ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/liblzf/lzf_c.c
(Generate patch)

Comparing liblzf/lzf_c.c (file contents):
Revision 1.14 by root, Thu Jun 21 22:11:34 2007 UTC vs.
Revision 1.22 by root, Tue Nov 13 10:41:52 2007 UTC

1/* 1/*
2 * Copyright (c) 2000-2005 Marc Alexander Lehmann <schmorp@schmorp.de> 2 * Copyright (c) 2000-2007 Marc Alexander Lehmann <schmorp@schmorp.de>
3 * 3 *
4 * Redistribution and use in source and binary forms, with or without modifica- 4 * Redistribution and use in source and binary forms, with or without modifica-
5 * tion, are permitted provided that the following conditions are met: 5 * tion, are permitted provided that the following conditions are met:
6 * 6 *
7 * 1. Redistributions of source code must retain the above copyright notice, 7 * 1. Redistributions of source code must retain the above copyright notice,
8 * this list of conditions and the following disclaimer. 8 * this list of conditions and the following disclaimer.
9 * 9 *
10 * 2. Redistributions in binary form must reproduce the above copyright 10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the 11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution. 12 * documentation and/or other materials provided with the distribution.
13 *
14 * 3. The name of the author may not be used to endorse or promote products
15 * derived from this software without specific prior written permission.
16 * 13 *
17 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED 14 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED
18 * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MER- 15 * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MER-
19 * CHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO 16 * CHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
20 * EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPE- 17 * EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPE-
… …
24 * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTH- 21 * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTH-
25 * ERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED 22 * ERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
26 * OF THE POSSIBILITY OF SUCH DAMAGE. 23 * OF THE POSSIBILITY OF SUCH DAMAGE.
27 * 24 *
28 * Alternatively, the contents of this file may be used under the terms of 25 * Alternatively, the contents of this file may be used under the terms of
29 * the GNU General Public License version 2 (the "GPL"), in which case the 26 * the GNU General Public License ("GPL") version 2 or any later version,
30 * provisions of the GPL are applicable instead of the above. If you wish to 27 * in which case the provisions of the GPL are applicable instead of
31 * allow the use of your version of this file only under the terms of the 28 * the above. If you wish to allow the use of your version of this file
32 * GPL and not to allow others to use your version of this file under the 29 * only under the terms of the GPL and not to allow others to use your
33 * BSD license, indicate your decision by deleting the provisions above and 30 * version of this file under the BSD license, indicate your decision
34 * replace them with the notice and other provisions required by the GPL. If 31 * by deleting the provisions above and replace them with the notice
35 * you do not delete the provisions above, a recipient may use your version 32 * and other provisions required by the GPL. If you do not delete the
33 * provisions above, a recipient may use your version of this file under
36 * of this file under either the BSD or the GPL. 34 * either the BSD or the GPL.
37 */ 35 */
38 36
39#include "lzfP.h" 37#include "lzfP.h"
40 38
41#define HSIZE (1 << (HLOG)) 39#define HSIZE (1 << (HLOG))
… …
47 * it works ;) 45 * it works ;)
48 */ 46 */
49#ifndef FRST 47#ifndef FRST
50# define FRST(p) (((p[0]) << 8) | p[1]) 48# define FRST(p) (((p[0]) << 8) | p[1])
51# define NEXT(v,p) (((v) << 8) | p[2]) 49# define NEXT(v,p) (((v) << 8) | p[2])
50# if ULTRA_FAST
51# define IDX(h) (((h >> (3*8 - HLOG)) - h ) & (HSIZE - 1))
52# elif VERY_FAST
53# define IDX(h) (((h >> (3*8 - HLOG)) - h*5) & (HSIZE - 1))
54# else
52# define IDX(h) ((((h ^ (h << 5)) >> (3*8 - HLOG)) - h*5) & (HSIZE - 1)) 55# define IDX(h) ((((h ^ (h << 5)) >> (3*8 - HLOG)) - h*5) & (HSIZE - 1))
56# endif
53/*# define IDX(h) ((ip[0] * 121 ^ ip[1] * 33 ^ ip[2] * 1) & (HSIZE-1))*/ 57/*# define IDX(h) ((ip[0] * 121 ^ ip[1] * 33 ^ ip[2] * 1) & (HSIZE-1))*/
54#endif 58#endif
55/* 59/*
56 * IDX works because it is very similar to a multiplicative hash, e.g. 60 * IDX works because it is very similar to a multiplicative hash, e.g.
