strlen.c revision 9ad2a73ed65d39c3a638ff9bd2d73250b41212c6
1/* 2 * Copyright (C) 2008 The Android Open Source Project 3 * All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * * Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * * Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in 12 * the documentation and/or other materials provided with the 13 * distribution. 14 * 15 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 16 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 17 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS 18 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE 19 * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, 20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, 21 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS 22 * OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED 23 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 24 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT 25 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 26 * SUCH DAMAGE. 27 */ 28#include <string.h> 29#include <stdint.h> 30#include <machine/cpu-features.h> 31 32size_t strlen(const char *s) 33{ 34 __builtin_prefetch(s); 35 __builtin_prefetch(s+32); 36 37 union { 38 const char *b; 39 const uint32_t *w; 40 uintptr_t i; 41 } u; 42 43 // these are some scratch variables for the asm code below 44 uint32_t v, t; 45 46 // initialize the string length to zero 47 size_t l = 0; 48 49 // align the pointer to a 32-bit word boundary 50 u.b = s; 51 while (u.i & 0x3) { 52 if (__builtin_expect(*u.b++ == 0, 0)) { 53 goto done; 54 } 55 l++; 56 } 57 58 // loop for each word, testing if it contains a zero byte 59 // if so, exit the loop and update the length. 60 // We need to process 32 bytes per loop to schedule PLD properly 61 // and achieve the maximum bus speed. 62 asm( 63 "ldr %[v], [%[s]], #4 \n" 64 "sub %[l], %[l], %[s] \n" 65 "0: \n" 66#if __ARM_HAVE_PLD 67 "pld [%[s], #64] \n" 68#endif 69 "sub %[t], %[v], %[mask], lsr #7\n" 70 "and %[t], %[t], %[mask] \n" 71 "bics %[t], %[t], %[v] \n" 72 "it eq \n" 73 "ldreq %[v], [%[s]], #4 \n" 74#if !defined(__OPTIMIZE_SIZE__) 75 "bne 1f \n" 76 "sub %[t], %[v], %[mask], lsr #7\n" 77 "and %[t], %[t], %[mask] \n" 78 "bics %[t], %[t], %[v] \n" 79 "it eq \n" 80 "ldreq %[v], [%[s]], #4 \n" 81 "bne 1f \n" 82 "sub %[t], %[v], %[mask], lsr #7\n" 83 "and %[t], %[t], %[mask] \n" 84 "bics %[t], %[t], %[v] \n" 85 "it eq \n" 86 "ldreq %[v], [%[s]], #4 \n" 87 "bne 1f \n" 88 "sub %[t], %[v], %[mask], lsr #7\n" 89 "and %[t], %[t], %[mask] \n" 90 "bics %[t], %[t], %[v] \n" 91 "it eq \n" 92 "ldreq %[v], [%[s]], #4 \n" 93 "bne 1f \n" 94 "sub %[t], %[v], %[mask], lsr #7\n" 95 "and %[t], %[t], %[mask] \n" 96 "bics %[t], %[t], %[v] \n" 97 "it eq \n" 98 "ldreq %[v], [%[s]], #4 \n" 99 "bne 1f \n" 100 "sub %[t], %[v], %[mask], lsr #7\n" 101 "and %[t], %[t], %[mask] \n" 102 "bics %[t], %[t], %[v] \n" 103 "it eq \n" 104 "ldreq %[v], [%[s]], #4 \n" 105 "bne 1f \n" 106 "sub %[t], %[v], %[mask], lsr #7\n" 107 "and %[t], %[t], %[mask] \n" 108 "bics %[t], %[t], %[v] \n" 109 "it eq \n" 110 "ldreq %[v], [%[s]], #4 \n" 111 "bne 1f \n" 112 "sub %[t], %[v], %[mask], lsr #7\n" 113 "and %[t], %[t], %[mask] \n" 114 "bics %[t], %[t], %[v] \n" 115 "it eq \n" 116 "ldreq %[v], [%[s]], #4 \n" 117#endif 118 "beq 0b \n" 119 "1: \n" 120 "add %[l], %[l], %[s] \n" 121 "tst %[v], #0xFF \n" 122 "beq 2f \n" 123 "add %[l], %[l], #1 \n" 124 "tst %[v], #0xFF00 \n" 125 "beq 2f \n" 126 "add %[l], %[l], #1 \n" 127 "tst %[v], #0xFF0000 \n" 128 "it ne \n" 129 "addne %[l], %[l], #1 \n" 130 "2: \n" 131 : [l]"=&r"(l), [v]"=&r"(v), [t]"=&r"(t), [s]"=&r"(u.b) 132 : "%[l]"(l), "%[s]"(u.b), [mask]"r"(0x80808080UL) 133 : "cc" 134 ); 135 136done: 137 return l; 138} 139