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) 2005, 2010, Oracle and/or its affiliates. All rights reserved. 23 * Copyright (c) 2011, 2018 by Delphix. All rights reserved. 24 * Copyright 2011 Nexenta Systems, Inc. All rights reserved. 25 * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved. 26 * Copyright 2013 Saso Kiselkov. All rights reserved. 27 * Copyright (c) 2014 Integros [integros.com] 28 * Copyright 2017 Joyent, Inc. 29 * Copyright (c) 2017 Datto Inc. 30 */ 31 32 #ifndef _SYS_SPA_H 33 #define _SYS_SPA_H 34 35 #include <sys/avl.h> 36 #include <sys/zfs_context.h> 37 #include <sys/nvpair.h> 38 #include <sys/sysevent.h> 39 #include <sys/sysmacros.h> 40 #include <sys/types.h> 41 #include <sys/fs/zfs.h> 42 #include <sys/dmu.h> 43 44 #ifdef __cplusplus 45 extern "C" { 46 #endif 47 48 /* 49 * Forward references that lots of things need. 50 */ 51 typedef struct spa spa_t; 52 typedef struct vdev vdev_t; 53 typedef struct metaslab metaslab_t; 54 typedef struct metaslab_group metaslab_group_t; 55 typedef struct metaslab_class metaslab_class_t; 56 typedef struct zio zio_t; 57 typedef struct zilog zilog_t; 58 typedef struct spa_aux_vdev spa_aux_vdev_t; 59 typedef struct ddt ddt_t; 60 typedef struct ddt_entry ddt_entry_t; 61 struct dsl_pool; 62 struct dsl_dataset; 63 64 /* 65 * General-purpose 32-bit and 64-bit bitfield encodings. 66 */ 67 #define BF32_DECODE(x, low, len) P2PHASE((x) >> (low), 1U << (len)) 68 #define BF64_DECODE(x, low, len) P2PHASE((x) >> (low), 1ULL << (len)) 69 #define BF32_ENCODE(x, low, len) (P2PHASE((x), 1U << (len)) << (low)) 70 #define BF64_ENCODE(x, low, len) (P2PHASE((x), 1ULL << (len)) << (low)) 71 72 #define BF32_GET(x, low, len) BF32_DECODE(x, low, len) 73 #define BF64_GET(x, low, len) BF64_DECODE(x, low, len) 74 75 #define BF32_SET(x, low, len, val) do { \ 76 ASSERT3U(val, <, 1U << (len)); \ 77 ASSERT3U(low + len, <=, 32); \ 78 (x) ^= BF32_ENCODE((x >> low) ^ (val), low, len); \ 79 _NOTE(CONSTCOND) } while (0) 80 81 #define BF64_SET(x, low, len, val) do { \ 82 ASSERT3U(val, <, 1ULL << (len)); \ 83 ASSERT3U(low + len, <=, 64); \ 84 ((x) ^= BF64_ENCODE((x >> low) ^ (val), low, len)); \ 85 _NOTE(CONSTCOND) } while (0) 86 87 #define BF32_GET_SB(x, low, len, shift, bias) \ 88 ((BF32_GET(x, low, len) + (bias)) << (shift)) 89 #define BF64_GET_SB(x, low, len, shift, bias) \ 90 ((BF64_GET(x, low, len) + (bias)) << (shift)) 91 92 #define BF32_SET_SB(x, low, len, shift, bias, val) do { \ 93 ASSERT(IS_P2ALIGNED(val, 1U << shift)); \ 94 ASSERT3S((val) >> (shift), >=, bias); \ 95 BF32_SET(x, low, len, ((val) >> (shift)) - (bias)); \ 96 _NOTE(CONSTCOND) } while (0) 97 #define BF64_SET_SB(x, low, len, shift, bias, val) do { \ 98 ASSERT(IS_P2ALIGNED(val, 1ULL << shift)); \ 99 ASSERT3S((val) >> (shift), >=, bias); \ 100 BF64_SET(x, low, len, ((val) >> (shift)) - (bias)); \ 101 _NOTE(CONSTCOND) } while (0) 102 103 /* 104 * We currently support block sizes from 512 bytes to 16MB. 105 * The benefits of larger blocks, and thus larger IO, need to be weighed 106 * against the cost of COWing a giant block to modify one byte, and the 107 * large latency of reading or writing a large block. 