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  * Copyright (c) 2009, 2010, Oracle and/or its affiliates. All rights reserved.
  23  */
  24 
  25 #ifndef _SYS_DDT_H
  26 #define _SYS_DDT_H
  27 
  28 #include <sys/sysmacros.h>
  29 #include <sys/types.h>
  30 #include <sys/fs/zfs.h>
  31 #include <sys/zio.h>
  32 #include <sys/dmu.h>
  33 
  34 #ifdef  __cplusplus
  35 extern "C" {
  36 #endif
  37 
  38 /*
  39  * On-disk DDT formats, in the desired search order (newest version first).
  40  */
  41 enum ddt_type {
  42         DDT_TYPE_ZAP = 0,
  43         DDT_TYPES
  44 };
  45 
  46 /*
  47  * DDT classes, in the desired search order (highest replication level first).
  48  */
  49 enum ddt_class {
  50         DDT_CLASS_DITTO = 0,
  51         DDT_CLASS_DUPLICATE,
  52         DDT_CLASS_UNIQUE,
  53         DDT_CLASSES
  54 };
  55 
  56 #define DDT_TYPE_CURRENT                0
  57 
  58 #define DDT_COMPRESS_BYTEORDER_MASK     0x80
  59 #define DDT_COMPRESS_FUNCTION_MASK      0x7f
  60 
  61 /*
  62  * On-disk ddt entry:  key (name) and physical storage (value).
  63  */
  64 typedef struct ddt_key {
  65         zio_cksum_t     ddk_cksum;      /* 256-bit block checksum */
  66         /*
  67          * Encoded with logical & physical size, and compression, as follows:
  68          *   +-------+-------+-------+-------+-------+-------+-------+-------+
  69          *   |   0   |   0   |   0   | comp  |     PSIZE     |     LSIZE     |
  70          *   +-------+-------+-------+-------+-------+-------+-------+-------+
  71          */
  72         uint64_t        ddk_prop;
  73 } ddt_key_t;
  74 
  75 #define DDK_GET_LSIZE(ddk)      \
  76         BF64_GET_SB((ddk)->ddk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1)
  77 #define DDK_SET_LSIZE(ddk, x)   \
  78         BF64_SET_SB((ddk)->ddk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1, x)
  79 
  80 #define DDK_GET_PSIZE(ddk)      \
  81         BF64_GET_SB((ddk)->ddk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1)
  82 #define DDK_SET_PSIZE(ddk, x)   \
  83         BF64_SET_SB((ddk)->ddk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1, x)
  84 
  85 #define DDK_GET_COMPRESS(ddk)           BF64_GET((ddk)->ddk_prop, 32, 8)
  86 #define DDK_SET_COMPRESS(ddk, x)        BF64_SET((ddk)->ddk_prop, 32, 8, x)
  87 
  88 #define DDT_KEY_WORDS   (sizeof (ddt_key_t) / sizeof (uint64_t))
  89 
  90 typedef struct ddt_phys {
  91         dva_t           ddp_dva[SPA_DVAS_PER_BP];
  92         uint64_t        ddp_refcnt;
  93         uint64_t        ddp_phys_birth;
  94 } ddt_phys_t;
  95 
  96 enum ddt_phys_type {
  97         DDT_PHYS_DITTO = 0,
  98         DDT_PHYS_SINGLE = 1,
  99         DDT_PHYS_DOUBLE = 2,
 100         DDT_PHYS_TRIPLE = 3,
 101         DDT_PHYS_TYPES
 102 };
 103 
 104 /*
 105  * In-core ddt entry
 106  */
 107 struct ddt_entry {
 108         ddt_key_t       dde_key;
 109         ddt_phys_t      dde_phys[DDT_PHYS_TYPES];
