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 2006 Sun Microsystems, Inc. All rights reserved. 23 * Use is subject to license terms. 24 */ 25 26 #ifndef _SYS_SPA_H 27 #define _SYS_SPA_H 28 29 #pragma ident "%Z%%M% %I% %E% SMI" 30 31 #include <sys/avl.h> 32 #include <sys/zfs_context.h> 33 #include <sys/nvpair.h> 34 #include <sys/sysmacros.h> 35 #include <sys/types.h> 36 #include <sys/fs/zfs.h> 37 38 #ifdef __cplusplus 39 extern "C" { 40 #endif 41 42 /* 43 * Forward references that lots of things need. 44 */ 45 typedef struct spa spa_t; 46 typedef struct vdev vdev_t; 47 typedef struct metaslab metaslab_t; 48 typedef struct zilog zilog_t; 49 typedef struct traverse_handle traverse_handle_t; 50 struct dsl_pool; 51 52 /* 53 * General-purpose 32-bit and 64-bit bitfield encodings. 54 */ 55 #define BF32_DECODE(x, low, len) P2PHASE((x) >> (low), 1U << (len)) 56 #define BF64_DECODE(x, low, len) P2PHASE((x) >> (low), 1ULL << (len)) 57 #define BF32_ENCODE(x, low, len) (P2PHASE((x), 1U << (len)) << (low)) 58 #define BF64_ENCODE(x, low, len) (P2PHASE((x), 1ULL << (len)) << (low)) 59 60 #define BF32_GET(x, low, len) BF32_DECODE(x, low, len) 61 #define BF64_GET(x, low, len) BF64_DECODE(x, low, len) 62 63 #define BF32_SET(x, low, len, val) \ 64 ((x) ^= BF32_ENCODE((x >> low) ^ (val), low, len)) 65 #define BF64_SET(x, low, len, val) \ 66 ((x) ^= BF64_ENCODE((x >> low) ^ (val), low, len)) 67 68 #define BF32_GET_SB(x, low, len, shift, bias) \ 69 ((BF32_GET(x, low, len) + (bias)) << (shift)) 70 #define BF64_GET_SB(x, low, len, shift, bias) \ 71 ((BF64_GET(x, low, len) + (bias)) << (shift)) 72 73 #define BF32_SET_SB(x, low, len, shift, bias, val) \ 74 BF32_SET(x, low, len, ((val) >> (shift)) - (bias)) 75 #define BF64_SET_SB(x, low, len, shift, bias, val) \ 76 BF64_SET(x, low, len, ((val) >> (shift)) - (bias)) 77 78 /* 79 * We currently support nine block sizes, from 512 bytes to 128K. 80 * We could go higher, but the benefits are near-zero and the cost 81 * of COWing a giant block to modify one byte would become excessive. 82 */ 83 #define SPA_MINBLOCKSHIFT 9 84 #define SPA_MAXBLOCKSHIFT 17 85 #define SPA_MINBLOCKSIZE (1ULL << SPA_MINBLOCKSHIFT) 86 #define SPA_MAXBLOCKSIZE (1ULL << SPA_MAXBLOCKSHIFT) 87 88 #define SPA_BLOCKSIZES (SPA_MAXBLOCKSHIFT - SPA_MINBLOCKSHIFT + 1) 89 90 /* 91 * The DVA size encodings for LSIZE and PSIZE support blocks up to 32MB. 92 * The ASIZE encoding should be at least 64 times larger (6 more bits) 93 * to support up to 4-way RAID-Z mirror mode with worst-case gang block 94 * overhead, three DVAs per bp, plus one more bit in case we do anything 95 * else that expands the ASIZE. 96 */ 97 #define SPA_LSIZEBITS 16 /* LSIZE up to 32M (2^16 * 512) */ 98 #define SPA_PSIZEBITS 16 /* PSIZE up to 32M (2^16 * 512) */ 99 #define SPA_ASIZEBITS 24 /* ASIZE up to 64 times larger */ 100 101 /* 102 * All SPA data is represented by 128-bit data virtual addresses (DVAs). 