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         uint64_t        ddk_prop;       /* LSIZE, PSIZE, compression */
  67 } ddt_key_t;
  68 
  69 /*
  70  * ddk_prop layout:
  71  *
  72  *      +-------+-------+-------+-------+-------+-------+-------+-------+
  73  *      |   0   |   0   |   0   | comp  |     PSIZE     |     LSIZE     |
  74  *      +-------+-------+-------+-------+-------+-------+-------+-------+
  75  */
  76 #define DDK_GET_LSIZE(ddk)      \
  77         BF64_GET_SB((ddk)->ddk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1)
  78 #define DDK_SET_LSIZE(ddk, x)   \
  79         BF64_SET_SB((ddk)->ddk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1, x)
  80 
  81 #define DDK_GET_PSIZE(ddk)      \
  82         BF64_GET_SB((ddk)->ddk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1)
  83 #define DDK_SET_PSIZE(ddk, x)   \
  84         BF64_SET_SB((ddk)->ddk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1, x)
  85 
  86 #define DDK_GET_COMPRESS(ddk)           BF64_GET((ddk)->ddk_prop, 32, 8)
  87 #define DDK_SET_COMPRESS(ddk, x)        BF64_SET((ddk)->ddk_prop, 32, 8, x)
  88 
  89 #define DDT_KEY_WORDS   (sizeof (ddt_key_t) / sizeof (uint64_t))
  90 
  91 typedef struct ddt_phys {
  92         dva_t           ddp_dva[SPA_DVAS_PER_BP];
  93         uint64_t        ddp_refcnt;
  94         uint64_t        ddp_phys_birth;
  95 } ddt_phys_t;
  96 
  97 enum ddt_phys_type {
  98         DDT_PHYS_DITTO = 0,
  99         DDT_PHYS_SINGLE = 1,
 100         DDT_PHYS_DOUBLE = 2,
 101         DDT_PHYS_TRIPLE = 3,
 102         DDT_PHYS_TYPES
 103 };
 104 
 105 /*
 106  * In-core ddt entry
 107  */
 108 struct ddt_entry {
 109         ddt_key_t       dde_key;
 110         ddt_phys_t      dde_phys[DDT_PHYS_TYPES];
 111         zio_t           *dde_lead_zio[DDT_PHYS_TYPES];
 112         void            *dde_repair_data;
 113         enum ddt_type   dde_type;
 114         enum ddt_class  dde_class;
 115         uint8_t         dde_loading;
 116         uint8_t         dde_loaded;
 117         kcondvar_t      dde_cv;
 118         avl_node_t      dde_node;
 119 };
 120 
 121 /*
 122  * In-core ddt
 123  */
 124 struct ddt {
 125         kmutex_t        ddt_lock;
 126         avl_tree_t      ddt_tree;
 127         avl_tree_t      ddt_repair_tree;
 128         enum zio_checksum ddt_checksum;
 129         spa_t           *ddt_spa;
 130         objset_t        *ddt_os;
 131         uint64_t        ddt_stat_object;
 132         uint64_t        ddt_object[DDT_TYPES][DDT_CLASSES];
 133         ddt_histogram_t ddt_histogram[DDT_TYPES][DDT_CLASSES];
 134         ddt_histogram_t ddt_histogram_cache[DDT_TYPES][DDT_CLASSES];
 135         ddt_object_t    ddt_object_stats[DDT_TYPES][DDT_CLASSES];
 136         avl_node_t      ddt_node;
 137 };
 138 
 139 /*
 140  * In-core and on-disk bookmark for DDT walks
 141  */
 142 typedef struct ddt_bookmark {
 143         uint64_t        ddb_class;
 144         uint64_t        ddb_type;
 145         uint64_t        ddb_checksum;
 146         uint64_t        ddb_cursor;
 147 } ddt_bookmark_t;
 148 
 149 /*
 150  * Ops vector to access a specific DDT object type.
