libkern/zlib/adler32.c

   1 /*
   2  * Copyright (c) 2008 Apple Inc. All rights reserved.
   3  *
   4  * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
   5  *
   6  * This file contains Original Code and/or Modifications of Original Code
   7  * as defined in and that are subject to the Apple Public Source License
   8  * Version 2.0 (the 'License'). You may not use this file except in
   9  * compliance with the License. The rights granted to you under the License
  10  * may not be used to create, or enable the creation or redistribution of,
  11  * unlawful or unlicensed copies of an Apple operating system, or to
  12  * circumvent, violate, or enable the circumvention or violation of, any
  13  * terms of an Apple operating system software license agreement.
  14  *
  15  * Please obtain a copy of the License at
  16  * http://www.opensource.apple.com/apsl/ and read it before using this file.
  17  *
  18  * The Original Code and all software distributed under the License are
  19  * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  20  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  21  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  22  * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
  23  * Please see the License for the specific language governing rights and
  24  * limitations under the License.
  25  *
  26  * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
  27  */
  28 /* adler32.c -- compute the Adler-32 checksum of a data stream
  29  * Copyright (C) 1995-2004 Mark Adler
  30  * For conditions of distribution and use, see copyright notice in zlib.h
  31  */
  32
  33 /* @(#) $Id$ */
  34
  35 #include <stdint.h> // For uintptr_t.
  36
  37
  38 #define ZLIB_INTERNAL
  39 #if KERNEL
  40     #include <libkern/zlib.h>
  41 #else
  42     #include "zlib.h"
  43 #endif /* KERNEL */
  44
  45 #if defined _ARM_ARCH_6
  46         extern uLong adler32_vec(uLong adler, uLong sum2, const Bytef *buf, uInt len);
  47 #endif
  48
  49 #define BASE 65521UL    /* largest prime smaller than 65536 */
  50 #define NMAX 5552
  51 /* NMAX is the largest n such that 255n(n+1)/2 + (n+1)(BASE-1) <= 2^32-1 */
  52
  53 #define DO1(buf,i)  {adler += (buf)[i]; sum2 += adler;}
  54 #define DO2(buf,i)  DO1(buf,i); DO1(buf,i+1);
  55 #define DO4(buf,i)  DO2(buf,i); DO2(buf,i+2);
  56 #define DO8(buf,i)  DO4(buf,i); DO4(buf,i+4);
  57 #define DO16(buf)   DO8(buf,0); DO8(buf,8);
  58
  59 /* use NO_DIVIDE if your processor does not do division in hardware */
  60 #ifdef NO_DIVIDE
  61 #  define MOD(a) \
  62     do { \
  63         if (a >= (BASE << 16)) a -= (BASE << 16); \
  64         if (a >= (BASE << 15)) a -= (BASE << 15); \
  65         if (a >= (BASE << 14)) a -= (BASE << 14); \
  66         if (a >= (BASE << 13)) a -= (BASE << 13); \
  67         if (a >= (BASE << 12)) a -= (BASE << 12); \
  68         if (a >= (BASE << 11)) a -= (BASE << 11); \
  69         if (a >= (BASE << 10)) a -= (BASE << 10); \
  70         if (a >= (BASE << 9)) a -= (BASE << 9); \
  71         if (a >= (BASE << 8)) a -= (BASE << 8); \
  72         if (a >= (BASE << 7)) a -= (BASE << 7); \
  73         if (a >= (BASE << 6)) a -= (BASE << 6); \
  74         if (a >= (BASE << 5)) a -= (BASE << 5); \
  75         if (a >= (BASE << 4)) a -= (BASE << 4); \
  76         if (a >= (BASE << 3)) a -= (BASE << 3); \
  77         if (a >= (BASE << 2)) a -= (BASE << 2); \
  78         if (a >= (BASE << 1)) a -= (BASE << 1); \
  79         if (a >= BASE) a -= BASE; \
  80     } while (0)
  81 #  define MOD4(a) \
  82     do { \
  83         if (a >= (BASE << 4)) a -= (BASE << 4); \
  84         if (a >= (BASE << 3)) a -= (BASE << 3); \
  85         if (a >= (BASE << 2)) a -= (BASE << 2); \
  86         if (a >= (BASE << 1)) a -= (BASE << 1); \