57 * ((h * 57321 >> (3*8 - HLOG)) & (HSIZE - 1)) 61 * ((h * 57321 >> (3*8 - HLOG)) & (HSIZE - 1))
… …
70 74
71#define MAX_LIT (1 << 5) 75#define MAX_LIT (1 << 5)
72#define MAX_OFF (1 << 13) 76#define MAX_OFF (1 << 13)
73#define MAX_REF ((1 << 8) + (1 << 3)) 77#define MAX_REF ((1 << 8) + (1 << 3))
74 78
75#if (__i386 || __amd64) && __GNUC__ >= 3 79#if __GNUC__ >= 3
76# define lzf_movsb(dst, src, len) \ 80# define expect(expr,value) __builtin_expect ((expr),(value))
77 asm ("rep movsb" \ 81# define inline inline
78 : "=D" (dst), "=S" (src), "=c" (len) \ 82#else
79 : "0" (dst), "1" (src), "2" (len)); 83# define expect(expr,value) (expr)
84# define inline static
80#endif 85#endif
86
87#define expect_false(expr) expect ((expr) != 0, 0)
88#define expect_true(expr) expect ((expr) != 0, 1)
81 89
82/* 90/*
83 * compressed format 91 * compressed format
84 * 92 *
85 * 000LLLLL <L+1> ; literal 93 * 000LLLLL <L+1> ; literal
… …
104 u8 *op = (u8 *)out_data; 112 u8 *op = (u8 *)out_data;
105 const u8 *in_end = ip + in_len; 113 const u8 *in_end = ip + in_len;
106 u8 *out_end = op + out_len; 114 u8 *out_end = op + out_len;
107 const u8 *ref; 115 const u8 *ref;
108 116
109 unsigned int hval = FRST (ip); 117 unsigned int hval;
110 unsigned long off; 118 unsigned long off;
111 int lit = 0; 119 int lit;
120
121 if (!in_len || !out_len)
122 return 0;
112 123
113#if INIT_HTAB 124#if INIT_HTAB
114 memset (htab, 0, sizeof (htab)); 125 memset (htab, 0, sizeof (htab));
115# if 0 126# if 0
116 for (hslot = htab; hslot < htab + HSIZE; hslot++) 127 for (hslot = htab; hslot < htab + HSIZE; hslot++)
117 *hslot++ = ip; 128 *hslot++ = ip;
118# endif 129# endif
119#endif 130#endif
120 131
121 for (;;) 132 lit = 0; op++; /* start run */
133
134 hval = FRST (ip);
135 while (ip < in_end - 2)
122 { 136 {
123 if (ip < in_end - 2) 137 hval = NEXT (hval, ip);
138 hslot = htab + IDX (hval);
139 ref = *hslot; *hslot = ip;
140
141 if (1
142#if INIT_HTAB && !USE_MEMCPY
143 && ref < ip /* the next test will actually take care of this, but this is faster */
144#endif
145 && (off = ip - ref - 1) < MAX_OFF
146 && ip + 4 < in_end
147 && ref > (u8 *)in_data
148#if STRICT_ALIGN
149 && ref[0] == ip[0]
150 && ref[1] == ip[1]
151 && ref[2] == ip[2]
152#else
153 && *(u16 *)ref == *(u16 *)ip
154 && ref[2] == ip[2]
155#endif
156 )
124 { 157 {
125 hval = NEXT (hval, ip);
126 hslot = htab + IDX (hval);
127 ref = *hslot; *hslot = ip;
128
129 if (1
130#if INIT_HTAB && !USE_MEMCPY
131 && ref < ip /* the next test will actually take care of this, but this is faster */
132#endif
133 && (off = ip - ref - 1) < MAX_OFF
134 && ip + 4 < in_end
135 && ref > (u8 *)in_data
136#if STRICT_ALIGN
137 && ref[0] == ip[0]
138 && ref[1] == ip[1]
139 && ref[2] == ip[2]
140#else
141 && *(u16 *)ref == *(u16 *)ip
142 && ref[2] == ip[2]
143#endif
144 )
145 {
146 /* match found at *ref++ */ 158 /* match found at *ref++ */
147 unsigned int len = 2; 159 unsigned int len = 2;
148 unsigned int maxlen = in_end - ip - len; 160 unsigned int maxlen = in_end - ip - len;
149 maxlen = maxlen > MAX_REF ? MAX_REF : maxlen; 161 maxlen = maxlen > MAX_REF ? MAX_REF : maxlen;
150 162
151 if (op + lit + 1 + 3 >= out_end) 163 if (expect_false (op + 1 + 3 >= out_end))
152 return 0; 164 return 0;
165
166 op [- lit - 1] = lit - 1; /* stop run */
167 op -= !lit; /* undo run if length is zero */
168
169 for (;;)
170 {
171 if (expect_true (maxlen > 16))
172 {
173 len++; if (ref [len] != ip [len]) break;
174 len++; if (ref [len] != ip [len]) break;