 110         zio_t           *dde_lead_zio[DDT_PHYS_TYPES];
 111         void            *dde_repair_data;
 112         enum ddt_type   dde_type;
 113         enum ddt_class  dde_class;
 114         uint8_t         dde_loading;
 115         uint8_t         dde_loaded;
 116         kcondvar_t      dde_cv;
 117         avl_node_t      dde_node;
 118 };
 119 
 120 /*
 121  * In-core ddt
 122  */
 123 struct ddt {
 124         kmutex_t        ddt_lock;
 125         avl_tree_t      ddt_tree;
 126         avl_tree_t      ddt_repair_tree;
 127         enum zio_checksum ddt_checksum;
 128         spa_t           *ddt_spa;
 129         objset_t        *ddt_os;
 130         uint64_t        ddt_stat_object;
 131         uint64_t        ddt_object[DDT_TYPES][DDT_CLASSES];
 132         ddt_histogram_t ddt_histogram[DDT_TYPES][DDT_CLASSES];
 133         ddt_histogram_t ddt_histogram_cache[DDT_TYPES][DDT_CLASSES];
 134         ddt_object_t    ddt_object_stats[DDT_TYPES][DDT_CLASSES];
 135         avl_node_t      ddt_node;
 136 };
 137 
 138 /*
 139  * In-core and on-disk bookmark for DDT walks
 140  */
 141 typedef struct ddt_bookmark {
 142         uint64_t        ddb_class;
 143         uint64_t        ddb_type;
 144         uint64_t        ddb_checksum;
 145         uint64_t        ddb_cursor;
 146 } ddt_bookmark_t;
 147 
 148 /*
 149  * Ops vector to access a specific DDT object type.
 150  */
 151 typedef struct ddt_ops {
 152         char ddt_op_name[32];
 153         int (*ddt_op_create)(objset_t *os, uint64_t *object, dmu_tx_t *tx,
 154             boolean_t prehash);
 155         int (*ddt_op_destroy)(objset_t *os, uint64_t object, dmu_tx_t *tx);
 156         int (*ddt_op_lookup)(objset_t *os, uint64_t object, ddt_entry_t *dde);
 157         void (*ddt_op_prefetch)(objset_t *os, uint64_t object,
 158             ddt_entry_t *dde);
 159         int (*ddt_op_update)(objset_t *os, uint64_t object, ddt_entry_t *dde,
 160             dmu_tx_t *tx);
 161         int (*ddt_op_remove)(objset_t *os, uint64_t object, ddt_entry_t *dde,
 162             dmu_tx_t *tx);
 163         int (*ddt_op_walk)(objset_t *os, uint64_t object, ddt_entry_t *dde,
 164             uint64_t *walk);
 165         uint64_t (*ddt_op_count)(objset_t *os, uint64_t object);
 166 } ddt_ops_t;
 167 
 168 #define DDT_NAMELEN     80
 169 
 170 extern void ddt_object_name(ddt_t *ddt, enum ddt_type type,
 171     enum ddt_class class, char *name);
 172 extern int ddt_object_walk(ddt_t *ddt, enum ddt_type type,
 173     enum ddt_class class, uint64_t *walk, ddt_entry_t *dde);
 174 extern uint64_t ddt_object_count(ddt_t *ddt, enum ddt_type type,
 175     enum ddt_class class);
 176 extern int ddt_object_info(ddt_t *ddt, enum ddt_type type,
 177     enum ddt_class class, dmu_object_info_t *);
 178 extern boolean_t ddt_object_exists(ddt_t *ddt, enum ddt_type type,
 179     enum ddt_class class);
 180 
 181 extern void ddt_bp_fill(const ddt_phys_t *ddp, blkptr_t *bp,
