usr/src/lib/libc/i386/gen/strlen.s

   1 /*
   2  * CDDL HEADER START
   3  *
   4  * The contents of this file are subject to the terms of the
   5  * Common Development and Distribution License (the "License").
   6  * You may not use this file except in compliance with the License.
   7  *
   8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
   9  * or http://www.opensolaris.org/os/licensing.
  10  * See the License for the specific language governing permissions
  11  * and limitations under the License.
  12  *
  13  * When distributing Covered Code, include this CDDL HEADER in each
  14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
  15  * If applicable, add the following below this CDDL HEADER, with the
  16  * fields enclosed by brackets "[]" replaced with your own identifying
  17  * information: Portions Copyright [yyyy] [name of copyright owner]
  18  *
  19  * CDDL HEADER END
  20  */
  21 /*
  22  * Copyright 2005 Sun Microsystems, Inc.  All rights reserved.
  23  * Use is subject to license terms.
  24  */
  25
  26         .file   "strlen.s"
  27
  28 /
  29 / strlen(s)
  30 /
  31 / Returns the number of non-NULL bytes in string argument.
  32 /
  33 /
  34 / Fast assembly language version of the following C-program strlen
  35 / which represents the `standard' for the C-library.
  36 /
  37 /       size_t
  38 /       strlen(const char *s)
  39 /       {
  40 /               const char      *s0 = s + 1;
  41 /
  42 /               while (*s++ != '\0')
  43 /                       ;
  44 /               return (s - s0);
  45 /       }
  46 /
  47 / In this assembly language version, the following expression is used
  48 / to check if a 32-bit word data contains a null byte or not:
  49 /       (((A & 0x7f7f7f7f) + 0x7f7f7f7f) | A) & 0x80808080
  50 / If the above expression geneates a value other than 0x80808080,
  51 / that means the 32-bit word data contains a null byte.
  52 /
  53
  54 #include "SYS.h"
  55
  56         ENTRY(strlen)
  57         mov     4(%esp), %edx           / src in %edx
  58         mov     %edx, %eax              / cpy src to %eax
  59
  60         and     $3, %edx                / is src aligned?
  61         jz      countbytes
  62                                         / work byte-wise until aligned
  63         cmpb    $0, (%eax)              / is *src == 0 ?
  64         jz      done
  65         inc     %eax                    / increment src
  66         cmp     $3, %edx                / if aligned, jump to word-wise check
  67         jz      countbytes
  68         cmpb    $0, (%eax)
  69         jz      done
  70         inc     %eax
  71         cmp     $2, %edx
  72         jz      countbytes
  73         cmpb    $0, (%eax)
  74         jz      done
  75         inc     %eax
  76
  77         .align    16
  78
  79 countbytes:
  80         mov     (%eax), %ecx            / load wrd
  81         add     $4, %eax                / increment src by 4 (bytes in word)
  82         lea     -0x01010101(%ecx), %edx / (wrd - 0x01010101)
  83         not     %ecx                    / ~wrd
  84         and     $0x80808080, %ecx       / ~wrd & 0x80808080
  85         and     %edx, %ecx              / (wrd - 0x01010101) & ~wrd & 0x80808080
  86         jz      countbytes              / if zero, no null byte found -- cont
  87
  88 has_zero_byte:
  89         bsfl    %ecx, %ecx              / find first set bit (null byte)
  90         shr     $3, %ecx                / switch bit position to byte posn
  91         lea     -4(%eax, %ecx, 1), %eax / undo pre-increment and count bytes
  92 done:
  93         sub     4(%esp), %eax           / return (src - old_src)
  94         ret
  95         SET_SIZE(strlen)