175 len++; if (ref [len] != ip [len]) break;
176 len++; if (ref [len] != ip [len]) break;
177
178 len++; if (ref [len] != ip [len]) break;
179 len++; if (ref [len] != ip [len]) break;
180 len++; if (ref [len] != ip [len]) break;
181 len++; if (ref [len] != ip [len]) break;
182
183 len++; if (ref [len] != ip [len]) break;
184 len++; if (ref [len] != ip [len]) break;
185 len++; if (ref [len] != ip [len]) break;
186 len++; if (ref [len] != ip [len]) break;
187
188 len++; if (ref [len] != ip [len]) break;
189 len++; if (ref [len] != ip [len]) break;
190 len++; if (ref [len] != ip [len]) break;
191 len++; if (ref [len] != ip [len]) break;
192 }
153 193
154 do 194 do
155 len++; 195 len++;
156 while (len < maxlen && ref[len] == ip[len]); 196 while (len < maxlen && ref[len] == ip[len]);
157 197
158 if (lit)
159 { 198 break;
160 *op++ = lit - 1;
161 lit = -lit;
162 do
163 *op++ = ip[lit];
164 while (++lit);
165 } 199 }
166 200
167 len -= 2; 201 len -= 2;
168 ip++; 202 ip++;
169 203
170 if (len < 7) 204 if (len < 7)
171 { 205 {
172 *op++ = (off >> 8) + (len << 5); 206 *op++ = (off >> 8) + (len << 5);
173 } 207 }
174 else 208 else
175 { 209 {
176 *op++ = (off >> 8) + ( 7 << 5); 210 *op++ = (off >> 8) + ( 7 << 5);
177 *op++ = len - 7; 211 *op++ = len - 7;
178 } 212 }
179 213
180 *op++ = off; 214 *op++ = off;
181 215
182#if ULTRA_FAST || VERY_FAST 216#if ULTRA_FAST || VERY_FAST
183 ip += len; 217 ip += len;
184#if VERY_FAST && !ULTRA_FAST 218#if VERY_FAST && !ULTRA_FAST
185 --ip; 219 --ip;
186#endif 220#endif
187 hval = FRST (ip); 221 hval = FRST (ip);
188 222
223 hval = NEXT (hval, ip);
224 htab[IDX (hval)] = ip;
225 ip++;
226
227#if VERY_FAST && !ULTRA_FAST
228 hval = NEXT (hval, ip);
229 htab[IDX (hval)] = ip;
230 ip++;
231#endif
232#else
233 do
234 {
189 hval = NEXT (hval, ip); 235 hval = NEXT (hval, ip);
190 htab[IDX (hval)] = ip; 236 htab[IDX (hval)] = ip;
191 ip++; 237 ip++;
192
193#if VERY_FAST && !ULTRA_FAST
194 hval = NEXT (hval, ip);
195 htab[IDX (hval)] = ip;
196 ip++;
197#endif
198#else
199 do
200 {
201 hval = NEXT (hval, ip);
202 htab[IDX (hval)] = ip;
203 ip++;
204 } 238 }
205 while (len--); 239 while (len--);
206#endif 240#endif
241 lit = 0; op++; /* start run */
207 continue; 242 continue;
208 }
209 } 243 }
210 else if (ip == in_end)
211 break;
212 244
213 /* one more literal byte we must copy */ 245 /* one more literal byte we must copy */
246
247 if (expect_false (op >= out_end))
248 return 0;
249
214 lit++; 250 lit++;
215 ip++; 251 *op++ = *ip++;
216 252
217 if (lit == MAX_LIT) 253 if (expect_false (lit == MAX_LIT))
218 { 254 {
219 if (op + 1 + MAX_LIT >= out_end) 255 op [- lit - 1] = lit - 1; /* stop run */
220 return 0; 256 lit = 0; op++; /* start run */
221
222 *op++ = MAX_LIT - 1;
223
224#ifdef lzf_movsb
225 ip -= lit;
226 lzf_movsb (op, ip, lit);
227#else
228 lit = -lit;
229 do
230 *op++ = ip[lit];
231 while (++lit);
232#endif
233 } 257 }
234 } 258 }
235 259
236 if (lit) 260 if (op + lit + 2 >= out_end)
261 return 0;
262
263 while (ip < in_end)
237 { 264 {
238 if (op + lit + 1 >= out_end) 265 lit++;
239 return 0;
240
241 *op++ = lit - 1; 266 *op++ = *ip++;
242#ifdef lzf_movsb
243 ip -= lit;
244 lzf_movsb (op, ip, lit);
245#else
246 lit = -lit;
247 do
248 *op++ = ip[lit];
249 while (++lit);
250#endif
251 } 267 }
252 268
269 op [- lit - 1] = lit - 1;
270
253 return op - (u8 *) out_data; 271 return op - (u8 *)out_data;
254} 272}
255 273

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines