xref: /freebsd/sys/contrib/openzfs/module/zfs/lzjb.c (revision 61145dc2b94f12f6a47344fb9aac702321880e43)
1*61145dc2SMartin Matuska // SPDX-License-Identifier: CDDL-1.0
2eda14cbcSMatt Macy /*
3eda14cbcSMatt Macy  * CDDL HEADER START
4eda14cbcSMatt Macy  *
5eda14cbcSMatt Macy  * The contents of this file are subject to the terms of the
6eda14cbcSMatt Macy  * Common Development and Distribution License (the "License").
7eda14cbcSMatt Macy  * You may not use this file except in compliance with the License.
8eda14cbcSMatt Macy  *
9eda14cbcSMatt Macy  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
10271171e0SMartin Matuska  * or https://opensource.org/licenses/CDDL-1.0.
11eda14cbcSMatt Macy  * See the License for the specific language governing permissions
12eda14cbcSMatt Macy  * and limitations under the License.
13eda14cbcSMatt Macy  *
14eda14cbcSMatt Macy  * When distributing Covered Code, include this CDDL HEADER in each
15eda14cbcSMatt Macy  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
16eda14cbcSMatt Macy  * If applicable, add the following below this CDDL HEADER, with the
17eda14cbcSMatt Macy  * fields enclosed by brackets "[]" replaced with your own identifying
18eda14cbcSMatt Macy  * information: Portions Copyright [yyyy] [name of copyright owner]
19eda14cbcSMatt Macy  *
20eda14cbcSMatt Macy  * CDDL HEADER END
21eda14cbcSMatt Macy  */
22eda14cbcSMatt Macy 
23eda14cbcSMatt Macy /*
24eda14cbcSMatt Macy  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
25eda14cbcSMatt Macy  */
26eda14cbcSMatt Macy 
27eda14cbcSMatt Macy /*
28eda14cbcSMatt Macy  * We keep our own copy of this algorithm for 3 main reasons:
29eda14cbcSMatt Macy  *	1. If we didn't, anyone modifying common/os/compress.c would
30eda14cbcSMatt Macy  *         directly break our on disk format
31eda14cbcSMatt Macy  *	2. Our version of lzjb does not have a number of checks that the
32eda14cbcSMatt Macy  *         common/os version needs and uses
33eda14cbcSMatt Macy  *	3. We initialize the lempel to ensure deterministic results,
34eda14cbcSMatt Macy  *	   so that identical blocks can always be deduplicated.
35eda14cbcSMatt Macy  * In particular, we are adding the "feature" that compress() can
36eda14cbcSMatt Macy  * take a destination buffer size and returns the compressed length, or the
37eda14cbcSMatt Macy  * source length if compression would overflow the destination buffer.
38eda14cbcSMatt Macy  */
39eda14cbcSMatt Macy 
40eda14cbcSMatt Macy #include <sys/zfs_context.h>
41eda14cbcSMatt Macy #include <sys/zio_compress.h>
42eda14cbcSMatt Macy 
43eda14cbcSMatt Macy #define	MATCH_BITS	6
44eda14cbcSMatt Macy #define	MATCH_MIN	3
45eda14cbcSMatt Macy #define	MATCH_MAX	((1 << MATCH_BITS) + (MATCH_MIN - 1))
46eda14cbcSMatt Macy #define	OFFSET_MASK	((1 << (16 - MATCH_BITS)) - 1)
47eda14cbcSMatt Macy #define	LEMPEL_SIZE	1024
48eda14cbcSMatt Macy 
49e2df9bb4SMartin Matuska static size_t
zfs_lzjb_compress_buf(void * s_start,void * d_start,size_t s_len,size_t d_len,int n)50e2df9bb4SMartin Matuska zfs_lzjb_compress_buf(void *s_start, void *d_start, size_t s_len,
51e2df9bb4SMartin Matuska     size_t d_len, int n)
52eda14cbcSMatt Macy {
53e92ffd9bSMartin Matuska 	(void) n;
54eda14cbcSMatt Macy 	uchar_t *src = s_start;
55eda14cbcSMatt Macy 	uchar_t *dst = d_start;
56eda14cbcSMatt Macy 	uchar_t *cpy;
57eda14cbcSMatt Macy 	uchar_t *copymap = NULL;
58eda14cbcSMatt Macy 	int copymask = 1 << (NBBY - 1);
59eda14cbcSMatt Macy 	int mlen, offset, hash;
60eda14cbcSMatt Macy 	uint16_t *hp;
61eda14cbcSMatt Macy 	uint16_t *lempel;
62eda14cbcSMatt Macy 
63eda14cbcSMatt Macy 	lempel = kmem_zalloc(LEMPEL_SIZE * sizeof (uint16_t), KM_SLEEP);
64eda14cbcSMatt Macy 	while (src < (uchar_t *)s_start + s_len) {
65eda14cbcSMatt Macy 		if ((copymask <<= 1) == (1 << NBBY)) {
66eda14cbcSMatt Macy 			if (dst >= (uchar_t *)d_start + d_len - 1 - 2 * NBBY) {
67eda14cbcSMatt Macy 				kmem_free(lempel,
68eda14cbcSMatt Macy 				    LEMPEL_SIZE*sizeof (uint16_t));
69eda14cbcSMatt Macy 				return (s_len);
70eda14cbcSMatt Macy 			}
71eda14cbcSMatt Macy 			copymask = 1;
72eda14cbcSMatt Macy 			copymap = dst;
73eda14cbcSMatt Macy 			*dst++ = 0;
74eda14cbcSMatt Macy 		}
75eda14cbcSMatt Macy 		if (src > (uchar_t *)s_start + s_len - MATCH_MAX) {
76eda14cbcSMatt Macy 			*dst++ = *src++;
77eda14cbcSMatt Macy 			continue;
78eda14cbcSMatt Macy 		}
79eda14cbcSMatt Macy 		hash = (src[0] << 16) + (src[1] << 8) + src[2];
80eda14cbcSMatt Macy 		hash += hash >> 9;
81eda14cbcSMatt Macy 		hash += hash >> 5;
82eda14cbcSMatt Macy 		hp = &lempel[hash & (LEMPEL_SIZE - 1)];
83eda14cbcSMatt Macy 		offset = (intptr_t)(src - *hp) & OFFSET_MASK;
84eda14cbcSMatt Macy 		*hp = (uint16_t)(uintptr_t)src;
85eda14cbcSMatt Macy 		cpy = src - offset;
86eda14cbcSMatt Macy 		if (cpy >= (uchar_t *)s_start && cpy != src &&
87eda14cbcSMatt Macy 		    src[0] == cpy[0] && src[1] == cpy[1] && src[2] == cpy[2]) {
88eda14cbcSMatt Macy 			*copymap |= copymask;
89eda14cbcSMatt Macy 			for (mlen = MATCH_MIN; mlen < MATCH_MAX; mlen++)
90eda14cbcSMatt Macy 				if (src[mlen] != cpy[mlen])
91eda14cbcSMatt Macy 					break;
92eda14cbcSMatt Macy 			*dst++ = ((mlen - MATCH_MIN) << (NBBY - MATCH_BITS)) |
93eda14cbcSMatt Macy 			    (offset >> NBBY);
94eda14cbcSMatt Macy 			*dst++ = (uchar_t)offset;
95eda14cbcSMatt Macy 			src += mlen;
96eda14cbcSMatt Macy 		} else {
97eda14cbcSMatt Macy 			*dst++ = *src++;
98eda14cbcSMatt Macy 		}
99eda14cbcSMatt Macy 	}
100eda14cbcSMatt Macy 
101eda14cbcSMatt Macy 	kmem_free(lempel, LEMPEL_SIZE * sizeof (uint16_t));
102eda14cbcSMatt Macy 	return (dst - (uchar_t *)d_start);
103eda14cbcSMatt Macy }
104eda14cbcSMatt Macy 
105e2df9bb4SMartin Matuska static int
zfs_lzjb_decompress_buf(void * s_start,void * d_start,size_t s_len,size_t d_len,int n)106e2df9bb4SMartin Matuska zfs_lzjb_decompress_buf(void *s_start, void *d_start,
107e2df9bb4SMartin Matuska     size_t s_len, size_t d_len, int n)
108eda14cbcSMatt Macy {
109e92ffd9bSMartin Matuska 	(void) s_len, (void) n;
110eda14cbcSMatt Macy 	uchar_t *src = s_start;
111eda14cbcSMatt Macy 	uchar_t *dst = d_start;
112eda14cbcSMatt Macy 	uchar_t *d_end = (uchar_t *)d_start + d_len;
113eda14cbcSMatt Macy 	uchar_t *cpy;
114eda14cbcSMatt Macy 	uchar_t copymap = 0;
115eda14cbcSMatt Macy 	int copymask = 1 << (NBBY - 1);
116eda14cbcSMatt Macy 
117eda14cbcSMatt Macy 	while (dst < d_end) {
118eda14cbcSMatt Macy 		if ((copymask <<= 1) == (1 << NBBY)) {
119eda14cbcSMatt Macy 			copymask = 1;
120eda14cbcSMatt Macy 			copymap = *src++;
121eda14cbcSMatt Macy 		}
122eda14cbcSMatt Macy 		if (copymap & copymask) {
123eda14cbcSMatt Macy 			int mlen = (src[0] >> (NBBY - MATCH_BITS)) + MATCH_MIN;
124eda14cbcSMatt Macy 			int offset = ((src[0] << NBBY) | src[1]) & OFFSET_MASK;
125eda14cbcSMatt Macy 			src += 2;
126eda14cbcSMatt Macy 			if ((cpy = dst - offset) < (uchar_t *)d_start)
127eda14cbcSMatt Macy 				return (-1);
128eda14cbcSMatt Macy 			while (--mlen >= 0 && dst < d_end)
129eda14cbcSMatt Macy 				*dst++ = *cpy++;
130eda14cbcSMatt Macy 		} else {
131eda14cbcSMatt Macy 			*dst++ = *src++;
132eda14cbcSMatt Macy 		}
133eda14cbcSMatt Macy 	}
134eda14cbcSMatt Macy 	return (0);
135eda14cbcSMatt Macy }
136e2df9bb4SMartin Matuska 
137e2df9bb4SMartin Matuska ZFS_COMPRESS_WRAP_DECL(zfs_lzjb_compress)
138e2df9bb4SMartin Matuska ZFS_DECOMPRESS_WRAP_DECL(zfs_lzjb_decompress)
139