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