103 * The members of the dva_t should be considered opaque outside the SPA. 104 */ 105 typedef struct dva { 106 uint64_t dva_word[2]; 107 } dva_t; 108 109 /* 110 * Each block has a 256-bit checksum -- strong enough for cryptographic hashes. 111 */ 112 typedef struct zio_cksum { 113 uint64_t zc_word[4]; 114 } zio_cksum_t; 115 116 /* 117 * Each block is described by its DVAs, time of birth, checksum, etc. 118 * The word-by-word, bit-by-bit layout of the blkptr is as follows: 119 * 120 * 64 56 48 40 32 24 16 8 0 121 * +-------+-------+-------+-------+-------+-------+-------+-------+ 122 * 0 | vdev1 | GRID | ASIZE | 123 * +-------+-------+-------+-------+-------+-------+-------+-------+ 124 * 1 |G| offset1 | 125 * +-------+-------+-------+-------+-------+-------+-------+-------+ 126 * 2 | vdev2 | GRID | ASIZE | 127 * +-------+-------+-------+-------+-------+-------+-------+-------+ 128 * 3 |G| offset2 | 129 * +-------+-------+-------+-------+-------+-------+-------+-------+ 130 * 4 | vdev3 | GRID | ASIZE | 131 * +-------+-------+-------+-------+-------+-------+-------+-------+ 132 * 5 |G| offset3 | 133 * +-------+-------+-------+-------+-------+-------+-------+-------+ 134 * 6 |E| lvl | type | cksum | comp | PSIZE | LSIZE | 135 * +-------+-------+-------+-------+-------+-------+-------+-------+ 136 * 7 | padding | 137 * +-------+-------+-------+-------+-------+-------+-------+-------+ 138 * 8 | padding | 139 * +-------+-------+-------+-------+-------+-------+-------+-------+ 140 * 9 | padding | 141 * +-------+-------+-------+-------+-------+-------+-------+-------+ 142 * a | birth txg | 143 * +-------+-------+-------+-------+-------+-------+-------+-------+ 144 * b | fill count | 145 * +-------+-------+-------+-------+-------+-------+-------+-------+ 146 * c | checksum[0] | 147 * +-------+-------+-------+-------+-------+-------+-------+-------+ 148 * d | checksum[1] | 149 * +-------+-------+-------+-------+-------+-------+-------+-------+ 150 * e | checksum[2] | 151 * +-------+-------+-------+-------+-------+-------+-------+-------+ 152 * f | checksum[3] | 153 * +-------+-------+-------+-------+-------+-------+-------+-------+ 154 * 155 * Legend: 156 * 157 * vdev virtual device ID 158 * offset offset into virtual device 159 * LSIZE logical size 160 * PSIZE physical size (after compression) 161 * ASIZE allocated size (including RAID-Z parity and gang block headers) 162 * GRID RAID-Z layout information (reserved for future use) 163 * cksum checksum function 164 * comp compression function 165 * G gang block indicator 166 * E endianness 167 * type DMU object type 168 * lvl level of indirection 169 * birth txg transaction group in which the block was born 170 * fill count number of non-zero blocks under this bp 171 * checksum[4] 256-bit checksum of the data this bp describes 172 */ 173 typedef struct blkptr { 174 dva_t blk_dva[3]; /* 128-bit Data Virtual Address */ 175 uint64_t blk_prop; /* size, compression, type, etc */ 176 uint64_t blk_pad[3]; /* Extra