xref: /illumos-gate/usr/src/uts/common/os/compress.c (revision 581cede61ac9c14d8d4ea452562a567189eead78)
1 /*
2  * CDDL HEADER START
3  *
4  * The contents of this file are subject to the terms of the
5  * Common Development and Distribution License, Version 1.0 only
6  * (the "License").  You may not use this file except in compliance
7  * with the License.
8  *
9  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
10  * or http://www.opensolaris.org/os/licensing.
11  * See the License for the specific language governing permissions
12  * and limitations under the License.
13  *
14  * When distributing Covered Code, include this CDDL HEADER in each
15  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
16  * If applicable, add the following below this CDDL HEADER, with the
17  * fields enclosed by brackets "[]" replaced with your own identifying
18  * information: Portions Copyright [yyyy] [name of copyright owner]
19  *
20  * CDDL HEADER END
21  */
22 /*
23  * Copyright (c) 1998 by Sun Microsystems, Inc.
24  * All rights reserved.
25  */
26 
27 #pragma ident	"%Z%%M%	%I%	%E% SMI"
28 
29 /*
30  * NOTE: this file is compiled into the kernel, cprboot, and savecore.
31  * Therefore it must compile in kernel, boot, and userland source context;
32  * so if you ever change this code, avoid references to external symbols.
33  *
34  * This compression algorithm is a derivative of LZRW1, which I'll call
35  * LZJB in the classic LZ* spirit.  All LZ* (Lempel-Ziv) algorithms are
36  * based on the same basic principle: when a "phrase" (sequences of bytes)
37  * is repeated in a data stream, we can save space by storing a reference to
38  * the previous instance of that phrase (a "copy item") rather than storing
39  * the phrase itself (a "literal item").  The compressor remembers phrases
40  * in a simple hash table (the "Lempel history") that maps three-character
41  * sequences (the minimum match) to the addresses where they were last seen.
42  *
43  * A copy item must encode both the length and the location of the matching
44  * phrase so that decompress() can reconstruct the original data stream.
45  * For example, here's how we'd encode "yadda yadda yadda, blah blah blah"
46  * (with "_" replacing spaces for readability):
47  *
48  * Original:
49  *
50  * y a d d a _ y a d d a _ y a d d a , _ b l a h _ b l a h _ b l a h
51  *
52  * Compressed:
53  *
54  * y a d d a _ 6 11 , _ b l a h 5 10
55  *
56  * In the compressed output, the "6 11" simply means "to get the original
57  * data, execute memmove(ptr, ptr - 6, 11)".  Note that in this example,
58  * the match at "6 11" actually extends beyond the current location and
59  * overlaps it.  That's OK; like memmove(), decompress() handles overlap.
60  *
61  * There's still one more thing decompress() needs to know, which is how to
62  * distinguish literal items from copy items.  We encode this information
63  * in an 8-bit bitmap that precedes each 8 items of output; if the Nth bit
64  * is set, then the Nth item is a copy item.  Thus the full encoding for
65  * the example above would be:
66  *
67  * 0x40 y a d d a _ 6 11 , 0x20 _ b l a h 5 10
68  *
69  * Finally, the "6 11" isn't really encoded as the two byte values 6 and 11
70  * in the output stream because, empirically, we get better compression by
71  * dedicating more bits to offset, fewer to match length.  LZJB uses 6 bits
72  * to encode the match length, 10 bits to encode the offset.  Since copy-item
73  * encoding consumes 2 bytes, we don't generate copy items unless the match
74  * length is at least 3; therefore, we can store (length - 3) in the 6-bit
75  * match length field, which extends the maximum match from 63 to 66 bytes.
76  * Thus the 2-byte encoding for a copy item is as follows:
77  *
78  *	byte[0] = ((length - 3) << 2) | (offset >> 8);
79  *	byte[1] = (uint8_t)offset;
80  *
81  * In our example above, an offset of 6 with length 11 would be encoded as:
82  *
83  *	byte[0] = ((11 - 3) << 2) | (6 >> 8) = 0x20
84  *	byte[1] = (uint8_t)6 = 0x6
85  *
86  * Similarly, an offset of 5 with length 10 would be encoded as:
87  *
88  *	byte[0] = ((10 - 3) << 2) | (5 >> 8) = 0x1c
89  *	byte[1] = (uint8_t)5 = 0x5
90  *
91  * Putting it all together, the actual LZJB output for our example is:
92  *
93  * 0x40 y a d d a _ 0x2006 , 0x20 _ b l a h 0x1c05
94  *
95  * The main differences between LZRW1 and LZJB are as follows:
96  *
97  * (1) LZRW1 is sloppy about buffer overruns.  LZJB never reads past the
98  *     end of its input, and never writes past the end of its output.
99  *
100  * (2) LZJB allows a maximum match length of 66 (vs. 18 for LZRW1), with
101  *     the trade-off being a shorter look-behind (1K vs. 4K for LZRW1).
102  *
103  * (3) LZJB records only the low-order 16 bits of pointers in the Lempel
104  *     history (which is all we need since the maximum look-behind is 1K),
105  *     and uses only 256 hash entries (vs. 4096 for LZRW1).  This makes
106  *     the compression hash small enough to allocate on the stack, which
107  *     solves two problems: (1) it saves 64K of kernel/cprboot memory,
108  *     and (2) it makes the code MT-safe without any locking, since we
109  *     don't have multiple threads sharing a common hash table.
