string/strchrnul.c

   1 /* Copyright (C) 1991-2019 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 (const char *s, int c_in)
  37 {
  38   const unsigned char *char_ptr;
  39   const unsigned long int *longword_ptr;
  40   unsigned long int longword, magic_bits, charmask;
  41   unsigned char c;
  42
  43   c = (unsigned char) c_in;
  44
  45   /* Handle the first few characters by reading one character at a time.
  46      Do this until CHAR_PTR is aligned on a longword boundary.  */
  47   for (char_ptr = (const unsigned char *) s;
  48        ((unsigned long int) char_ptr & (sizeof (longword) - 1)) != 0;
  49        ++char_ptr)
  50     if (*char_ptr == c || *char_ptr == '\0')
  51       return (void *) char_ptr;
  52
  53   /* All these elucidatory comments refer to 4-byte longwords,
  54      but the theory applies equally well to 8-byte longwords.  */
  55
  56   longword_ptr = (unsigned long int *) char_ptr;
  57
  58   /* Bits 31, 24, 16, and 8 of this number are zero.  Call these bits
  59      the "holes."  Note that there is a hole just to the left of
  60      each byte, with an extra at the end:
  61
  62      bits:  01111110 11111110 11111110 11111111
  63      bytes: AAAAAAAA BBBBBBBB CCCCCCCC DDDDDDDD
  64
  65      The 1-bits make sure that carries propagate to the next 0-bit.
  66      The 0-bits provide holes for carries to fall into.  */
  67   magic_bits = -1;
  68   magic_bits = magic_bits / 0xff * 0xfe << 1 >> 1 | 1;
  69
  70   /* Set up a longword, each of whose bytes is C.  */
  71   charmask = c | (c << 8);
  72   charmask |= charmask << 16;
  73   if (sizeof (longword) > 4)
  74     /* Do the shift in two steps to avoid a warning if long has 32 bits.  */
  75     charmask |= (charmask << 16) << 16;
  76   if (sizeof (longword) > 8)
  77     abort ();
  78
  79   /* Instead of the traditional loop which tests each character,
  80      we will test a longword at a time.  The tricky part is testing
  81      if *any of the four* bytes in the longword in question are zero.  */
  82   for (;;)
  83     {
  84       /* We tentatively exit the loop if adding MAGIC_BITS to
  85          LONGWORD fails to change any of the hole bits of LONGWORD.
  86
  87          1) Is this safe?  Will it catch all the zero bytes?
  88          Suppose there is a byte with all zeros.  Any carry bits
  89          propagating from its left will fall into the hole at its
  90          least significant bit and stop.  Since there will be no
  91          carry from its most significant bit, the LSB of the
  92          byte to the left will be unchanged, and the zero will be
  93          detected.
  94
  95          2) Is this worthwhile?  Will it ignore everything except
  96          zero bytes?  Suppose every byte of LONGWORD has a bit set
  97          somewhere.  There will be a carry into bit 8.  If bit 8
  98          is set, this will carry into bit 16.  If bit 8 is clear,
  99          one of bits 9-15 must be set, so there will be a carry
 100          into bit 16.  Similarly, there will be a carry into bit
 101          24.  If one of bits 24-30 is set, there will be a carry
 102          into bit 31, so all of the hole bits will be changed.
 103
 104          The one misfire occurs when bits 24-30 are clear and bit
 105          31 is set; in this case, the hole at bit 31 is not
 106          changed.  If we had access to the processor carry flag,
 107          we could close this loophole by putting the fourth hole
 108          at bit 32!
 109
 110          So it ignores everything except 128's, when they're aligned
 111          properly.
 112
 113          3) But wait!  Aren't we looking for C as well as zero?
 114          Good point.  So what we do is XOR LONGWORD with a longword,
 115          each of whose bytes is C.  This turns each byte that is C
 116          into a zero.  */
 117
 118       longword = *longword_ptr++;
 119
 120       /* Add MAGIC_BITS to LONGWORD.  */
 121       if ((((longword + magic_bits)
 122
 123             /* Set those bits that were unchanged by the addition.  */
 124             ^ ~longword)
 125
 126            /* Look at only the hole bits.  If any of the hole bits
 127               are unchanged, most likely one of the bytes was a
 128               zero.  */
 129            & ~magic_bits) != 0
 130
 131           /* That caught zeroes.  Now test for C.  */
 132           || ((((longword ^ charmask) + magic_bits) ^ ~(longword ^ charmask))
 133               & ~magic_bits) != 0)
 134         {
 135           /* Which of the bytes was C or zero?
 136              If none of them were, it was a misfire; continue the search.  */
 137
 138           const unsigned char *cp = (const unsigned char *) (longword_ptr - 1);
 139
 140           if (*cp == c || *cp == '\0')
 141             return (char *) cp;
 142           if (*++cp == c || *cp == '\0')
 143             return (char *) cp;
 144           if (*++cp == c || *cp == '\0')
 145             return (char *) cp;
 146           if (*++cp == c || *cp == '\0')
 147             return (char *) cp;
 148           if (sizeof (longword) > 4)
 149             {
 150               if (*++cp == c || *cp == '\0')
 151                 return (char *) cp;
 152               if (*++cp == c || *cp == '\0')
 153                 return (char *) cp;
 154               if (*++cp == c || *cp == '\0')
 155                 return (char *) cp;
 156               if (*++cp == c || *cp == '\0')
 157                 return (char *) cp;
 158             }
 159         }
 160     }
 161
 162   /* This should never happen.  */
 163   return NULL;
 164 }
 165
 166 weak_alias (__strchrnul, strchrnul)