space for the future */ 177 uint64_t blk_birth; /* transaction group at birth */ 178 uint64_t blk_fill; /* fill count */ 179 zio_cksum_t blk_cksum; /* 256-bit checksum */ 180 } blkptr_t; 181 182 #define SPA_BLKPTRSHIFT 7 /* blkptr_t is 128 bytes */ 183 #define SPA_DVAS_PER_BP 3 /* Number of DVAs in a bp */ 184 185 /* 186 * Macros to get and set fields in a bp or DVA. 187 */ 188 #define DVA_GET_ASIZE(dva) \ 189 BF64_GET_SB((dva)->dva_word[0], 0, 24, SPA_MINBLOCKSHIFT, 0) 190 #define DVA_SET_ASIZE(dva, x) \ 191 BF64_SET_SB((dva)->dva_word[0], 0, 24, SPA_MINBLOCKSHIFT, 0, x) 192 193 #define DVA_GET_GRID(dva) BF64_GET((dva)->dva_word[0], 24, 8) 194 #define DVA_SET_GRID(dva, x) BF64_SET((dva)->dva_word[0], 24, 8, x) 195 196 #define DVA_GET_VDEV(dva) BF64_GET((dva)->dva_word[0], 32, 32) 197 #define DVA_SET_VDEV(dva, x) BF64_SET((dva)->dva_word[0], 32, 32, x) 198 199 #define DVA_GET_OFFSET(dva) \ 200 BF64_GET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0) 201 #define DVA_SET_OFFSET(dva, x) \ 202 BF64_SET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0, x) 203 204 #define DVA_GET_GANG(dva) BF64_GET((dva)->dva_word[1], 63, 1) 205 #define DVA_SET_GANG(dva, x) BF64_SET((dva)->dva_word[1], 63, 1, x) 206 207 #define BP_GET_LSIZE(bp) \ 208 (BP_IS_HOLE(bp) ? 0 : \ 209 BF64_GET_SB((bp)->blk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1)) 210 #define BP_SET_LSIZE(bp, x) \ 211 BF64_SET_SB((bp)->blk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1, x) 212 213 #define BP_GET_PSIZE(bp) \ 214 BF64_GET_SB((bp)->blk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1) 215 #define BP_SET_PSIZE(bp, x) \ 216 BF64_SET_SB((bp)->blk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1, x) 217 218 #define BP_GET_COMPRESS(bp) BF64_GET((bp)->blk_prop, 32, 8) 219 #define BP_SET_COMPRESS(bp, x) BF64_SET((bp)->blk_prop, 32, 8, x) 220 221 #define BP_GET_CHECKSUM(bp) BF64_GET((bp)->blk_prop, 40, 8) 222 #define BP_SET_CHECKSUM(bp, x) BF64_SET((bp)->blk_prop, 40, 8, x) 223 224 #define BP_GET_TYPE(bp) BF64_GET((bp)->blk_prop, 48, 8) 225 #define BP_SET_TYPE(bp, x) BF64_SET((bp)->blk_prop, 48, 8, x) 226 227 #define BP_GET_LEVEL(bp) BF64_GET((bp)->blk_prop, 56, 5) 228 #define BP_SET_LEVEL(bp, x) BF64_SET((bp)->blk_prop, 56, 5, x) 229 230 #define BP_GET_BYTEORDER(bp) (0 - BF64_GET((bp)->blk_prop, 63, 1)) 231 #define BP_SET_BYTEORDER(bp, x) BF64_SET((bp)->blk_prop, 63, 1, x) 232 233 #define BP_GET_ASIZE(bp) \ 234 (DVA_GET_ASIZE(&(bp)->blk_dva[0]) + DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \ 235 DVA_GET_ASIZE(&(bp)->blk_dva[2])) 236 237 #define BP_GET_UCSIZE(bp) \ 238 ((BP_GET_LEVEL(bp) > 0 || dmu_ot[BP_GET_TYPE(bp)].ot_metadata) ? \ 239 BP_GET_PSIZE(bp) : BP_GET_LSIZE(bp)); 240 241 #define BP_GET_NDVAS(bp) \ 242 (!!DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \ 243 !!DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \ 244 !!DVA_GET_ASIZE(&(bp)->blk_dva[2])) 245 246 #define BP_COUNT_GANG(bp) \ 