zoukankan      html  css  js  c++  java
  • glibc-2.19 之 strlen 实现

    前几天遇到一个有意思的问题,实现strlen 不考虑线程安全:

    下面是我的实现:

    1 size_t strlen(const char* s)
    2 {
    3     const char* p = s;
    4     while (*p++);
    5     return p-1-s;
    6 }

    Glibc 2.19 的实现:
    针对此实现,函数头部分没太明白, size_t strlen (str) const char *str; 详细情况参见Glibc 2.19, 下面的实现还是比较经典的兼顾性能和体系机构。

     1 /* Return the length of the null-terminated string STR.  Scan for
     2    the null terminator quickly by testing four bytes at a time.  */
     3 size_t
     4 strlen (str)
     5      const char *str;
     6 {
     7   const char *char_ptr;
     8   const unsigned long int *longword_ptr;
     9   unsigned long int longword, himagic, lomagic;
    10 
    11   /* Handle the first few characters by reading one character at a time.
    12      Do this until CHAR_PTR is aligned on a longword boundary.  */
    13   for (char_ptr = str; ((unsigned long int) char_ptr
    14             & (sizeof (longword) - 1)) != 0;
    15        ++char_ptr)
    16     if (*char_ptr == '')
    17       return char_ptr - str;
    18 
    19   /* All these elucidatory comments refer to 4-byte longwords,
    20      but the theory applies equally well to 8-byte longwords.  */
    21 
    22   longword_ptr = (unsigned long int *) char_ptr;
    23 
    24   /* Bits 31, 24, 16, and 8 of this number are zero.  Call these bits
    25      the "holes."  Note that there is a hole just to the left of
    26      each byte, with an extra at the end:
    27 
    28      bits:  01111110 11111110 11111110 11111111
    29      bytes: AAAAAAAA BBBBBBBB CCCCCCCC DDDDDDDD
    30 
    31      The 1-bits make sure that carries propagate to the next 0-bit.
    32      The 0-bits provide holes for carries to fall into.  */
    33   himagic = 0x80808080L;
    34   lomagic = 0x01010101L;
    35   if (sizeof (longword) > 4)
    36     {
    37       /* 64-bit version of the magic.  */
    38       /* Do the shift in two steps to avoid a warning if long has 32 bits.  */
    39       himagic = ((himagic << 16) << 16) | himagic;
    40       lomagic = ((lomagic << 16) << 16) | lomagic;
    41     }
    42   if (sizeof (longword) > 8)
    43     abort ();
    44 
    45   /* Instead of the traditional loop which tests each character,
    46      we will test a longword at a time.  The tricky part is testing
    47      if *any of the four* bytes in the longword in question are zero.  */
    48   for (;;)
    49     {
    50       longword = *longword_ptr++;
    51 
    52       if (((longword - lomagic) & ~longword & himagic) != 0)
    53     {
    54       /* Which of the bytes was the zero?  If none of them were, it was
    55          a misfire; continue the search.  */
    56 
    57       const char *cp = (const char *) (longword_ptr - 1);
    58 
    59       if (cp[0] == 0)
    60         return cp - str;
    61       if (cp[1] == 0)
    62         return cp - str + 1;
    63       if (cp[2] == 0)
    64         return cp - str + 2;
    65       if (cp[3] == 0)
    66         return cp - str + 3;
    67       if (sizeof (longword) > 4)
    68         {
    69           if (cp[4] == 0)
    70         return cp - str + 4;
    71           if (cp[5] == 0)
    72         return cp - str + 5;
    73           if (cp[6] == 0)
    74         return cp - str + 6;
    75           if (cp[7] == 0)
    76         return cp - str + 7;
    77         }
    78     }
    79     }
    80 }
  • 相关阅读:
    C++内置类型对象之间的转换
    快速排序
    面试题7:用两个栈实现队列
    面试题6:重建二叉树
    poj 3264(线段树)
    poj 3038
    poj 并查集
    poj 1270(toposort)
    poj 2503(字符串)
    poj 3687(拓扑排序)
  • 原文地址:https://www.cnblogs.com/lihuibng/p/3871267.html
Copyright © 2011-2022 走看看