  87         if (a >= BASE) a -= BASE; \
  88     } while (0)
  89 #else
  90 #  define MOD(a) a %= BASE
  91 #  define MOD4(a) a %= BASE
  92 #endif
  93
  94 /* ========================================================================= */
  95 uLong ZEXPORT adler32(adler, buf, len)
  96     uLong adler;
  97     const Bytef *buf;
  98     uInt len;
  99 {
 100     unsigned long sum2;
 101 #if !defined _ARM_ARCH_6
 102     unsigned n;
 103 #endif
 104
 105     /* split Adler-32 into component sums */
 106     sum2 = (adler >> 16) & 0xffff;
 107     adler &= 0xffff;
 108
 109     /* in case user likes doing a byte at a time, keep it fast */
 110     if (len == 1) {
 111         adler += buf[0];
 112         if (adler >= BASE)
 113             adler -= BASE;
 114         sum2 += adler;
 115         if (sum2 >= BASE)
 116             sum2 -= BASE;
 117         return adler | (sum2 << 16);
 118     }
 119
 120     /* initial Adler-32 value (deferred check for len == 1 speed) */
 121     if (buf == Z_NULL)
 122         return 1L;
 123
 124     /* in case short lengths are provided, keep it somewhat fast */
 125     if (len < 16) {
 126         while (len--) {
 127             adler += *buf++;
 128             sum2 += adler;
 129         }
 130         if (adler >= BASE)
 131             adler -= BASE;
 132         MOD4(sum2);             /* only added so many BASE's */
 133         return adler | (sum2 << 16);
 134     }
 135
 136 #if defined _ARM_ARCH_6
 137     /* align buf to 16-byte boundary */
 138     while (((uintptr_t)buf)&15) { /* not on a 16-byte boundary */
 139         len--;
 140         adler += *buf++;
 141         sum2 += adler;
 142         if (adler >= BASE) adler -= BASE;
 143         MOD4(sum2);             /* only added so many BASE's */
 144     }
 145
 146     return adler32_vec(adler, sum2, buf, len);      // armv7 neon vectorized implementation
 147
 148 #else   //  _ARM_ARCH_6
 149
 150     /* do length NMAX blocks -- requires just one modulo operation */
 151     while (len >= NMAX) {
 152         len -= NMAX;
 153         n = NMAX / 16;          /* NMAX is divisible by 16 */
 154         do {
 155             DO16(buf);          /* 16 sums unrolled */
 156             buf += 16;
 157         } while (--n);
 158         MOD(adler);
 159         MOD(sum2);
 160     }
 161
 162     /* do remaining bytes (less than NMAX, still just one modulo) */
 163     if (len) {                  /* avoid modulos if none remaining */
 164         while (len >= 16) {
 165             len -= 16;
 166             DO16(buf);
 167             buf += 16;
 168         }
 169         while (len--) {
 170             adler += *buf++;
 171             sum2 += adler;
 172         }
 173         MOD(adler);
 174         MOD(sum2);
 175     }
 176
 177     /* return recombined sums */
 178     return adler | (sum2 << 16);
 179
 180 #endif  // _ARM_ARCH_6
 181 }
 182
 183 /* ========================================================================= */
 184 uLong ZEXPORT adler32_combine(adler1, adler2, len2)
 185     uLong adler1;
 186     uLong adler2;
 187     z_off_t len2;
 188 {
 189     unsigned long sum1;
 190     unsigned long sum2;
 191     unsigned rem;
 192
 193     /* the derivation of this formula is left as an exercise for the reader */
 194     rem = (unsigned)(len2 % BASE);
 195     sum1 = adler1 & 0xffff;
 196     sum2 = rem * sum1;
 197     MOD(sum2);
 198     sum1 += (adler2 & 0xffff) + BASE - 1;
 199     sum2 += ((adler1 >> 16) & 0xffff) + ((adler2 >> 16) & 0xffff) + BASE - rem;
 200     if (sum1 > BASE) sum1 -= BASE;
 201     if (sum1 > BASE) sum1 -= BASE;
 202     if (sum2 > (BASE << 1)) sum2 -= (BASE << 1);
 203     if (sum2 > BASE) sum2 -= BASE;
 204     return sum1 | (sum2 << 16);
 205 }