1 |
root |
1.1 |
/* |
2 |
root |
1.15 |
* Copyright (c) 2000-2007 Marc Alexander Lehmann <schmorp@schmorp.de> |
3 |
root |
1.1 |
* |
4 |
|
|
* Redistribution and use in source and binary forms, with or without modifica- |
5 |
|
|
* tion, are permitted provided that the following conditions are met: |
6 |
|
|
* |
7 |
|
|
* 1. Redistributions of source code must retain the above copyright notice, |
8 |
|
|
* this list of conditions and the following disclaimer. |
9 |
|
|
* |
10 |
|
|
* 2. Redistributions in binary form must reproduce the above copyright |
11 |
|
|
* notice, this list of conditions and the following disclaimer in the |
12 |
|
|
* documentation and/or other materials provided with the distribution. |
13 |
|
|
* |
14 |
|
|
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED |
15 |
|
|
* WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MER- |
16 |
|
|
* CHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO |
17 |
|
|
* EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPE- |
18 |
|
|
* CIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, |
19 |
|
|
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; |
20 |
|
|
* OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, |
21 |
|
|
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTH- |
22 |
|
|
* ERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED |
23 |
|
|
* OF THE POSSIBILITY OF SUCH DAMAGE. |
24 |
pcg |
1.6 |
* |
25 |
|
|
* Alternatively, the contents of this file may be used under the terms of |
26 |
|
|
* the GNU General Public License version 2 (the "GPL"), in which case the |
27 |
|
|
* provisions of the GPL are applicable instead of the above. If you wish to |
28 |
|
|
* allow the use of your version of this file only under the terms of the |
29 |
|
|
* GPL and not to allow others to use your version of this file under the |
30 |
|
|
* BSD license, indicate your decision by deleting the provisions above and |
31 |
|
|
* replace them with the notice and other provisions required by the GPL. If |
32 |
|
|
* you do not delete the provisions above, a recipient may use your version |
33 |
|
|
* of this file under either the BSD or the GPL. |
34 |
root |
1.1 |
*/ |
35 |
|
|
|
36 |
|
|
#include "lzfP.h" |
37 |
|
|
|
38 |
|
|
#define HSIZE (1 << (HLOG)) |
39 |
|
|
|
40 |
|
|
/* |
41 |
|
|
* don't play with this unless you benchmark! |
42 |
|
|
* decompression is not dependent on the hash function |
43 |
|
|
* the hashing function might seem strange, just believe me |
44 |
|
|
* it works ;) |
45 |
|
|
*/ |
46 |
root |
1.10 |
#ifndef FRST |
47 |
|
|
# define FRST(p) (((p[0]) << 8) | p[1]) |
48 |
|
|
# define NEXT(v,p) (((v) << 8) | p[2]) |
49 |
|
|
# define IDX(h) ((((h ^ (h << 5)) >> (3*8 - HLOG)) - h*5) & (HSIZE - 1)) |
50 |
root |
1.13 |
/*# define IDX(h) ((ip[0] * 121 ^ ip[1] * 33 ^ ip[2] * 1) & (HSIZE-1))*/ |
51 |
root |
1.10 |
#endif |
52 |
root |
1.1 |
/* |
53 |
|
|
* IDX works because it is very similar to a multiplicative hash, e.g. |
54 |
root |
1.7 |
