1 |
/* |
2 |
* Copyright (c) 2000-2007 Marc Alexander Lehmann <schmorp@schmorp.de> |
3 |
* |
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 |
* |
25 |
* Alternatively, the contents of this file may be used under the terms of |
26 |
* the GNU General Public License ("GPL") version 2 or any later version, |
27 |
* in which case the provisions of the GPL are applicable instead of |
28 |
* the above. If you wish to allow the use of your version of this file |
29 |
* only under the terms of the GPL and not to allow others to use your |
30 |
* version of this file under the BSD license, indicate your decision |
31 |
* by deleting the provisions above and replace them with the notice |
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 |
34 |
* either the BSD or the GPL. |
35 |
*/ |
36 |
|
37 |
#include "lzfP.h" |
38 |
|
39 |
#define HSIZE (1 << (HLOG)) |
40 |
|
41 |
/* |
42 |
* don't play with this unless you benchmark! |
43 |
* decompression is not dependent on the hash function |
44 |
* the hashing function might seem strange, just believe me |
45 |
* it works ;) |
46 |
*/ |
47 |
#ifndef FRST |
48 |
# define FRST(p) (((p[0]) << 8) | p[1]) |
49 |
# define NEXT(v,p) (((v) << 8) | p[2]) |
50 |
# define IDX(h) ((((h ^ (h << 5)) >> (3*8 - HLOG)) - h*5) & (HSIZE - 1)) |
51 |
/*# define IDX(h) ((ip[0] * 121 ^ ip[1] * 33 ^ ip[2] * 1) & (HSIZE-1))*/ |
52 |
#endif |
53 |
/* |
54 |
* IDX works because it is very similar to a multiplicative hash, e.g. |
55 |
* ((h * 57321 >> (3*8 - HLOG)) & (HSIZE - 1)) |
56 |
* the latter is also quite fast on newer CPUs, and compresses similarly. |
57 |
* |
58 |
* the next one is also quite good, albeit slow ;) |
59 |
* (int)(cos(h & 0xffffff) * 1e6) |
60 |
*/ |
61 |
|
62 |
#if 0 |
63 |
/* original lzv-like hash function, much worse and thus slower */ |
64 |
# define FRST(p) (p[0] << 5) ^ p[1] |
65 |
# define NEXT(v,p) ((v) << 5) ^ p[2] |
66 |
# define IDX(h) ((h) & (HSIZE - 1)) |
67 |
#endif |
68 |
|
69 |
#define MAX_LIT (1 << 5) |
70 |
#define MAX_OFF (1 << 13) |
71 |
#define MAX_REF ((1 << 8) + (1 << 3)) |
72 |
|
73 |
#if (__i386 || __amd64) && __GNUC__ >= 3 |
74 |
# define lzf_movsb(dst, src, len) \ |
75 |
asm ("rep movsb" \ |
76 |
: "=D" (dst), "=S" (src), "=c" (len) \ |
77 |
: "0" (dst), "1" (src), "2" (len)); |
78 |
#endif |
79 |
|
80 |
/* |
81 |
* compressed format |
82 |
* |
83 |
* 000LLLLL <L+1> ; literal |
84 |
* LLLooooo oooooooo ; backref L |
85 |
* 111ooooo LLLLLLLL oooooooo ; backref L+7 |
86 |
* |
87 |
*/ |
88 |
|
89 |
unsigned int |
90 |
lzf_compress (const void *const in_data, unsigned int in_len, |
91 |
void *out_data, unsigned int out_len |
92 |
#if LZF_STATE_ARG |
93 |
, LZF_STATE htab |
94 |
#endif |
95 |
) |
96 |
{ |
97 |
#if !LZF_STATE_ARG |
98 |
LZF_STATE htab; |
99 |
#endif |
100 |
const u8 **hslot; |
101 |
const u8 *ip = (const u8 *)in_data; |
102 |
u8 *op = (u8 *)out_data; |
103 |
const u8 *in_end = ip + in_len; |
104 |
u8 *out_end = op + out_len; |
105 |
const u8 *ref; |
106 |
|
107 |
unsigned int hval = FRST (ip); |
108 |
unsigned long off; |
