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, Version 1.0 only 6 * (the "License"). You may not use this file except in compliance 7 * with the License. 8 * 9 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE 10 * or http://www.opensolaris.org/os/licensing. 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 * Copyright 2005 Sun Microsystems, Inc. All rights reserved. 24 * Use is subject to license terms. 25 */ 26 27 #ifndef _SYS_VDEV_IMPL_H 28 #define _SYS_VDEV_IMPL_H 29 30 #pragma ident "%Z%%M% %I% %E% SMI" 31 32 #include <sys/avl.h> 33 #include <sys/dmu.h> 34 #include <sys/metaslab.h> 35 #include <sys/nvpair.h> 36 #include <sys/space_map.h> 37 #include <sys/vdev.h> 38 #include <sys/dkio.h> 39 #include <sys/uberblock_impl.h> 40 41 #ifdef __cplusplus 42 extern "C" { 43 #endif 44 45 /* 46 * Virtual device descriptors. 47 * 48 * All storage pool operations go through the virtual device framework, 49 * which provides data replication and I/O scheduling. 50 */ 51 52 /* 53 * Forward declarations that lots of things need. 54 */ 55 typedef struct vdev_queue vdev_queue_t; 56 typedef struct vdev_cache vdev_cache_t; 57 typedef struct vdev_cache_entry vdev_cache_entry_t; 58 59 /* 60 * Virtual device operations 61 */ 62 typedef int vdev_open_func_t(vdev_t *vd, uint64_t *size, uint64_t *ashift); 63 typedef void vdev_close_func_t(vdev_t *vd); 64 typedef uint64_t vdev_asize_func_t(vdev_t *vd, uint64_t psize); 65 typedef void vdev_io_start_func_t(zio_t *zio); 66 typedef void vdev_io_done_func_t(zio_t *zio); 67 typedef void vdev_state_change_func_t(vdev_t *vd, int, int); 68 69 typedef struct vdev_ops { 70 vdev_open_func_t *vdev_op_open; 71 vdev_close_func_t *vdev_op_close; 72 vdev_asize_func_t *vdev_op_asize; 73 vdev_io_start_func_t *vdev_op_io_start; 74 vdev_io_done_func_t *vdev_op_io_done; 75 vdev_state_change_func_t *vdev_op_state_change; 76 char vdev_op_type[16]; 77 boolean_t vdev_op_leaf; 78 } vdev_ops_t; 79 80 /* 81 * Virtual device properties 82 */ 83 struct vdev_cache_entry { 84 char *ve_data; 85 uint64_t ve_offset; 86 uint64_t ve_lastused; 87 avl_node_t ve_offset_node; 88 avl_node_t ve_lastused_node; 89 uint32_t ve_hits; 90 uint16_t ve_missed_update; 91 zio_t *ve_fill_io; 92 }; 93 94 struct vdev_cache { 95 uint64_t vc_size; 96 uint64_t vc_bshift; 97 uint64_t vc_blocksize; 98 uint64_t vc_max; 99 avl_tree_t vc_offset_tree; 100 avl_tree_t vc_lastused_tree; 101 kmutex_t vc_lock; 102 }; 103 104 struct vdev_queue { 105 uint64_t vq_min_pending; 106 uint64_t vq_max_pending; 107 uint64_t vq_agg_limit; 108 uint64_t vq_time_shift; 109 uint64_t vq_ramp_rate; 110 avl_tree_t vq_deadline_tree; 111 avl_tree_t vq_read_tree; 112 avl_tree_t vq_write_tree; 113 avl_tree_t vq_pending_tree; 114 kmutex_t vq_lock; 115 }; 116 117 /* 118 * Virtual device descriptor 119 */ 120 struct vdev { 121 /* 122 * Common to all vdev types. 