 151  */
 152 typedef struct ddt_ops {
 153         char ddt_op_name[32];
 154         int (*ddt_op_create)(objset_t *os, uint64_t *object, dmu_tx_t *tx,
 155             boolean_t prehash);
 156         int (*ddt_op_destroy)(objset_t *os, uint64_t object, dmu_tx_t *tx);
 157         int (*ddt_op_lookup)(objset_t *os, uint64_t object, ddt_entry_t *dde);
 158         void (*ddt_op_prefetch)(objset_t *os, uint64_t object,
 159             ddt_entry_t *dde);
 160         int (*ddt_op_update)(objset_t *os, uint64_t object, ddt_entry_t *dde,
 161             dmu_tx_t *tx);
 162         int (*ddt_op_remove)(objset_t *os, uint64_t object, ddt_entry_t *dde,
 163             dmu_tx_t *tx);
 164         int (*ddt_op_walk)(objset_t *os, uint64_t object, ddt_entry_t *dde,
 165             uint64_t *walk);
 166         uint64_t (*ddt_op_count)(objset_t *os, uint64_t object);
 167 } ddt_ops_t;
 168 
 169 #define DDT_NAMELEN     80
 170 
 171 extern void ddt_object_name(ddt_t *ddt, enum ddt_type type,
 172     enum ddt_class class, char *name);
 173 extern int ddt_object_walk(ddt_t *ddt, enum ddt_type type,
 174     enum ddt_class class, uint64_t *walk, ddt_entry_t *dde);
 175 extern uint64_t ddt_object_count(ddt_t *ddt, enum ddt_type type,
 176     enum ddt_class class);
 177 extern int ddt_object_info(ddt_t *ddt, enum ddt_type type,
 178     enum ddt_class class, dmu_object_info_t *);
 179 extern boolean_t ddt_object_exists(ddt_t *ddt, enum ddt_type type,
 180     enum ddt_class class);
 181 
 182 extern void ddt_bp_fill(const ddt_phys_t *ddp, blkptr_t *bp,
 183     uint64_t txg);
 184 extern void ddt_bp_create(enum zio_checksum checksum, const ddt_key_t *ddk,
 185     const ddt_phys_t *ddp, blkptr_t *bp);
 186 
 187 extern void ddt_key_fill(ddt_key_t *ddk, const blkptr_t *bp);
 188 
 189 extern void ddt_phys_fill(ddt_phys_t *ddp, const blkptr_t *bp);
 190 extern void ddt_phys_clear(ddt_phys_t *ddp);
 191 extern void ddt_phys_addref(ddt_phys_t *ddp);
 192 extern void ddt_phys_decref(ddt_phys_t *ddp);
 193 extern void ddt_phys_free(ddt_t *ddt, ddt_key_t *ddk, ddt_phys_t *ddp,
 194     uint64_t txg);
 195 extern ddt_phys_t *ddt_phys_select(const ddt_entry_t *dde, const blkptr_t *bp);
 196 extern uint64_t ddt_phys_total_refcnt(const ddt_entry_t *dde);
 197 
 198 extern void ddt_stat_add(ddt_stat_t *dst, const ddt_stat_t *src, uint64_t neg);
 199 
 200 extern void ddt_histogram_add(ddt_histogram_t *dst, const ddt_histogram_t *src);
 201 extern void ddt_histogram_stat(ddt_stat_t *dds, const ddt_histogram_t *ddh);
 202 extern boolean_t ddt_histogram_empty(const ddt_histogram_t *ddh);
 203 extern void ddt_get_dedup_object_stats(spa_t *spa, ddt_object_t *ddo);
 204 extern void ddt_get_dedup_histogram(spa_t *spa, ddt_histogram_t *ddh);
 205 extern void ddt_get_dedup_stats(spa_t *spa, ddt_stat_t *dds_total);
 206 
 207 extern uint64_t ddt_get_dedup_dspace(spa_t *spa);
 208 extern uint64_t ddt_get_pool_dedup_ratio(spa_t *spa);
 209 
 210 extern int ddt_ditto_copies_needed(ddt_t *ddt, ddt_entry_t *dde,
 211     ddt_phys_t *ddp_willref);
 212 extern int ddt_ditto_copies_present(ddt_entry_t *dde);
 213 
 214 extern size_t ddt_compress(void *src, uchar_t *dst, size_t s_len, size_t d_len);
 215 extern void ddt_decompress(uchar_t *src, void *dst, size_t s_len, size_t d_len);
 216 
 217 extern ddt_t *ddt_select(spa_t *spa, const blkptr_t *bp);
 218 extern void ddt_enter(ddt_t *ddt);
 219 extern void ddt_exit(ddt_t *ddt);
 220 extern ddt_entry_t *ddt_lookup(ddt_t *ddt, const blkptr_t *bp, boolean_t add);
 221 extern void ddt_prefetch(spa_t *spa, const blkptr_t *bp);
 222 extern void ddt_remove(ddt_t *ddt, ddt_entry_t *dde);
 223 
 224 extern boolean_t ddt_class_contains(spa_t *spa, enum ddt_class max_class,
 225     const blkptr_t *bp);
 226 
 227 extern ddt_entry_t *ddt_repair_start(ddt_t *ddt, const blkptr_t *bp);
 228 extern void ddt_repair_done(ddt_t *ddt, ddt_entry_t *dde);
 229 
 230 extern int ddt_entry_compare(const void *x1, const void *x2);
 231 
 232 extern void ddt_create(spa_t *spa);
 233 extern int ddt_load(spa_t *spa);
 234 extern void ddt_unload(spa_t *spa);
 235 extern void ddt_sync(spa_t *spa, uint64_t txg);
 236 extern int ddt_walk(spa_t *spa, ddt_bookmark_t *ddb, ddt_entry_t *dde);
 237 extern int ddt_object_update(ddt_t *ddt, enum ddt_type type,
 238     enum ddt_class class, ddt_entry_t *dde, dmu_tx_t *tx);
 239 
 240 extern const ddt_ops_t ddt_zap_ops;
 241 
 242 #ifdef  __cplusplus
 243 }
 244 #endif
 245 
 246 #endif  /* _SYS_DDT_H */