1 | /* strchr - find a character in a string |
2 | |
3 | Copyright (C) 2014-2022 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 chrin w1 |
31 | #define result x0 |
32 | |
33 | #define src x2 |
34 | #define tmp1 x1 |
35 | #define wtmp2 w3 |
36 | #define tmp3 x3 |
37 | |
38 | #define vrepchr v0 |
39 | #define vdata v1 |
40 | #define qdata q1 |
41 | #define vhas_nul v2 |
42 | #define vhas_chr v3 |
43 | #define vrepmask v4 |
44 | #define vrepmask2 v5 |
45 | #define vend v6 |
46 | #define dend d6 |
47 | |
48 | /* Core algorithm. |
49 | |
50 | For each 16-byte chunk we calculate a 64-bit syndrome value with four bits |
51 | per byte. For even bytes, bits 0-1 are set if the relevant byte matched the |
52 | requested character, bits 2-3 are set if the byte is NUL (or matched), and |
53 | bits 4-7 are not used and must be zero if none of bits 0-3 are set). Odd |
54 | bytes set bits 4-7 so that adjacent bytes can be merged. Since the bits |
55 | in the syndrome reflect the order in which things occur in the original |
56 | string, counting trailing zeros identifies exactly which byte matched. */ |
57 | |
58 | ENTRY (strchr) |
59 | PTR_ARG (0) |
60 | bic src, srcin, 15 |
61 | dup vrepchr.16b, chrin |
62 | ld1 {vdata.16b}, [src] |
63 | mov wtmp2, 0x3003 |
64 | dup vrepmask.8h, wtmp2 |
65 | cmeq vhas_nul.16b, vdata.16b, 0 |
66 | cmeq vhas_chr.16b, vdata.16b, vrepchr.16b |
67 | mov wtmp2, 0xf00f |
68 | dup vrepmask2.8h, wtmp2 |
69 | |
70 | bit vhas_nul.16b, vhas_chr.16b, vrepmask.16b |
71 | and vhas_nul.16b, vhas_nul.16b, vrepmask2.16b |
72 | lsl tmp3, srcin, 2 |
73 | addp vend.16b, vhas_nul.16b, vhas_nul.16b /* 128->64 */ |
74 | |
75 | fmov tmp1, dend |
76 | lsr tmp1, tmp1, tmp3 |
77 | cbz tmp1, L(loop) |
78 | |
79 | rbit tmp1, tmp1 |
80 | clz tmp1, tmp1 |
81 | /* Tmp1 is an even multiple of 2 if the target character was |
82 | found first. Otherwise we've found the end of string. */ |
83 | tst tmp1, 2 |
84 | add result, srcin, tmp1, lsr 2 |
85 | csel result, result, xzr, eq |
86 | ret |
87 | |
88 | .p2align 4 |
89 | L(loop): |
90 | ldr qdata, [src, 16]! |
91 | cmeq vhas_chr.16b, vdata.16b, vrepchr.16b |
92 | cmhs vhas_nul.16b, vhas_chr.16b, vdata.16b |
93 | umaxp vend.16b, vhas_nul.16b, vhas_nul.16b |
94 | fmov tmp1, dend |
95 | cbz tmp1, L(loop) |
96 | |
97 | #ifdef __AARCH64EB__ |
98 | bif vhas_nul.16b, vhas_chr.16b, vrepmask.16b |
99 | and vhas_nul.16b, vhas_nul.16b, vrepmask2.16b |
100 | addp vend.16b, vhas_nul.16b, vhas_nul.16b /* 128->64 */ |
101 | fmov tmp1, dend |
102 | #else |
103 | bit vhas_nul.16b, vhas_chr.16b, vrepmask.16b |
104 | and vhas_nul.16b, vhas_nul.16b, vrepmask2.16b |
105 | addp vend.16b, vhas_nul.16b, vhas_nul.16b /* 128->64 */ |
106 | fmov tmp1, dend |
107 | rbit tmp1, tmp1 |
108 | #endif |
109 | clz tmp1, tmp1 |
110 | /* Tmp1 is an even multiple of 2 if the target character was |
111 | found first. Otherwise we've found the end of string. */ |
112 | tst tmp1, 2 |
113 | add result, src, tmp1, lsr 2 |
114 | csel result, result, xzr, eq |
115 | ret |
116 | |
117 | END (strchr) |
118 | libc_hidden_builtin_def (strchr) |
119 | weak_alias (strchr, index) |
120 | |