123 */ 124 uint64_t vdev_id; /* child number in vdev parent */ 125 uint64_t vdev_guid; /* unique ID for this vdev */ 126 uint64_t vdev_guid_sum; /* self guid + all child guids */ 127 uint64_t vdev_asize; /* allocatable device capacity */ 128 uint64_t vdev_ashift; /* block alignment shift */ 129 uint64_t vdev_state; /* see VDEV_STATE_* #defines */ 130 vdev_ops_t *vdev_ops; /* vdev operations */ 131 spa_t *vdev_spa; /* spa for this vdev */ 132 void *vdev_tsd; /* type-specific data */ 133 vdev_t *vdev_top; /* top-level vdev */ 134 vdev_t *vdev_parent; /* parent vdev */ 135 vdev_t **vdev_child; /* array of children */ 136 uint64_t vdev_children; /* number of children */ 137 space_map_t vdev_dtl_map; /* dirty time log in-core state */ 138 space_map_t vdev_dtl_scrub; /* DTL for scrub repair writes */ 139 vdev_stat_t vdev_stat; /* virtual device statistics */ 140 141 /* 142 * Top-level vdev state. 143 */ 144 uint64_t vdev_ms_array; /* metaslab array object */ 145 uint64_t vdev_ms_shift; /* metaslab size shift */ 146 uint64_t vdev_ms_count; /* number of metaslabs */ 147 metaslab_group_t *vdev_mg; /* metaslab group */ 148 metaslab_t **vdev_ms; /* metaslab array */ 149 space_map_obj_t *vdev_smo; /* metaslab space map array */ 150 txg_list_t vdev_ms_list; /* per-txg dirty metaslab lists */ 151 txg_list_t vdev_dtl_list; /* per-txg dirty DTL lists */ 152 txg_node_t vdev_txg_node; /* per-txg dirty vdev linkage */ 153 uint8_t vdev_dirty[TXG_SIZE]; /* per-txg dirty flags */ 154 int vdev_is_dirty; /* on config dirty list? */ 155 list_node_t vdev_dirty_node; /* config dirty list */ 156 zio_t *vdev_io_retry; /* I/O retry list */ 157 list_t vdev_io_pending; /* I/O pending list */ 158 159 /* 160 * Leaf vdev state. 161 */ 162 uint64_t vdev_psize; /* physical device capacity */ 163 space_map_obj_t vdev_dtl; /* dirty time log on-disk state */ 164 txg_node_t vdev_dtl_node; /* per-txg dirty DTL linkage */ 165 char *vdev_path; /* vdev path (if any) */ 166 char *vdev_devid; /* vdev devid (if any) */ 167 uint64_t vdev_fault_arg; /* fault injection paramater */ 168 int vdev_fault_mask; /* zio types to fault */ 169 uint8_t vdev_fault_mode; /* fault injection mode */ 170 uint8_t vdev_cache_active; /* vdev_cache and vdev_queue */ 171 uint8_t vdev_offline; /* device taken offline? */ 172 uint8_t vdev_detached; /* device detached? */ 173 vdev_queue_t vdev_queue; /* I/O deadline schedule queue */ 174 vdev_cache_t vdev_cache; /* physical block cache */ 175 176 /* 177 * For DTrace to work in userland (libzpool) context, these fields must 178 * remain at the end of the structure. DTrace will use the kernel's 179 * CTF definition for 'struct vdev', and since the size of a kmutex_t is 180 * larger in userland, the offsets for the rest fields would be 181 * incorrect. 182 */ 183 kmutex_t vdev_dtl_lock; /* vdev_dtl_{map,resilver} */ 184 kmutex_t vdev_dirty_lock; /* vdev_dirty[] */ 185 kmutex_t vdev_io_lock; /* vdev_io_pending list */ 186 kcondvar_t vdev_io_cv; /* vdev_io_pending list empty? */ 187 kmutex_t vdev_stat_lock; /* vdev_stat */ 188 }; 189 190 #define VDEV_SKIP_SIZE (8 << 10) 191 #define VDEV_BOOT_HEADER_SIZE (8 << 10) 192 #define VDEV_PHYS_SIZE (112 << 10) 193 #define VDEV_UBERBLOCKS ((128 << 10) >> UBERBLOCK_SHIFT) 