 182     uint64_t txg);
 183 extern void ddt_bp_create(enum zio_checksum checksum, const ddt_key_t *ddk,
 184     const ddt_phys_t *ddp, blkptr_t *bp);
 185 
 186 extern void ddt_key_fill(ddt_key_t *ddk, const blkptr_t *bp);
 187 
 188 extern void ddt_phys_fill(ddt_phys_t *ddp, const blkptr_t *bp);
 189 extern void ddt_phys_clear(ddt_phys_t *ddp);
 190 extern void ddt_phys_addref(ddt_phys_t *ddp);
 191 extern void ddt_phys_decref(ddt_phys_t *ddp);
 192 extern void ddt_phys_free(ddt_t *ddt, ddt_key_t *ddk, ddt_phys_t *ddp,
 193     uint64_t txg);
 194 extern ddt_phys_t *ddt_phys_select(const ddt_entry_t *dde, const blkptr_t *bp);
 195 extern uint64_t ddt_phys_total_refcnt(const ddt_entry_t *dde);
 196 
 197 extern void ddt_stat_add(ddt_stat_t *dst, const ddt_stat_t *src, uint64_t neg);
 198 
 199 extern void ddt_histogram_add(ddt_histogram_t *dst, const ddt_histogram_t *src);
 200 extern void ddt_histogram_stat(ddt_stat_t *dds, const ddt_histogram_t *ddh);
 201 extern boolean_t ddt_histogram_empty(const ddt_histogram_t *ddh);
 202 extern void ddt_get_dedup_object_stats(spa_t *spa, ddt_object_t *ddo);
 203 extern void ddt_get_dedup_histogram(spa_t *spa, ddt_histogram_t *ddh);
 204 extern void ddt_get_dedup_stats(spa_t *spa, ddt_stat_t *dds_total);
 205 
 206 extern uint64_t ddt_get_dedup_dspace(spa_t *spa);
 207 extern uint64_t ddt_get_pool_dedup_ratio(spa_t *spa);
 208 
 209 extern int ddt_ditto_copies_needed(ddt_t *ddt, ddt_entry_t *dde,
 210     ddt_phys_t *ddp_willref);
 211 extern int ddt_ditto_copies_present(ddt_entry_t *dde);
 212 
 213 extern size_t ddt_compress(void *src, uchar_t *dst, size_t s_len, size_t d_len);
 214 extern void ddt_decompress(uchar_t *src, void *dst, size_t s_len, size_t d_len);
 215 
 216 extern ddt_t *ddt_select(spa_t *spa, const blkptr_t *bp);
 217 extern void ddt_enter(ddt_t *ddt);
 218 extern void ddt_exit(ddt_t *ddt);
 219 extern ddt_entry_t *ddt_lookup(ddt_t *ddt, const blkptr_t *bp, boolean_t add);
 220 extern void ddt_prefetch(spa_t *spa, const blkptr_t *bp);
 221 extern void ddt_remove(ddt_t *ddt, ddt_entry_t *dde);
 222 
 223 extern boolean_t ddt_class_contains(spa_t *spa, enum ddt_class max_class,
 224     const blkptr_t *bp);
 225 
 226 extern ddt_entry_t *ddt_repair_start(ddt_t *ddt, const blkptr_t *bp);
 227 extern void ddt_repair_done(ddt_t *ddt, ddt_entry_t *dde);
 228 
 229 extern int ddt_entry_compare(const void *x1, const void *x2);
 230 
 231 extern void ddt_create(spa_t *spa);
 232 extern int ddt_load(spa_t *spa);
 233 extern void ddt_unload(spa_t *spa);
 234 extern void ddt_sync(spa_t *spa, uint64_t txg);
 235 extern int ddt_walk(spa_t *spa, ddt_bookmark_t *ddb, ddt_entry_t *dde);
 236 extern int ddt_object_update(ddt_t *ddt, enum ddt_type type,
 237     enum ddt_class class, ddt_entry_t *dde, dmu_tx_t *tx);
 238 
 239 extern const ddt_ops_t ddt_zap_ops;
 240 
 241 #ifdef  __cplusplus
 242 }
 243 #endif
 244 
 245 #endif  /* _SYS_DDT_H */