110  *
111  * (4) LZJB is faster at both compression and decompression, has a
112  *     better compression ratio, and is somewhat simpler than LZRW1.
113  *
114  * Finally, note that LZJB is non-deterministic: given the same input,
115  * two calls to compress() may produce different output.  This is a
116  * general characteristic of most Lempel-Ziv derivatives because there's
117  * no need to initialize the Lempel history; not doing so saves time.
118  */
119 
120 #include <sys/types.h>
121 
122 #define	MATCH_BITS	6
123 #define	MATCH_MIN	3
124 #define	MATCH_MAX	((1 << MATCH_BITS) + (MATCH_MIN - 1))
125 #define	OFFSET_MASK	((1 << (16 - MATCH_BITS)) - 1)
126 #define	LEMPEL_SIZE	256
127 
128 size_t
129 compress(void *s_start, void *d_start, size_t s_len)
130 {
131 	uchar_t *src = s_start;
132 	uchar_t *dst = d_start;
133 	uchar_t *cpy, *copymap;
134 	int copymask = 1 << (NBBY - 1);
135 	int mlen, offset;
136 	uint16_t *hp;
137 	uint16_t lempel[LEMPEL_SIZE];	/* uninitialized; see above */
138 
139 	while (src < (uchar_t *)s_start + s_len) {
140 		if ((copymask <<= 1) == (1 << NBBY)) {
141 			if (dst >= (uchar_t *)d_start + s_len - 1 - 2 * NBBY) {
142 				mlen = s_len;
143 				for (src = s_start, dst = d_start; mlen; mlen--)
144 					*dst++ = *src++;
145 				return (s_len);
146 			}
147 			copymask = 1;
148 			copymap = dst;
149 			*dst++ = 0;
150 		}
151 		if (src > (uchar_t *)s_start + s_len - MATCH_MAX) {
152 			*dst++ = *src++;
153 			continue;
154 		}
155 		hp = &lempel[((src[0] + 13) ^ (src[1] - 13) ^ src[2]) &
156 		    (LEMPEL_SIZE - 1)];
157 		offset = (intptr_t)(src - *hp) & OFFSET_MASK;
158 		*hp = (uint16_t)(uintptr_t)src;
159 		cpy = src - offset;
160 		if (cpy >= (uchar_t *)s_start && cpy != src &&
161 		    src[0] == cpy[0] && src[1] == cpy[1] && src[2] == cpy[2]) {
162 			*copymap |= copymask;
163 			for (mlen = MATCH_MIN; mlen < MATCH_MAX; mlen++)
164 				if (src[mlen] != cpy[mlen])
165 					break;
166 			*dst++ = ((mlen - MATCH_MIN) << (NBBY - MATCH_BITS)) |
167 			    (offset >> NBBY);
168 			*dst++ = (uchar_t)offset;
169 			src += mlen;
170 		} else {
171 			*dst++ = *src++;
172 		}
173 	}
174 	return (dst - (uchar_t *)d_start);
175 }
176 
177 size_t
178 decompress(void *s_start, void *d_start, size_t s_len, size_t d_len)
179 {
180 	uchar_t *src = s_start;
181 	uchar_t *dst = d_start;
182 	uchar_t *s_end = (uchar_t *)s_start + s_len;
183 	uchar_t *d_end = (uchar_t *)d_start + d_len;
184 	uchar_t *cpy, copymap;
185 	int copymask = 1 << (NBBY - 1);
186 
187 	if (s_len >= d_len) {
188 		size_t d_rem = d_len;
189 		while (d_rem-- != 0)
190 			*dst++ = *src++;
191 		return (d_len);
192 	}
193 
194 	while (src < s_end && dst < d_end) {
195 		if ((copymask <<= 1) == (1 << NBBY)) {
196 			copymask = 1;
197 			copymap = *src++;
198 		}
199 		if (copymap & copymask) {
200 			int mlen = (src[0] >> (NBBY - MATCH_BITS)) + MATCH_MIN;
201 			int offset = ((src[0] << NBBY) | src[1]) & OFFSET_MASK;
202 			src += 2;
203 			if ((cpy = dst - offset) >= (uchar_t *)d_start)
204 				while (--mlen >= 0 && dst < d_end)
205 					*dst++ = *cpy++;
206 			else
207 				/*
208 				 * offset before start of destination buffer
209 				 * indicates corrupt source data
210 				 */
211 				return (dst - (uchar_t *)d_start);
212 		} else {
213 			*dst++ = *src++;
214 		}
215 	}
216 	return (dst - (uchar_t *)d_start);
217 }
218 
219 uint32_t
220 checksum32(void *cp_arg, size_t length)
221 {
222 	uchar_t *cp, *ep;
223 	uint32_t sum = 0;
224 
225 	for (cp = cp_arg, ep = cp + length; cp < ep; cp++)
226 		sum = ((sum >> 1) | (sum << 31)) + *cp;
227 	return (sum);
228 }
229