string/strchr.c

   1 /* Copyright (C) 1991-2015 Free Software Foundation, Inc.
   2    This file is part of the GNU C Library.
   3    Based on strlen implementation by Torbjorn Granlund (tege@sics.se),
   4    with help from Dan Sahlin (dan@sics.se) and
   5    bug fix and commentary by Jim Blandy (jimb@ai.mit.edu);
   6    adaptation to strchr suggested by Dick Karpinski (dick@cca.ucsf.edu),
   7    and implemented by Roland McGrath (roland@ai.mit.edu).
   8
   9    The GNU C Library is free software; you can redistribute it and/or
  10    modify it under the terms of the GNU Lesser General Public
  11    License as published by the Free Software Foundation; either
  12    version 2.1 of the License, or (at your option) any later version.
  13
  14    The GNU C Library is distributed in the hope that it will be useful,
  15    but WITHOUT ANY WARRANTY; without even the implied warranty of
  16    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17    Lesser General Public License for more details.
  18
  19    You should have received a copy of the GNU Lesser General Public
  20    License along with the GNU C Library; if not, see
  21    <http://www.gnu.org/licenses/>.  */
  22
  23 #include <string.h>
  24 #include <stdlib.h>
  25
  26 #undef strchr
  27
  28 /* Find the first occurrence of C in S.  */
  29 char *
  30 strchr (const char *s, int c_in)
  31 {
  32   const unsigned char *char_ptr;
  33   const unsigned long int *longword_ptr;
  34   unsigned long int longword, magic_bits, charmask;
  35   unsigned char c;
  36
  37   c = (unsigned char) c_in;
  38
  39   /* Handle the first few characters by reading one character at a time.
  40      Do this until CHAR_PTR is aligned on a longword boundary.  */
  41   for (char_ptr = (const unsigned char *) s;
  42        ((unsigned long int) char_ptr & (sizeof (longword) - 1)) != 0;
  43        ++char_ptr)
  44     if (*char_ptr == c)
  45       return (void *) char_ptr;
  46     else if (*char_ptr == '\0')
  47       return NULL;
  48
  49   /* All these elucidatory comments refer to 4-byte longwords,
  50      but the theory applies equally well to 8-byte longwords.  */
  51
  52   longword_ptr = (unsigned long int *) char_ptr;
  53
  54   /* Bits 31, 24, 16, and 8 of this number are zero.  Call these bits
  55      the "holes."  Note that there is a hole just to the left of
  56      each byte, with an extra at the end:
  57
  58      bits:  01111110 11111110 11111110 11111111
  59      bytes: AAAAAAAA BBBBBBBB CCCCCCCC DDDDDDDD
  60
  61      The 1-bits make sure that carries propagate to the next 0-bit.
  62      The 0-bits provide holes for carries to fall into.  */
  63   magic_bits = -1;
  64   magic_bits = magic_bits / 0xff * 0xfe << 1 >> 1 | 1;
  65
  66   /* Set up a longword, each of whose bytes is C.  */
  67   charmask = c | (c << 8);
  68   charmask |= charmask << 16;
  69   if (sizeof (longword) > 4)
  70     /* Do the shift in two steps to avoid a warning if long has 32 bits.  */
  71     charmask |= (charmask << 16) << 16;
  72   if (sizeof (longword) > 8)
  73     abort ();
  74
  75   /* Instead of the traditional loop which tests each character,
  76      we will test a longword at a time.  The tricky part is testing
  77      if *any of the four* bytes in the longword in question are zero.  */
  78   for (;;)
  79     {
  80       /* We tentatively exit the loop if adding MAGIC_BITS to
  81          LONGWORD fails to change any of the hole bits of LONGWORD.
  82
  83          1) Is this safe?  Will it catch all the zero bytes?
  84          Suppose there is a byte with all zeros.  Any carry bits
  85          propagating from its left will fall into the hole at its
  86          least significant bit and stop.  Since there will be no
  87          carry from its most significant bit, the LSB of the
  88          byte to the left will be unchanged, and the zero will be
  89          detected.
  90
  91          2) Is this worthwhile?  Will it ignore everything except
  92          zero bytes?  Suppose every byte of LONGWORD has a bit set
  93          somewhere.  There will be a carry into bit 8.  If bit 8
  94          is set, this will carry into bit 16.  If bit 8 is clear,
  95          one of bits 9-15 must be set, so there will be a carry
  96          into bit 16.  Similarly, there will be a carry into bit
  97          24.  If one of bits 24-30 is set, there will be a carry
  98          into bit 31, so all of the hole bits will be changed.
  99
 100          The one misfire occurs when bits 24-30 are clear and bit
 101          31 is set; in this case, the hole at bit 31 is not
 102          changed.  If we had access to the processor carry flag,
 103          we could close this loophole by putting the fourth hole
 104          at bit 32!
 105
 106          So it ignores everything except 128's, when they're aligned
 107          properly.
 108
 109          3) But wait!  Aren't we looking for C as well as zero?
 110          Good point.  So what we do is XOR LONGWORD with a longword,
 111          each of whose bytes is C.  This turns each byte that is C
 112          into a zero.  */
 113
 114       longword = *longword_ptr++;
 115
 116       /* Add MAGIC_BITS to LONGWORD.  */
 117       if ((((longword + magic_bits)
 118
 119             /* Set those bits that were unchanged by the addition.  */
 120             ^ ~longword)
 121
 122            /* Look at only the hole bits.  If any of the hole bits
 123               are unchanged, most likely one of the bytes was a
 124               zero.  */
 125            & ~magic_bits) != 0 ||
 126
 127           /* That caught zeroes.  Now test for C.  */
 128           ((((longword ^ charmask) + magic_bits) ^ ~(longword ^ charmask))
 129            & ~magic_bits) != 0)
 130         {
 131           /* Which of the bytes was C or zero?
 132              If none of them were, it was a misfire; continue the search.  */
 133
 134           const unsigned char *cp = (const unsigned char *) (longword_ptr - 1);
 135
 136           if (*cp == c)
 137             return (char *) cp;
 138           else if (*cp == '\0')
 139             return NULL;
 140           if (*++cp == c)
 141             return (char *) cp;
 142           else if (*cp == '\0')
 143             return NULL;
 144           if (*++cp == c)
 145             return (char *) cp;
 146           else if (*cp == '\0')
 147             return NULL;
 148           if (*++cp == c)
 149             return (char *) cp;
 150           else if (*cp == '\0')
 151             return NULL;
 152           if (sizeof (longword) > 4)
 153             {
 154               if (*++cp == c)
 155                 return (char *) cp;
 156               else if (*cp == '\0')
 157                 return NULL;
 158               if (*++cp == c)
 159                 return (char *) cp;
 160               else if (*cp == '\0')
 161                 return NULL;
 162               if (*++cp == c)
 163                 return (char *) cp;
 164               else if (*cp == '\0')
 165                 return NULL;
 166               if (*++cp == c)
 167                 return (char *) cp;
 168               else if (*cp == '\0')
 169                 return NULL;
 170             }
 171         }
 172     }
 173
 174   return NULL;
 175 }
 176
 177 #ifdef weak_alias
 178 #undef index
 179 weak_alias (strchr, index)
 180 #endif
 181 libc_hidden_builtin_def (strchr)