247 (DVA_GET_GANG(&(bp)->blk_dva[0]) + \ 248 DVA_GET_GANG(&(bp)->blk_dva[1]) + \ 249 DVA_GET_GANG(&(bp)->blk_dva[2])) 250 251 #define DVA_EQUAL(dva1, dva2) \ 252 ((dva1)->dva_word[1] == (dva2)->dva_word[1] && \ 253 (dva1)->dva_word[0] == (dva2)->dva_word[0]) 254 255 #define ZIO_CHECKSUM_EQUAL(zc1, zc2) \ 256 (0 == (((zc1).zc_word[0] - (zc2).zc_word[0]) | \ 257 ((zc1).zc_word[1] - (zc2).zc_word[1]) | \ 258 ((zc1).zc_word[2] - (zc2).zc_word[2]) | \ 259 ((zc1).zc_word[3] - (zc2).zc_word[3]))) 260 261 262 #define DVA_IS_VALID(dva) (DVA_GET_ASIZE(dva) != 0) 263 264 #define ZIO_SET_CHECKSUM(zcp, w0, w1, w2, w3) \ 265 { \ 266 (zcp)->zc_word[0] = w0; \ 267 (zcp)->zc_word[1] = w1; \ 268 (zcp)->zc_word[2] = w2; \ 269 (zcp)->zc_word[3] = w3; \ 270 } 271 272 #define BP_IDENTITY(bp) (&(bp)->blk_dva[0]) 273 #define BP_IS_GANG(bp) DVA_GET_GANG(BP_IDENTITY(bp)) 274 #define BP_IS_HOLE(bp) ((bp)->blk_birth == 0) 275 276 #define BP_ZERO(bp) \ 277 { \ 278 (bp)->blk_dva[0].dva_word[0] = 0; \ 279 (bp)->blk_dva[0].dva_word[1] = 0; \ 280 (bp)->blk_dva[1].dva_word[0] = 0; \ 281 (bp)->blk_dva[1].dva_word[1] = 0; \ 282 (bp)->blk_dva[2].dva_word[0] = 0; \ 283 (bp)->blk_dva[2].dva_word[1] = 0; \ 284 (bp)->blk_prop = 0; \ 285 (bp)->blk_pad[0] = 0; \ 286 (bp)->blk_pad[1] = 0; \ 287 (bp)->blk_pad[2] = 0; \ 288 (bp)->blk_birth = 0; \ 289 (bp)->blk_fill = 0; \ 290 ZIO_SET_CHECKSUM(&(bp)->blk_cksum, 0, 0, 0, 0); \ 291 } 292 293 /* 294 * Note: the byteorder is either 0 or -1, both of which are palindromes. 295 * This simplifies the endianness handling a bit. 296 */ 297 #ifdef _BIG_ENDIAN 298 #define ZFS_HOST_BYTEORDER (0ULL) 299 #else 300 #define ZFS_HOST_BYTEORDER (-1ULL) 301 #endif 302 303 #define BP_SHOULD_BYTESWAP(bp) (BP_GET_BYTEORDER(bp) != ZFS_HOST_BYTEORDER) 304 305 #define BP_SPRINTF_LEN 320 306 307 #include <sys/dmu.h> 308 309 /* 310 * Routines found in spa.c 311 */ 312 313 /* state manipulation functions */ 314 extern int spa_open(const char *pool, spa_t **, void *tag); 315 extern int spa_get_stats(const char *pool, nvlist_t **config, 316 char *altroot, size_t buflen); 317 extern int spa_create(const char *pool, nvlist_t *config, const char *altroot); 318 extern int spa_import(const char *pool, nvlist_t *config, const char *altroot); 319 extern nvlist_t *spa_tryimport(nvlist_t *tryconfig); 320 extern int spa_destroy(char *pool); 321 extern int spa_export(char *pool, nvlist_t **oldconfig); 322 extern int spa_reset(char *pool); 323 extern void spa_async_request(spa_t *spa, int flag); 324 extern void spa_async_suspend(spa_t *spa); 325 extern void spa_async_resume(spa_t *spa); 326 extern spa_t *spa_inject_addref(char *pool); 327 extern void spa_inject_delref(spa_t *spa); 328 329 #define SPA_ASYNC_REOPEN 0x01 330 #define SPA_ASYNC_REPLACE_DONE 0x02 331 #define SPA_ASYNC_SCRUB 0x04 332 #define SPA_ASYNC_RESILVER 0x08 333 #define SPA_ASYNC_CONFIG_UPDATE 0x10 334 335 /* device manipulation */ 336 extern int spa_vdev_add(spa_t *spa, nvlist_t *nvroot); 337 extern int spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot, 338 int replacing); 339 extern int spa_vdev_detach(spa_t *spa, uint64_t guid, int replace_done); 340 extern int spa_vdev_remove(spa_t *spa, uint64_t guid, boolean_t unspare); 341 extern int spa_vdev_setpath(spa_t *spa, uint64_t guid, const char *newpath); 342 343 /* spare state (which is global across all pools) */ 344 extern void spa_spare_add(uint64_t guid); 345 extern void spa_spare_remove(uint64_t guid); 346 extern boolean_t spa_spare_inuse(uint64_t guid); 347 348 /* scrubbing */ 349 extern int spa_scrub(spa_t *spa, pool_scrub_type_t type, boolean_t force); 350 extern void spa_scrub_suspend(spa_t *spa); 351 extern void spa_scrub_resume(spa_t *spa); 352 extern void spa_scrub_restart(spa_t *spa, uint64_t txg); 353 extern void spa_scrub_throttle(spa_t *spa, int direction); 354 355 /* spa syncing */ 356 extern void spa_sync(spa_t *spa, uint64_t txg); /* only for DMU use */ 357 extern void spa_sync_allpools(void); 358 359 /* 360 * SPA configuration functions in spa_config.c 361 */ 362 363 #define SPA_CONFIG_UPDATE_POOL 0 364 #define SPA_CONFIG_UPDATE_VDEVS 1 365 366 extern void spa_config_sync(void); 367 extern void spa_config_load(void); 368 extern nvlist_t *spa_all_configs(uint64_t *); 369 extern void spa_config_set(spa_t *spa, nvlist_t *config); 370 extern nvlist_t *spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg, 371 int getstats); 372 extern void spa_config_update(spa_t *spa, int what); 373 374 /* 375 * Miscellaneous SPA routines in spa_misc.c 376 */ 377 378 /* Namespace manipulation */ 379 extern spa_t *spa_lookup(const char *name); 380 extern spa_t *spa_add(const char *name, const char *altroot); 381 extern void spa_remove(spa_t *spa); 382 extern spa_t *spa_next(spa_t *prev); 383 384 /* Refcount functions */ 385 extern void spa_open_ref(spa_t *spa, void *tag); 386 extern void spa_close(spa_t *spa, void *tag); 387 extern boolean_t spa_refcount_zero(spa_t *spa); 388 389 /* Pool configuration lock */ 390 extern void spa_config_enter(spa_t *spa, krw_t rw, void *tag); 391 extern void spa_config_exit(spa_t *spa, void *tag); 392 extern boolean_t spa_config_held(spa_t *spa, krw_t rw); 393 394 /* Pool vdev add/remove lock */ 395 extern uint64_t spa_vdev_enter(spa_t *spa); 396 extern int spa_vdev_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error); 397 398 /* Accessor functions */ 399 extern krwlock_t *spa_traverse_rwlock(spa_t *spa); 400 extern int spa_traverse_wanted(spa_t *spa); 401 extern struct dsl_pool *spa_get_dsl(spa_t *spa); 402 extern blkptr_t *spa_get_rootblkptr(spa_t *spa); 403 extern void spa_set_rootblkptr(spa_t *spa, const blkptr_t *bp); 404 extern void spa_altroot(spa_t *, char *, size_t); 405 extern int spa_sync_pass(spa_t *spa); 406 extern char *spa_name(spa_t *spa); 407 extern uint64_t spa_guid(spa_t *spa); 408 extern uint64_t spa_last_synced_txg(spa_t *spa); 409 extern uint64_t spa_first_txg(spa_t *spa); 410 extern uint64_t spa_version(spa_t *spa); 411 extern int spa_state(spa_t *spa); 412 extern uint64_t spa_freeze_txg(spa_t *spa); 413 struct metaslab_class; 414 extern struct metaslab_class *spa_metaslab_class_select(spa_t *spa); 415 extern uint64_t spa_get_alloc(spa_t *spa); 416 extern uint64_t spa_get_space(spa_t *spa); 417 extern uint64_t spa_get_dspace(spa_t *spa); 418 extern uint64_t spa_get_asize(spa_t *spa, uint64_t lsize); 419 extern uint64_t spa_version(spa_t *spa); 420 extern int spa_max_replication(spa_t *spa); 421 extern int spa_busy(void); 422 423 /* Miscellaneous support routines */ 424 extern int spa_rename(const char *oldname, const char *newname); 425 extern boolean_t spa_guid_exists(uint64_t pool_guid, uint64_t device_guid); 426 extern char *spa_strdup(const char *); 427 extern void spa_strfree(char *); 428 extern uint64_t spa_get_random(uint64_t range); 429 extern void sprintf_blkptr(char *buf, int len, const blkptr_t *bp); 430 extern void spa_freeze(spa_t *spa); 431 extern void spa_upgrade(spa_t *spa); 432 extern void spa_evict_all(void); 433 extern vdev_t *spa_lookup_by_guid(spa_t *spa, uint64_t guid); 434 extern boolean_t spa_has_spare(spa_t *, uint64_t guid); 435 extern uint64_t bp_get_dasize(spa_t *spa, const blkptr_t *bp); 436 437 /* history logging */ 438 extern void spa_history_create_obj(spa_t *spa, dmu_tx_t *tx); 439 extern int spa_history_get(spa_t *spa, uint64_t *offset, uint64_t *len_read, 440 char *his_buf); 441 extern int spa_history_log(spa_t *spa, const char *his_buf, 442 uint64_t pool_create); 443 444 /* error handling */ 445 struct zbookmark; 446 struct zio; 447 extern void spa_log_error(spa_t *spa, struct zio *zio); 448 extern void zfs_ereport_post(const char *class, spa_t *spa, vdev_t *vd, 449 struct zio *zio, uint64_t stateoroffset, uint64_t length); 450 extern void zfs_post_ok(spa_t *spa, vdev_t *vd); 451 extern uint64_t spa_get_errlog_size(spa_t *spa); 452 extern int spa_get_errlog(spa_t *spa, void *uaddr, size_t *count); 453 extern void spa_errlog_rotate(spa_t *spa); 454 extern void spa_errlog_drain(spa_t *spa); 455 extern void spa_errlog_sync(spa_t *spa, uint64_t txg); 456 extern int spa_bookmark_name(spa_t *spa, struct zbookmark *zb, 457 nvlist_t *nvl); 458 extern void spa_get_errlists(spa_t *spa, avl_tree_t *last, avl_tree_t *scrub); 459 460 /* Initialization and termination */ 461 extern void spa_init(int flags); 462 extern void spa_fini(void); 463 464 #ifdef ZFS_DEBUG 465 #define dprintf_bp(bp, fmt, ...) do { \ 466 if (zfs_flags & ZFS_DEBUG_DPRINTF) { \ 467 char __blkbuf[BP_SPRINTF_LEN]; \ 468 sprintf_blkptr(__blkbuf, BP_SPRINTF_LEN, (bp)); \ 469 dprintf(fmt " %s\n", __VA_ARGS__, __blkbuf); \ 470 } \ 471 _NOTE(CONSTCOND) } while (0) 472 #else 473 #define dprintf_bp(bp, fmt, ...) 474 #endif 475 476 extern int spa_mode; /* mode, e.g. FREAD | FWRITE */ 477 478 #ifdef __cplusplus 479 } 480 #endif 481 482 #endif /* _SYS_SPA_H */ 483