108 * 109 * Note that although blocks up to 16MB are supported, the recordsize 110 * property can not be set larger than zfs_max_recordsize (default 1MB). 111 * See the comment near zfs_max_recordsize in dsl_dataset.c for details. 112 * 113 * Note that although the LSIZE field of the blkptr_t can store sizes up 114 * to 32MB, the dnode's dn_datablkszsec can only store sizes up to 115 * 32MB - 512 bytes. Therefore, we limit SPA_MAXBLOCKSIZE to 16MB. 116 */ 117 #define SPA_MINBLOCKSHIFT 9 118 #define SPA_OLD_MAXBLOCKSHIFT 17 119 #define SPA_MAXBLOCKSHIFT 24 120 #define SPA_MINBLOCKSIZE (1ULL << SPA_MINBLOCKSHIFT) 121 #define SPA_OLD_MAXBLOCKSIZE (1ULL << SPA_OLD_MAXBLOCKSHIFT) 122 #define SPA_MAXBLOCKSIZE (1ULL << SPA_MAXBLOCKSHIFT) 123 124 /* 125 * Default maximum supported logical ashift. 126 * 127 * The current 8k allocation block size limit is due to the 8k 128 * aligned/sized operations performed by vdev_probe() on 129 * vdev_label->vl_pad2. Using another "safe region" for these tests 130 * would allow the limit to be raised to 16k, at the expense of 131 * only having 8 available uberblocks in the label area. 132 */ 133 #define SPA_MAXASHIFT 13 134 135 /* 136 * Default minimum supported logical ashift. 137 */ 138 #define SPA_MINASHIFT SPA_MINBLOCKSHIFT 139 140 /* 141 * Size of block to hold the configuration data (a packed nvlist) 142 */ 143 #define SPA_CONFIG_BLOCKSIZE (1ULL << 14) 144 145 /* 146 * The DVA size encodings for LSIZE and PSIZE support blocks up to 32MB. 147 * The ASIZE encoding should be at least 64 times larger (6 more bits) 148 * to support up to 4-way RAID-Z mirror mode with worst-case gang block 149 * overhead, three DVAs per bp, plus one more bit in case we do anything 150 * else that expands the ASIZE. 151 */ 152 #define SPA_LSIZEBITS 16 /* LSIZE up to 32M (2^16 * 512) */ 153 #define SPA_PSIZEBITS 16 /* PSIZE up to 32M (2^16 * 512) */ 154 #define SPA_ASIZEBITS 24 /* ASIZE up to 64 times larger */ 155 156 #define SPA_COMPRESSBITS 7 157 #define SPA_VDEVBITS 24 158 159 /* 160 * All SPA data is represented by 128-bit data virtual addresses (DVAs). 161 * The members of the dva_t should be considered opaque outside the SPA. 162 */ 163 typedef struct dva { 164 uint64_t dva_word[2]; 165 } dva_t; 166 167 /* 168 * Each block has a 256-bit checksum -- strong enough for cryptographic hashes. 169 */ 170 typedef struct zio_cksum { 171 uint64_t zc_word[4]; 172 } zio_cksum_t; 173 174 /* 175 * Some checksums/hashes need a 256-bit initialization salt. This salt is kept 176 * secret and is suitable for use in MAC algorithms as the key. 177 */ 178 typedef struct zio_cksum_salt { 179 uint8_t zcs_bytes[32]; 180 } zio_cksum_salt_t; 181 182 /* 183 * Each block is described by its DVAs, time of birth, checksum, etc. 184 * The word-by-word, bit-by-bit layout of the blkptr is as follows: 185 * 186 * 64 56 48 40 32 24 16 8 0 187 * +-------+-------+-------+-------+-------+-------+-------+-------+ 