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 (the "License").
6 * You may not use this file except in compliance with the License.
7 *
8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9 * or http://www.opensolaris.org/os/licensing.
10 * See the License for the specific language governing permissions
11 * and limitations under the License.
12 *
13 * When distributing Covered Code, include this CDDL HEADER in each
14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15 * If applicable, add the following below this CDDL HEADER, with the
16 * fields enclosed by brackets "[]" replaced with your own identifying
17 * information: Portions Copyright [yyyy] [name of copyright owner]
18 *
19 * CDDL HEADER END
20 */
21
22 /*
23 * Copyright (c) 2009, 2010, Oracle and/or its affiliates. All rights reserved.
24 * Copyright (c) 2018 by Delphix. All rights reserved.
25 */
26
27 #include <sys/zfs_context.h>
28 #include <sys/spa.h>
29 #include <sys/zio.h>
30 #include <sys/ddt.h>
31 #include <sys/zap.h>
32 #include <sys/dmu_tx.h>
33 #include <util/sscanf.h>
34
35 int ddt_zap_leaf_blockshift = 12;
36 int ddt_zap_indirect_blockshift = 12;
37
38 static int
ddt_zap_create(objset_t * os,uint64_t * objectp,dmu_tx_t * tx,boolean_t prehash)39 ddt_zap_create(objset_t *os, uint64_t *objectp, dmu_tx_t *tx, boolean_t prehash)
40 {
41 zap_flags_t flags = ZAP_FLAG_HASH64 | ZAP_FLAG_UINT64_KEY;
42
43 if (prehash)
44 flags |= ZAP_FLAG_PRE_HASHED_KEY;
45
46 *objectp = zap_create_flags(os, 0, flags, DMU_OT_DDT_ZAP,
47 ddt_zap_leaf_blockshift, ddt_zap_indirect_blockshift,
48 DMU_OT_NONE, 0, tx);
49
50 return (*objectp == 0 ? ENOTSUP : 0);
51 }
52
53 static int
ddt_zap_destroy(objset_t * os,uint64_t object,dmu_tx_t * tx)54 ddt_zap_destroy(objset_t *os, uint64_t object, dmu_tx_t *tx)
55 {
56 return (zap_destroy(os, object, tx));
57 }
58
59 static int
ddt_zap_lookup(objset_t * os,uint64_t object,ddt_entry_t * dde)60 ddt_zap_lookup(objset_t *os, uint64_t object, ddt_entry_t *dde)
61 {
62 uchar_t cbuf[sizeof (dde->dde_phys) + 1];
63 uint64_t one, csize;
64 int error;
65
66 error = zap_length_uint64(os, object, (uint64_t *)&dde->dde_key,
67 DDT_KEY_WORDS, &one, &csize);
68 if (error)
69 return (error);
70
71 ASSERT(one == 1);
72 ASSERT(csize <= sizeof (cbuf));
73
74 error = zap_lookup_uint64(os, object, (uint64_t *)&dde->dde_key,
75 DDT_KEY_WORDS, 1, csize, cbuf);
76 if (error)
77 return (error);
78
79 ddt_decompress(cbuf, dde->dde_phys, csize, sizeof (dde->dde_phys));
80
81 return (0);
82 }
83
84 static void
ddt_zap_prefetch(objset_t * os,uint64_t object,ddt_entry_t * dde)85 ddt_zap_prefetch(objset_t *os, uint64_t object, ddt_entry_t *dde)
86 {
87 (void) zap_prefetch_uint64(os, object, (uint64_t *)&dde->dde_key,
88 DDT_KEY_WORDS);
89 }
90
91 static int
ddt_zap_update(objset_t * os,uint64_t object,ddt_entry_t * dde,dmu_tx_t * tx)92 ddt_zap_update(objset_t *os, uint64_t object, ddt_entry_t *dde, dmu_tx_t *tx)
93 {
94 uchar_t cbuf[sizeof (dde->dde_phys) + 1];
95 uint64_t csize;
96
97 csize = ddt_compress(dde->dde_phys, cbuf,
98 sizeof (dde->dde_phys), sizeof (cbuf));
99
100 return (zap_update_uint64(os, object, (uint64_t *)&dde->dde_key,
101 DDT_KEY_WORDS, 1, csize, cbuf, tx));
102 }
103
104 static int
ddt_zap_remove(objset_t * os,uint64_t object,ddt_entry_t * dde,dmu_tx_t * tx)105 ddt_zap_remove(objset_t *os, uint64_t object, ddt_entry_t *dde, dmu_tx_t *tx)
106 {
107 return (zap_remove_uint64(os, object, (uint64_t *)&dde->dde_key,
108 DDT_KEY_WORDS, tx));
109 }
110
111 static int
ddt_zap_walk(objset_t * os,uint64_t object,ddt_entry_t * dde,uint64_t * walk)112 ddt_zap_walk(objset_t *os, uint64_t object, ddt_entry_t *dde, uint64_t *walk)
113 {
114 zap_cursor_t zc;
115 zap_attribute_t za;
116 int error;
117
118 if (*walk == 0) {
119 /*
120 * We don't want to prefetch the entire ZAP object, because
121 * it can be enormous. Also the primary use of DDT iteration
122 * is for scrubbing, in which case we will be issuing many
123 * scrub i/os for each ZAP block that we read in, so
124 * reading the ZAP is unlikely to be the bottleneck.
125 */
126 zap_cursor_init_noprefetch(&zc, os, object);
127 } else {
128 zap_cursor_init_serialized(&zc, os, object, *walk);
129 }
130 if ((error = zap_cursor_retrieve(&zc, &za)) == 0) {
131 uchar_t cbuf[sizeof (dde->dde_phys) + 1];
132 uint64_t csize = za.za_num_integers;
133 ASSERT(za.za_integer_length == 1);
134 error = zap_lookup_uint64(os, object, (uint64_t *)za.za_name,
135 DDT_KEY_WORDS, 1, csize, cbuf);
136 ASSERT(error == 0);
137 if (error == 0) {
138 ddt_decompress(cbuf, dde->dde_phys, csize,
139 sizeof (dde->dde_phys));
140 dde->dde_key = *(ddt_key_t *)za.za_name;
141 }
142 zap_cursor_advance(&zc);
143 *walk = zap_cursor_serialize(&zc);
144 }
145 zap_cursor_fini(&zc);
146 return (error);
147 }
148
149 static uint64_t
ddt_zap_count(objset_t * os,uint64_t object)150 ddt_zap_count(objset_t *os, uint64_t object)
151 {
152 uint64_t count = 0;
153
154 VERIFY(zap_count(os, object, &count) == 0);
155
156 return (count);
157 }
158
159 const ddt_ops_t ddt_zap_ops = {
160 "zap",
161 ddt_zap_create,
162 ddt_zap_destroy,
163 ddt_zap_lookup,
164 ddt_zap_prefetch,
165 ddt_zap_update,
166 ddt_zap_remove,
167 ddt_zap_walk,
168 ddt_zap_count,
169 };
170