* ((h * 57321 >> (3*8 - HLOG)) & (HSIZE - 1)) |
55 |
root |
1.12 |
* the latter is also quite fast on newer CPUs, and compresses similarly. |
56 |
root |
1.7 |
* |
57 |
root |
1.1 |
* the next one is also quite good, albeit slow ;) |
58 |
|
|
* (int)(cos(h & 0xffffff) * 1e6) |
59 |
|
|
*/ |
60 |
|
|
|
61 |
|
|
#if 0 |
62 |
root |
1.7 |
/* original lzv-like hash function, much worse and thus slower */ |
63 |
root |
1.1 |
# define FRST(p) (p[0] << 5) ^ p[1] |
64 |
root |
1.2 |
# define NEXT(v,p) ((v) << 5) ^ p[2] |
65 |
|
|
# define IDX(h) ((h) & (HSIZE - 1)) |
66 |
root |
1.1 |
#endif |
67 |
|
|
|
68 |
|
|
#define MAX_LIT (1 << 5) |
69 |
|
|
#define MAX_OFF (1 << 13) |
70 |
root |
1.12 |
#define MAX_REF ((1 << 8) + (1 << 3)) |
71 |
root |
1.1 |
|
72 |
root |
1.14 |
#if (__i386 || __amd64) && __GNUC__ >= 3 |
73 |
|
|
# define lzf_movsb(dst, src, len) \ |
74 |
|
|
asm ("rep movsb" \ |
75 |
|
|
: "=D" (dst), "=S" (src), "=c" (len) \ |
76 |
|
|
: "0" (dst), "1" (src), "2" (len)); |
77 |
|
|
#endif |
78 |
|
|
|
79 |
root |
1.1 |
/* |
80 |
|
|
* compressed format |
81 |
|
|
* |
82 |
|
|
* 000LLLLL <L+1> ; literal |
83 |
root |
1.10 |
* LLLooooo oooooooo ; backref L |
84 |
|
|
* 111ooooo LLLLLLLL oooooooo ; backref L+7 |
85 |
root |
1.1 |
* |
86 |
|
|
*/ |
87 |
|
|
|
88 |
|
|
unsigned int |
89 |
|
|
lzf_compress (const void *const in_data, unsigned int in_len, |
90 |
pcg |
1.4 |
void *out_data, unsigned int out_len |
91 |
pcg |
1.5 |
#if LZF_STATE_ARG |
92 |
root |
1.11 |
, LZF_STATE htab |
93 |
pcg |
1.4 |
#endif |
94 |
|
|
) |
95 |
root |
1.1 |
{ |
96 |
pcg |
1.5 |
#if !LZF_STATE_ARG |
97 |
pcg |
1.4 |
LZF_STATE htab; |
98 |
|
|
#endif |
99 |
root |
1.1 |
const u8 **hslot; |
100 |
pcg |
1.4 |
const u8 *ip = (const u8 *)in_data; |
101 |
|
|
u8 *op = (u8 *)out_data; |
102 |
root |
1.1 |
const u8 *in_end = ip + in_len; |
103 |
|
|
u8 *out_end = op + out_len; |
104 |
|
|
const u8 *ref; |
105 |
|
|
|
106 |
|
|
unsigned int hval = FRST (ip); |
107 |
|
|
unsigned long off; |
108 |
|
|
int lit = 0; |
109 |
|
|
|
110 |
|
|
#if INIT_HTAB |
111 |
root |
1.14 |
memset (htab, 0, sizeof (htab)); |
112 |
|
|
# if 0 |
113 |
|
|
for (hslot = htab; hslot < htab + HSIZE; hslot++) |
114 |
|
|
*hslot++ = ip; |
115 |
root |
1.1 |
# endif |
116 |
|
|
#endif |
117 |
|
|
|
118 |
pcg |
1.5 |
for (;;) |
119 |
root |
1.1 |
{ |
120 |
pcg |
1.5 |
if (ip < in_end - 2) |
121 |
|
|
{ |
122 |
|
|
hval = NEXT (hval, ip); |
123 |
|
|
hslot = htab + IDX (hval); |
124 |
|
|
ref = *hslot; *hslot = ip; |
125 |
root |
1.1 |
|
126 |
pcg |
1.5 |
if (1 |
127 |
root |
1.1 |
#if INIT_HTAB && !USE_MEMCPY |
128 |
pcg |
1.5 |
&& ref < ip /* the next test will actually take care of this, but this is faster */ |
129 |
root |
1.1 |
#endif |
130 |
pcg |
1.5 |
&& (off = ip - ref - 1) < MAX_OFF |
131 |
|
|
&& ip + 4 < in_end |
132 |
|
|
&& ref > (u8 *)in_data |
133 |
root |
1.1 |
#if STRICT_ALIGN |
134 |
pcg |
1.5 |
&& ref[0] == ip[0] |
135 |
|
|
&& ref[1] == ip[1] |
136 |
|
|
&& ref[2] == ip[2] |
137 |
root |