188 * 0 | pad | vdev1 | GRID | ASIZE | 189 * +-------+-------+-------+-------+-------+-------+-------+-------+ 190 * 1 |G| offset1 | 191 * +-------+-------+-------+-------+-------+-------+-------+-------+ 192 * 2 | pad | vdev2 | GRID | ASIZE | 193 * +-------+-------+-------+-------+-------+-------+-------+-------+ 194 * 3 |G| offset2 | 195 * +-------+-------+-------+-------+-------+-------+-------+-------+ 196 * 4 | pad | vdev3 | GRID | ASIZE | 197 * +-------+-------+-------+-------+-------+-------+-------+-------+ 198 * 5 |G| offset3 | 199 * +-------+-------+-------+-------+-------+-------+-------+-------+ 200 * 6 |BDX|lvl| type | cksum |E| comp| PSIZE | LSIZE | 201 * +-------+-------+-------+-------+-------+-------+-------+-------+ 202 * 7 | padding | 203 * +-------+-------+-------+-------+-------+-------+-------+-------+ 204 * 8 | padding | 205 * +-------+-------+-------+-------+-------+-------+-------+-------+ 206 * 9 | physical birth txg | 207 * +-------+-------+-------+-------+-------+-------+-------+-------+ 208 * a | logical birth txg | 209 * +-------+-------+-------+-------+-------+-------+-------+-------+ 210 * b | fill count | 211 * +-------+-------+-------+-------+-------+-------+-------+-------+ 212 * c | checksum[0] | 213 * +-------+-------+-------+-------+-------+-------+-------+-------+ 214 * d | checksum[1] | 215 * +-------+-------+-------+-------+-------+-------+-------+-------+ 216 * e | checksum[2] | 217 * +-------+-------+-------+-------+-------+-------+-------+-------+ 218 * f | checksum[3] | 219 * +-------+-------+-------+-------+-------+-------+-------+-------+ 220 * 221 * Legend: 222 * 223 * vdev virtual device ID 224 * offset offset into virtual device 225 * LSIZE logical size 226 * PSIZE physical size (after compression) 227 * ASIZE allocated size (including RAID-Z parity and gang block headers) 228 * GRID RAID-Z layout information (reserved for future use) 229 * cksum checksum function 230 * comp compression function 231 * G gang block indicator 232 * B byteorder (endianness) 233 * D dedup 234 * X encryption (on version 30, which is not supported) 235 * E blkptr_t contains embedded data (see below) 236 * lvl level of indirection 237 * type DMU object type 238 * phys birth txg when dva[0] was written; zero if same as logical birth txg 239 * note that typically all the dva's would be written in this 240 * txg, but they could be different if they were moved by 241 * device removal. 242 * log. birth transaction group in which the block was logically born 243 * fill count number of non-zero blocks under this bp 244 * checksum[4] 256-bit checksum of the data this bp describes 245 */ 246 247 /* 248 * "Embedded" blkptr_t's don't actually point to a block, instead they 249 * have a data payload embedded in the blkptr_t itself. See the comment 250 * in blkptr.c for more details. 