diff options
Diffstat (limited to 'adler32.c')
-rw-r--r-- | adler32.c | 115 |
1 files changed, 84 insertions, 31 deletions
@@ -12,12 +12,13 @@ | |||
12 | #define NMAX 5552 | 12 | #define NMAX 5552 |
13 | /* NMAX is the largest n such that 255n(n+1)/2 + (n+1)(BASE-1) <= 2^32-1 */ | 13 | /* NMAX is the largest n such that 255n(n+1)/2 + (n+1)(BASE-1) <= 2^32-1 */ |
14 | 14 | ||
15 | #define DO1(buf,i) {s1 += buf[i]; s2 += s1;} | 15 | #define DO1(buf,i) {adler += (buf)[i]; sum2 += adler;} |
16 | #define DO2(buf,i) DO1(buf,i); DO1(buf,i+1); | 16 | #define DO2(buf,i) DO1(buf,i); DO1(buf,i+1); |
17 | #define DO4(buf,i) DO2(buf,i); DO2(buf,i+2); | 17 | #define DO4(buf,i) DO2(buf,i); DO2(buf,i+2); |
18 | #define DO8(buf,i) DO4(buf,i); DO4(buf,i+4); | 18 | #define DO8(buf,i) DO4(buf,i); DO4(buf,i+4); |
19 | #define DO16(buf) DO8(buf,0); DO8(buf,8); | 19 | #define DO16(buf) DO8(buf,0); DO8(buf,8); |
20 | 20 | ||
21 | /* use NO_DIVIDE if your processor does not do division in hardware */ | ||
21 | #ifdef NO_DIVIDE | 22 | #ifdef NO_DIVIDE |
22 | # define MOD(a) \ | 23 | # define MOD(a) \ |
23 | do { \ | 24 | do { \ |
@@ -39,8 +40,17 @@ | |||
39 | if (a >= (BASE << 1)) a -= (BASE << 1); \ | 40 | if (a >= (BASE << 1)) a -= (BASE << 1); \ |
40 | if (a >= BASE) a -= BASE; \ | 41 | if (a >= BASE) a -= BASE; \ |
41 | } while (0) | 42 | } while (0) |
43 | # define MOD4(a) \ | ||
44 | do { \ | ||
45 | if (a >= (BASE << 4)) a -= (BASE << 4); \ | ||
46 | if (a >= (BASE << 3)) a -= (BASE << 3); \ | ||
47 | if (a >= (BASE << 2)) a -= (BASE << 2); \ | ||
48 | if (a >= (BASE << 1)) a -= (BASE << 1); \ | ||
49 | if (a >= BASE) a -= BASE; \ | ||
50 | } while (0) | ||
42 | #else | 51 | #else |
43 | # define MOD(a) a %= BASE | 52 | # define MOD(a) a %= BASE |
53 | # define MOD4(a) a %= BASE | ||
44 | #endif | 54 | #endif |
45 | 55 | ||
46 | /* ========================================================================= */ | 56 | /* ========================================================================= */ |
@@ -49,48 +59,91 @@ uLong ZEXPORT adler32(adler, buf, len) | |||
49 | const Bytef *buf; | 59 | const Bytef *buf; |
50 | uInt len; | 60 | uInt len; |
51 | { | 61 | { |
52 | unsigned long s1 = adler & 0xffff; | 62 | unsigned long sum2; |
53 | unsigned long s2 = (adler >> 16) & 0xffff; | 63 | unsigned n; |
54 | int k; | 64 | |
65 | /* split Adler-32 into component sums */ | ||
66 | sum2 = (adler >> 16) & 0xffff; | ||
67 | adler &= 0xffff; | ||
68 | |||
69 | /* in case user likes doing a byte at a time, keep it fast */ | ||
70 | if (len == 1) { | ||
71 | adler += buf[0]; | ||
72 | if (adler >= BASE) | ||
73 | adler -= BASE; | ||
74 | sum2 += adler; | ||
75 | if (sum2 >= BASE) | ||
76 | sum2 -= BASE; | ||
77 | return adler | (sum2 << 16); | ||
78 | } | ||
55 | 79 | ||
56 | if (buf == Z_NULL) return 1L; | 80 | /* initial Adler-32 value (deferred check for len == 1 speed) */ |
81 | if (buf == Z_NULL) | ||
82 | return 1L; | ||
57 | 83 | ||
58 | while (len > 0) { | 84 | /* in case short lengths are provided, keep it somewhat fast */ |
59 | k = len < NMAX ? (int)len : NMAX; | 85 | if (len < 16) { |
60 | len -= k; | 86 | while (len--) { |
61 | while (k >= 16) { | 87 | adler += *buf++; |
88 | sum2 += adler; | ||
89 | } | ||
90 | if (adler >= BASE) | ||
91 | adler -= BASE; | ||
92 | MOD4(sum2); /* only added so many BASE's */ | ||
93 | return adler | (sum2 << 16); | ||
94 | } | ||
95 | |||
96 | /* do length NMAX blocks -- requires just one modulo operation */ | ||
97 | while (len >= NMAX) { | ||
98 | len -= NMAX; | ||
99 | n = NMAX / 16; /* NMAX is divisible by 16 */ | ||
100 | do { | ||
101 | DO16(buf); /* 16 sums unrolled */ | ||
102 | buf += 16; | ||
103 | } while (--n); | ||
104 | MOD(adler); | ||
105 | MOD(sum2); | ||
106 | } | ||
107 | |||
108 | /* do remaining bytes (less than NMAX, still just one modulo) */ | ||
109 | if (len) { /* avoid modulos if none remaining */ | ||
110 | while (len >= 16) { | ||
111 | len -= 16; | ||
62 | DO16(buf); | 112 | DO16(buf); |
63 | buf += 16; | 113 | buf += 16; |
64 | k -= 16; | ||
65 | } | 114 | } |
66 | if (k != 0) do { | 115 | while (len--) { |
67 | s1 += *buf++; | 116 | adler += *buf++; |
68 | s2 += s1; | 117 | sum2 += adler; |
69 | } while (--k); | 118 | } |
70 | MOD(s1); | 119 | MOD(adler); |
71 | MOD(s2); | 120 | MOD(sum2); |
72 | } | 121 | } |
73 | return (s2 << 16) | s1; | 122 | |
123 | /* return recombined sums */ | ||
124 | return adler | (sum2 << 16); | ||
74 | } | 125 | } |
75 | 126 | ||
76 | /* ========================================================================= */ | 127 | /* ========================================================================= */ |
77 | uLong ZEXPORT adler32_combine(adler1, adler2, len2) | 128 | uLong ZEXPORT adler32_combine(adler1, adler2, len2) |
78 | uLong adler1; | 129 | uLong adler1; |
79 | uLong adler2; | 130 | uLong adler2; |
80 | uLong len2; | 131 | z_off_t len2; |
81 | { | 132 | { |
82 | unsigned long s1; | 133 | unsigned long sum1; |
83 | unsigned long s2; | 134 | unsigned long sum2; |
135 | unsigned rem; | ||
84 | 136 | ||
85 | len2 %= BASE; | 137 | /* the derivation of this formula is left as an exercise for the reader */ |
86 | s1 = adler1 & 0xffff; | 138 | rem = (unsigned)(len2 % BASE); |
87 | s2 = len2 * s1; | 139 | sum1 = adler1 & 0xffff; |
88 | MOD(s2); | 140 | sum2 = rem * sum1; |
89 | s1 += (adler2 & 0xffff) + BASE - 1; | 141 | MOD(sum2); |
90 | s2 += ((adler1 >> 16) & 0xffff) + ((adler2 >> 16) & 0xffff) + BASE - len2; | 142 | sum1 += (adler2 & 0xffff) + BASE - 1; |
91 | if (s1 > BASE) s1 -= BASE; | 143 | sum2 += ((adler1 >> 16) & 0xffff) + ((adler2 >> 16) & 0xffff) + BASE - rem; |
92 | if (s1 > BASE) s1 -= BASE; | 144 | if (sum1 > BASE) sum1 -= BASE; |
93 | if (s2 > (BASE << 1)) s2 -= (BASE << 1); | 145 | if (sum1 > BASE) sum1 -= BASE; |
94 | if (s2 > BASE) s2 -= BASE; | 146 | if (sum2 > (BASE << 1)) sum2 -= (BASE << 1); |
95 | return (s2 << 16) | s1; | 147 | if (sum2 > BASE) sum2 -= BASE; |
148 | return sum1 | (sum2 << 16); | ||
96 | } | 149 | } |