1.1 |
#else |
138 |
pcg |
1.5 |
&& *(u16 *)ref == *(u16 *)ip |
139 |
|
|
&& ref[2] == ip[2] |
140 |
root |
1.1 |
#endif |
141 |
pcg |
1.5 |
) |
142 |
|
|
{ |
143 |
|
|
/* match found at *ref++ */ |
144 |
|
|
unsigned int len = 2; |
145 |
|
|
unsigned int maxlen = in_end - ip - len; |
146 |
|
|
maxlen = maxlen > MAX_REF ? MAX_REF : maxlen; |
147 |
|
|
|
148 |
root |
1.9 |
if (op + lit + 1 + 3 >= out_end) |
149 |
|
|
return 0; |
150 |
|
|
|
151 |
pcg |
1.5 |
do |
152 |
|
|
len++; |
153 |
|
|
while (len < maxlen && ref[len] == ip[len]); |
154 |
|
|
|
155 |
|
|
if (lit) |
156 |
|
|
{ |
157 |
|
|
*op++ = lit - 1; |
158 |
|
|
lit = -lit; |
159 |
|
|
do |
160 |
|
|
*op++ = ip[lit]; |
161 |
|
|
while (++lit); |
162 |
|
|
} |
163 |
|
|
|
164 |
|
|
len -= 2; |
165 |
|
|
ip++; |
166 |
|
|
|
167 |
|
|
if (len < 7) |
168 |
|
|
{ |
169 |
|
|
*op++ = (off >> 8) + (len << 5); |
170 |
|
|
} |
171 |
|
|
else |
172 |
|
|
{ |
173 |
|
|
*op++ = (off >> 8) + ( 7 << 5); |
174 |
|
|
*op++ = len - 7; |
175 |
|
|
} |
176 |
root |
1.1 |
|
177 |
pcg |
1.5 |
*op++ = off; |
178 |
root |
1.1 |
|
179 |
root |
1.9 |
#if ULTRA_FAST || VERY_FAST |
180 |
pcg |
1.5 |
ip += len; |
181 |
root |
1.9 |
#if VERY_FAST && !ULTRA_FAST |
182 |
|
|
--ip; |
183 |
|
|
#endif |
184 |
pcg |
1.5 |
hval = FRST (ip); |
185 |
root |
1.9 |
|
186 |
|
|
hval = NEXT (hval, ip); |
187 |
|
|
htab[IDX (hval)] = ip; |
188 |
|
|
ip++; |
189 |
|
|
|
190 |
|
|
#if VERY_FAST && !ULTRA_FAST |
191 |
pcg |
1.5 |
hval = NEXT (hval, ip); |
192 |
|
|
htab[IDX (hval)] = ip; |
193 |
|
|
ip++; |
194 |
root |
1.9 |
#endif |
195 |
root |
1.1 |
#else |
196 |
pcg |
1.5 |
do |
197 |
|
|
{ |
198 |
|
|
hval = NEXT (hval, ip); |
199 |
|
|
htab[IDX (hval)] = ip; |
200 |
|
|
ip++; |
201 |
|
|
} |
202 |
|
|
while (len--); |
203 |
|
|
#endif |
204 |
|
|
continue; |
205 |
|
|
} |
206 |
|
|
} |
207 |
|
|
else if (ip == in_end) |
208 |
|
|
break; |
209 |
|
|
|
210 |
|
|
/* one more literal byte we must copy */ |
211 |
|
|
lit++; |
212 |
|
|
ip++; |
213 |
|
|
|
214 |
|
|
if (lit == MAX_LIT) |
215 |
|
|
{ |
216 |
|
|
if (op + 1 + MAX_LIT >= out_end) |
217 |
|
|
return 0; |
218 |
root |
1.1 |
|
219 |
pcg |
1.5 |
*op++ = MAX_LIT - 1; |
220 |
root |
1.14 |
|
221 |
|
|
#ifdef lzf_movsb |
222 |
|
|
ip -= lit; |
223 |
|
|
lzf_movsb (op, ip, lit); |
224 |
root |
1.1 |
#else |
225 |
pcg |
1.5 |
lit = -lit; |
226 |
|
|
do |
227 |
|
|
*op++ = ip[lit]; |
228 |
|
|
while (++lit); |
229 |
root |
1.1 |
#endif |
230 |
pcg |
1.5 |
} |
231 |
root |
1.1 |
} |
232 |
|
|
|
233 |
|
|
if (lit) |
234 |
|
|
{ |
235 |
|
|
if (op + lit + 1 >= out_end) |
236 |
|
|
return 0; |
237 |
|
|
|
238 |
|
|
*op++ = lit - 1; |
239 |
root |
1.14 |
#ifdef lzf_movsb |
240 |
|
|
ip -= lit; |
241 |
|
|
lzf_movsb (op, ip, lit); |
242 |
|
|
#else |
243 |
root |
1.1 |
lit = -lit; |
244 |
|
|
do |
245 |
|
|
*op++ = ip[lit]; |
246 |
|
|
while (++lit); |
247 |
root |
1.14 |
#endif |
248 |
root |
1.1 |
} |
249 |
|
|
|
250 |
|
|
return op - (u8 *) out_data; |
251 |
|
|
} |
252 |
root |
1.14 |
|