109 |
int lit = 0; |
110 |
|
111 |
#if INIT_HTAB |
112 |
memset (htab, 0, sizeof (htab)); |
113 |
# if 0 |
114 |
for (hslot = htab; hslot < htab + HSIZE; hslot++) |
115 |
*hslot++ = ip; |
116 |
# endif |
117 |
#endif |
118 |
|
119 |
for (;;) |
120 |
{ |
121 |
if (ip < in_end - 2) |
122 |
{ |
123 |
hval = NEXT (hval, ip); |
124 |
hslot = htab + IDX (hval); |
125 |
ref = *hslot; *hslot = ip; |
126 |
|
127 |
if (1 |
128 |
#if INIT_HTAB && !USE_MEMCPY |
129 |
&& ref < ip /* the next test will actually take care of this, but this is faster */ |
130 |
#endif |
131 |
&& (off = ip - ref - 1) < MAX_OFF |
132 |
&& ip + 4 < in_end |
133 |
&& ref > (u8 *)in_data |
134 |
#if STRICT_ALIGN |
135 |
&& ref[0] == ip[0] |
136 |
&& ref[1] == ip[1] |
137 |
&& ref[2] == ip[2] |
138 |
#else |
139 |
&& *(u16 *)ref == *(u16 *)ip |
140 |
&& ref[2] == ip[2] |
141 |
#endif |
142 |
) |
143 |
{ |
144 |
/* match found at *ref++ */ |
145 |
unsigned int len = 2; |
146 |
unsigned int maxlen = in_end - ip - len; |
147 |
maxlen = maxlen > MAX_REF ? MAX_REF : maxlen; |
148 |
|
149 |
if (op + lit + 1 + 3 >= out_end) |
150 |
return 0; |
151 |
|
152 |
do |
153 |
len++; |
154 |
while (len < maxlen && ref[len] == ip[len]); |
155 |
|
156 |
if (lit) |
157 |
{ |
158 |
*op++ = lit - 1; |
159 |
lit = -lit; |
160 |
do |
161 |
*op++ = ip[lit]; |
162 |
while (++lit); |
163 |
} |
164 |
|
165 |
len -= 2; |
166 |
ip++; |
167 |
|
168 |
if (len < 7) |
169 |
{ |
170 |
*op++ = (off >> 8) + (len << 5); |
171 |
} |
172 |
else |
173 |
{ |
174 |
*op++ = (off >> 8) + ( 7 << 5); |
175 |
*op++ = len - 7; |
176 |
} |
177 |
|
178 |
*op++ = off; |
179 |
|
180 |
#if ULTRA_FAST || VERY_FAST |
181 |
ip += len; |
182 |
#if VERY_FAST && !ULTRA_FAST |
183 |
--ip; |
184 |
#endif |
185 |
hval = FRST (ip); |
186 |
|
187 |
hval = NEXT (hval, ip); |
188 |
htab[IDX (hval)] = ip; |
189 |
ip++; |
190 |
|
191 |
#if VERY_FAST && !ULTRA_FAST |
192 |
hval = NEXT (hval, ip); |
193 |
htab[IDX (hval)] = ip; |
194 |
ip++; |
195 |
#endif |
196 |
#else |
197 |
do |
198 |
{ |
199 |
hval = NEXT (hval, ip); |
200 |
htab[IDX (hval)] = ip; |
201 |
ip++; |
202 |
} |
203 |
while (len--); |
204 |
#endif |
205 |
continue; |
206 |
} |
207 |
} |
208 |
else if (ip == in_end) |
209 |
break; |
210 |
|
211 |
/* one more literal byte we must copy */ |
212 |
lit++; |
213 |
ip++; |
214 |
|
215 |
if (lit == MAX_LIT) |
216 |
{ |
217 |
if (op + 1 + MAX_LIT >= out_end) |
218 |
return 0; |
219 |
|
220 |
*op++ = MAX_LIT - 1; |
221 |
|
222 |
#ifdef lzf_movsb |
223 |
ip -= lit; |
224 |
lzf_movsb (op, ip, lit); |
225 |
#else |
226 |
lit = -lit; |
227 |
do |
228 |
*op++ = ip[lit]; |
229 |
while (++lit); |
230 |
#endif |
231 |
} |
232 |
} |
233 |
|
234 |
if (lit) |
235 |
{ |
236 |
if (op + lit + 1 >= out_end) |
237 |
return 0; |
238 |
|
239 |
*op++ = lit - 1; |
240 |
#ifdef lzf_movsb |
241 |
ip -= lit; |
242 |
lzf_movsb (op, ip, lit); |
243 |
#else |
244 |
lit = -lit; |
245 |
do |
246 |
*op++ = ip[lit]; |
247 |
while (++lit); |
248 |
#endif |
249 |
} |
250 |
|
251 |
return op - (u8 *) out_data; |
252 |
} |
253 |
|