194 195 #define VDEV_BOOT_MAGIC 0x2f5b007b10c /* ZFS boot block */ 196 #define VDEV_BOOT_VERSION 1 /* version number */ 197 198 typedef struct vdev_boot_header { 199 uint64_t vb_magic; /* VDEV_BOOT_MAGIC */ 200 uint64_t vb_version; /* VDEV_BOOT_VERSION */ 201 uint64_t vb_offset; /* start offset (bytes) */ 202 uint64_t vb_size; /* size (bytes) */ 203 char vb_pad[VDEV_BOOT_HEADER_SIZE - 4 * sizeof (uint64_t)]; 204 } vdev_boot_header_t; 205 206 typedef struct vdev_phys { 207 char vp_nvlist[VDEV_PHYS_SIZE - sizeof (zio_block_tail_t)]; 208 zio_block_tail_t vp_zbt; 209 } vdev_phys_t; 210 211 typedef struct vdev_label { 212 char vl_pad[VDEV_SKIP_SIZE]; /* 8K */ 213 vdev_boot_header_t vl_boot_header; /* 8K */ 214 vdev_phys_t vl_vdev_phys; /* 112K */ 215 uberblock_phys_t vl_uberblock[VDEV_UBERBLOCKS]; /* 128K */ 216 } vdev_label_t; /* 256K total */ 217 218 /* 219 * Size and offset of embedded boot loader region on each label. 220 * The total size of the first two labels plus the boot area is 4MB. 221 */ 222 #define VDEV_BOOT_OFFSET (2 * sizeof (vdev_label_t)) 223 #define VDEV_BOOT_SIZE (7ULL << 19) /* 3.5M */ 224 225 /* 226 * vdev_dirty[] flags 227 */ 228 #define VDD_ALLOC 0x01 /* allocated from in this txg */ 229 #define VDD_FREE 0x02 /* freed to in this txg */ 230 #define VDD_ADD 0x04 /* added to the pool in this txg */ 231 #define VDD_DTL 0x08 /* dirty time log entry in this txg */ 232 233 /* 234 * Size of label regions at the start and end of each leaf device. 235 */ 236 #define VDEV_LABEL_START_SIZE (2 * sizeof (vdev_label_t) + VDEV_BOOT_SIZE) 237 #define VDEV_LABEL_END_SIZE (2 * sizeof (vdev_label_t)) 238 #define VDEV_LABELS 4 239 240 #define VDEV_ALLOC_LOAD 0 241 #define VDEV_ALLOC_ADD 1 242 243 /* 244 * Allocate or free a vdev 245 */ 246 extern vdev_t *vdev_alloc(spa_t *spa, nvlist_t *config, vdev_t *parent, 247 uint_t id, int alloctype); 248 extern void vdev_free(vdev_t *vd); 249 250 /* 251 * Add or remove children and parents 252 */ 253 extern void vdev_add_child(vdev_t *pvd, vdev_t *cvd); 254 extern void vdev_remove_child(vdev_t *pvd, vdev_t *cvd); 255 extern void vdev_compact_children(vdev_t *pvd); 256 extern vdev_t *vdev_add_parent(vdev_t *cvd, vdev_ops_t *ops); 257 extern void vdev_remove_parent(vdev_t *cvd); 258 259 /* 260 * vdev sync load and sync 261 */ 262 extern int vdev_load(vdev_t *vd, int import); 263 extern void vdev_sync(vdev_t *vd, uint64_t txg); 264 extern void vdev_sync_done(vdev_t *vd, uint64_t txg); 265 extern void vdev_dirty(vdev_t *vd, uint8_t flags, uint64_t txg); 266 267 /* 268 * Available vdev types. 269 */ 270 extern vdev_ops_t vdev_root_ops; 271 extern vdev_ops_t vdev_mirror_ops; 272 extern vdev_ops_t vdev_replacing_ops; 273 extern vdev_ops_t vdev_raidz_ops; 274 extern vdev_ops_t vdev_disk_ops; 275 extern vdev_ops_t vdev_file_ops; 276 extern vdev_ops_t vdev_missing_ops; 277 278 /* 279 * Common asize function 280 */ 281 extern uint64_t vdev_default_asize(vdev_t *vd, uint64_t psize); 282 283 #ifdef __cplusplus 284 } 285 #endif 286 287 #endif /* _SYS_VDEV_IMPL_H */ 288