(loop32_2) |
| 83 | sub src, src, 16 |
| 84 | .p2align 5 |
| 85 | L(loop32): |
| 86 | ldr qdata, [src, 32]! |
| 87 | cmeq vhas_chr.16b, vdata.16b, 0 |
| 88 | umaxp vend.16b, vhas_chr.16b, vhas_chr.16b /* 128->64 */ |
| 89 | fmov synd, dend |
| 90 | cbnz synd, L(end) |
| 91 | L(loop32_2): |
| 92 | ldr qdata, [src, 16] |
| 93 | subs cntrem, cntrem, 32 |
| 94 | cmeq vhas_chr.16b, vdata.16b, 0 |
| 95 | b.lo L(end_2) |
| 96 | umaxp vend.16b, vhas_chr.16b, vhas_chr.16b /* 128->64 */ |
| 97 | fmov synd, dend |
| 98 | cbz synd, L(loop32) |
| 99 | L(end_2): |
| 100 | add src, src, 16 |
| 101 | L(end): |
| 102 | shrn vend.8b, vhas_chr.8h, 4 /* 128->64 */ |
| 103 | sub result, src, srcin |
| 104 | fmov synd, dend |
| 105 | #ifndef __AARCH64EB__ |
| 106 | rbit synd, synd |
| 107 | #endif |
| 108 | clz synd, synd |
| 109 | add result, result, synd, lsr 2 |
| 110 | cmp cntin, result |
| 111 | csel result, cntin, result, ls |
| 112 | ret |
| 113 | |
| 114 | END (__strnlen) |
| 115 | libc_hidden_def (__strnlen) |
| 116 | weak_alias (__strnlen, strnlen) |
| 117 | libc_hidden_def (strnlen) |
| 118 | |