string/strchrnul.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 <memcopy.h>
  25 #include <stdlib.h>
  26
  27 #undef __strchrnul
  28 #undef strchrnul
  29
  30 #ifndef STRCHRNUL
  31 # define STRCHRNUL __strchrnul
  32 #endif
  33
  34 /* Find the first occurrence of C in S or the final NUL byte.  */
  35 char *
  36 STRCHRNUL (s, c_in)
  37      const char *s;
  38      int c_in;
  39 {
  40   const unsigned char *char_ptr;
  41   const unsigned long int *longword_ptr;
  42   unsigned long int longword, magic_bits, charmask;
  43   unsigned char c;
  44
  45   c = (unsigned char) c_in;
  46
  47   /* Handle the first few characters by reading one character at a time.
  48      Do this until CHAR_PTR is aligned on a longword boundary.  */
  49   for (char_ptr = (const unsigned char *) s;
  50        ((unsigned long int) char_ptr & (sizeof (longword) - 1)) != 0;
  51        ++char_ptr)
  52     if (*char_ptr == c || *char_ptr == '\0')
  53       return (void *) char_ptr;
  54
  55   /* All these elucidatory comments refer to 4-byte longwords,
  56      but the theory applies equally well to 8-byte longwords.  */
  57
  58   longword_ptr = (unsigned long int *) char_ptr;
  59
  60   /* Bits 31, 24, 16, and 8 of this number are zero.  Call these bits
  61      the "holes."  Note that there is a hole just to the left of
  62      each byte, with an extra at the end:
  63
  64      bits:  01111110 11111110 11111110 11111111
  65      bytes: AAAAAAAA BBBBBBBB CCCCCCCC DDDDDDDD
  66
  67      The 1-bits make sure that carries propagate to the next 0-bit.
  68      The 0-bits provide holes for carries to fall into.  */
  69   switch (sizeof (longword))
  70     {
  71     case 4: magic_bits = 0x7efefeffL; break;
  72     case 8: magic_bits = ((0x7efefefeL << 16) << 16) | 0xfefefeffL; break;
  73     default:
  74       abort ();
  75     }
  76
  77   /* Set up a longword, each of whose bytes is C.  */
  78   charmask = c | (c << 8);
  79   charmask |= charmask << 16;
  80   if (sizeof (longword) > 4)
  81     /* Do the shift in two steps to avoid a warning if long has 32 bits.  */
  82     charmask |= (charmask << 16) << 16;
  83   if (sizeof (longword) > 8)
  84     abort ();
  85
  86   /* Instead of the traditional loop which tests each character,
  87      we will test a longword at a time.  The tricky part is testing
  88      if *any of the four* bytes in the longword in question are zero.  */
  89   for (;;)
  90     {
  91       /* We tentatively exit the loop if adding MAGIC_BITS to
  92          LONGWORD fails to change any of the hole bits of LONGWORD.
  93
  94          1) Is this safe?  Will it catch all the zero bytes?
  95          Suppose there is a byte with all zeros.  Any carry bits
  96          propagating from its left will fall into the hole at its
  97          least significant bit and stop.  Since there will be no
  98          carry from its most significant bit, the LSB of the
  99          byte to the left will be unchanged, and the zero will be
 100          detected.
 101
 102          2) Is this worthwhile?  Will it ignore everything except
 103          zero bytes?  Suppose every byte of LONGWORD has a bit set
 104          somewhere.  There will be a carry into bit 8.  If bit 8
 105          is set, this will carry into bit 16.  If bit 8 is clear,
 106          one of bits 9-15 must be set, so there will be a carry
 107          into bit 16.  Similarly, there will be a carry into bit
 108          24.  If one of bits 24-30 is set, there will be a carry
 109          into bit 31, so all of the hole bits will be changed.
 110
 111          The one misfire occurs when bits 24-30 are clear and bit
 112          31 is set; in this case, the hole at bit 31 is not
 113          changed.  If we had access to the processor carry flag,
 114          we could close this loophole by putting the fourth hole
 115          at bit 32!
 116
 117          So it ignores everything except 128's, when they're aligned
 118          properly.
 119
 120          3) But wait!  Aren't we looking for C as well as zero?
 121          Good point.  So what we do is XOR LONGWORD with a longword,
 122          each of whose bytes is C.  This turns each byte that is C
 123          into a zero.  */
 124
 125       longword = *longword_ptr++;
 126
 127       /* Add MAGIC_BITS to LONGWORD.  */
 128       if ((((longword + magic_bits)
 129
 130             /* Set those bits that were unchanged by the addition.  */
 131             ^ ~longword)
 132
 133            /* Look at only the hole bits.  If any of the hole bits
 134               are unchanged, most likely one of the bytes was a
 135               zero.  */
 136            & ~magic_bits) != 0 ||
 137
 138           /* That caught zeroes.  Now test for C.  */
 139           ((((longword ^ charmask) + magic_bits) ^ ~(longword ^ charmask))
 140            & ~magic_bits) != 0)
 141         {
 142           /* Which of the bytes was C or zero?
 143              If none of them were, it was a misfire; continue the search.  */
 144
 145           const unsigned char *cp = (const unsigned char *) (longword_ptr - 1);
 146
 147           if (*cp == c || *cp == '\0')
 148             return (char *) cp;
 149           if (*++cp == c || *cp == '\0')
 150             return (char *) cp;
 151           if (*++cp == c || *cp == '\0')
 152             return (char *) cp;
 153           if (*++cp == c || *cp == '\0')
 154             return (char *) cp;
 155           if (sizeof (longword) > 4)
 156             {
 157               if (*++cp == c || *cp == '\0')
 158                 return (char *) cp;
 159               if (*++cp == c || *cp == '\0')
 160                 return (char *) cp;
 161               if (*++cp == c || *cp == '\0')
 162                 return (char *) cp;
 163               if (*++cp == c || *cp == '\0')
 164                 return (char *) cp;
 165             }
 166         }
 167     }
 168
 169   /* This should never happen.  */
 170   return NULL;
 171 }
 172
 173 weak_alias (__strchrnul, strchrnul)