251 * 252 * The blkptr_t is laid out as follows: 253 * 254 * 64 56 48 40 32 24 16 8 0 255 * +-------+-------+-------+-------+-------+-------+-------+-------+ 256 * 0 | payload | 257 * 1 | payload | 258 * 2 | payload | 259 * 3 | payload | 260 * 4 | payload | 261 * 5 | payload | 262 * +-------+-------+-------+-------+-------+-------+-------+-------+ 263 * 6 |BDX|lvl| type | etype |E| comp| PSIZE| LSIZE | 264 * +-------+-------+-------+-------+-------+-------+-------+-------+ 265 * 7 | payload | 266 * 8 | payload | 267 * 9 | payload | 268 * +-------+-------+-------+-------+-------+-------+-------+-------+ 269 * a | logical birth txg | 270 * +-------+-------+-------+-------+-------+-------+-------+-------+ 271 * b | payload | 272 * c | payload | 273 * d | payload | 274 * e | payload | 275 * f | payload | 276 * +-------+-------+-------+-------+-------+-------+-------+-------+ 277 * 278 * Legend: 279 * 280 * payload contains the embedded data 281 * B (byteorder) byteorder (endianness) 282 * D (dedup) padding (set to zero) 283 * X encryption (set to zero; see above) 284 * E (embedded) set to one 285 * lvl indirection level 286 * type DMU object type 287 * etype how to interpret embedded data (BP_EMBEDDED_TYPE_*) 288 * comp compression function of payload 289 * PSIZE size of payload after compression, in bytes 290 * LSIZE logical size of payload, in bytes 291 * note that 25 bits is enough to store the largest 292 * "normal" BP's LSIZE (2^16 * 2^9) in bytes 293 * log. birth transaction group in which the block was logically born 294 * 295 * Note that LSIZE and PSIZE are stored in bytes, whereas for non-embedded 296 * bp's they are stored in units of SPA_MINBLOCKSHIFT. 297 * Generally, the generic BP_GET_*() macros can be used on embedded BP's. 298 * The B, D, X, lvl, type, and comp fields are stored the same as with normal 299 * BP's so the BP_SET_* macros can be used with them. etype, PSIZE, LSIZE must 300 * be set with the BPE_SET_* macros. BP_SET_EMBEDDED() should be called before 301 * other macros, as they assert that they are only used on BP's of the correct 302 * "embedded-ness". 303 */ 304 305 #define BPE_GET_ETYPE(bp) \ 306 (ASSERT(BP_IS_EMBEDDED(bp)), \ 307 BF64_GET((bp)->blk_prop, 40, 8)) 308 #define BPE_SET_ETYPE(bp, t) do { \ 309 ASSERT(BP_IS_EMBEDDED(bp)); \ 310 BF64_SET((bp)->blk_prop, 40, 8, t); \ 311 _NOTE(CONSTCOND) } while (0) 312 313 #define BPE_GET_LSIZE(bp) \ 314 (ASSERT(BP_IS_EMBEDDED(bp)), \ 315 BF64_GET_SB((bp)->blk_prop, 0, 25, 0, 1)) 316 #define BPE_SET_LSIZE(bp, x) do { \ 317 ASSERT(BP_IS_EMBEDDED(bp)); \ 318 BF64_SET_SB((bp)->blk_prop, 0, 25, 0, 1, x); \ 319 _NOTE(CONSTCOND) } while (0) 320 321 #define BPE_GET_PSIZE(bp) \ 322 (ASSERT(BP_IS_EMBEDDED(bp)), \ 323 BF64_GET_SB((bp)->blk_prop, 25, 7, 0, 1)) 324 #define BPE_SET_PSIZE(bp, x) do { \ 325 ASSERT(BP_IS_EMBEDDED(bp)); \ 326 BF64_SET_SB((bp)->blk_prop, 25, 7, 0, 1, x); \ 327 _NOTE(CONSTCOND) } while (0) 328 329 typedef enum bp_embedded_type { 330 BP_EMBEDDED_TYPE_DATA, 331 BP_EMBEDDED_TYPE_RESERVED, /* Reserved for an unintegrated feature. */ 332 NUM_BP_EMBEDDED_TYPES = BP_EMBEDDED_TYPE_RESERVED 333 } bp_embedded_type_t; 334 335 #define BPE_NUM_WORDS 14 336 #define BPE_PAYLOAD_SIZE (BPE_NUM_WORDS * sizeof (uint64_t)) 337 #define BPE_IS_PAYLOADWORD(bp, wp) \ 338 ((wp) != &(bp)->blk_prop && (wp) != &(bp)->blk_birth) 339 340 #define SPA_BLKPTRSHIFT 7 /* blkptr_t is 128 bytes */ 341 #define SPA_DVAS_PER_BP 3 /* Number of DVAs in a bp */ 342 #define SPA_SYNC_MIN_VDEVS 3 /* min vdevs to update during sync */ 343 344 /* 345 * A block is a hole when it has either 1) never been written to, or 346 * 2) is zero-filled. In both cases, ZFS can return all zeroes for all reads 347 * without physically allocating disk space. Holes are represented in the 348 * blkptr_t structure by zeroed blk_dva. Correct checking for holes is 349 * done through the BP_IS_HOLE macro. For holes, the logical size, level, 350 * DMU object type, and birth times are all also stored for holes that 351 * were written to at some point (i.e. were punched after having been filled). 352 */ 353 typedef struct blkptr { 354 dva_t blk_dva[SPA_DVAS_PER_BP]; /* Data Virtual Addresses */ 355 uint64_t blk_prop; /* size, compression, type, etc */ 356 uint64_t blk_pad[2]; /* Extra space for the future */ 357 uint64_t blk_phys_birth; /* txg when block was allocated */ 358 uint64_t blk_birth; /* transaction group at birth */ 359 uint64_t blk_fill; /* fill count */ 360 zio_cksum_t blk_cksum; /* 256-bit checksum */ 361 } blkptr_t; 362 363 /* 364 * Macros to get and set fields in a bp or DVA. 365 */ 366 #define DVA_GET_ASIZE(dva) \ 367 BF64_GET_SB((dva)->dva_word[0], 0, SPA_ASIZEBITS, SPA_MINBLOCKSHIFT, 0) 368 #define DVA_SET_ASIZE(dva, x) \ 369 BF64_SET_SB((dva)->dva_word[0], 0, SPA_ASIZEBITS, \ 370 SPA_MINBLOCKSHIFT, 0, x) 371 372 #define DVA_GET_GRID(dva) BF64_GET((dva)->dva_word[0], 24, 8) 373 #define DVA_SET_GRID(dva, x) BF64_SET((dva)->dva_word[0], 24, 8, x) 374 375 #define DVA_GET_VDEV(dva) BF64_GET((dva)->dva_word[0], 32, SPA_VDEVBITS) 376 #define DVA_SET_VDEV(dva, x) \ 377 BF64_SET((dva)->dva_word[0], 32, SPA_VDEVBITS, x) 378 379 #define DVA_GET_OFFSET(dva) \ 380 BF64_GET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0) 381 #define DVA_SET_OFFSET(dva, x) \ 382 BF64_SET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0, x) 383 384 #define DVA_GET_GANG(dva) BF64_GET((dva)->dva_word[1], 63, 1) 385 #define DVA_SET_GANG(dva, x) BF64_SET((dva)->dva_word[1], 63, 1, x) 386 387 #define BP_GET_LSIZE(bp) \ 388 (BP_IS_EMBEDDED(bp) ? \ 389 (BPE_GET_ETYPE(bp) == BP_EMBEDDED_TYPE_DATA ? BPE_GET_LSIZE(bp) : 0): \ 390 BF64_GET_SB((bp)->blk_prop, 0, SPA_LSIZEBITS, SPA_MINBLOCKSHIFT, 1)) 391 #define BP_SET_LSIZE(bp, x) do { \ 392 ASSERT(!BP_IS_EMBEDDED(bp)); \ 393 BF64_SET_SB((bp)->blk_prop, \ 394 0, SPA_LSIZEBITS, SPA_MINBLOCKSHIFT, 1, x); \ 395 _NOTE(CONSTCOND) } while (0) 396 397 #define BP_GET_PSIZE(bp) \ 398 (BP_IS_EMBEDDED(bp) ? 0 : \ 399 BF64_GET_SB((bp)->blk_prop, 16, SPA_PSIZEBITS, SPA_MINBLOCKSHIFT, 1)) 400 #define BP_SET_PSIZE(bp, x) do { \ 401 ASSERT(!BP_IS_EMBEDDED(bp)); \ 402 BF64_SET_SB((bp)->blk_prop, \ 403 16, SPA_PSIZEBITS, SPA_MINBLOCKSHIFT, 1, x); \ 404 _NOTE(CONSTCOND) } while (0) 405 406 #define BP_GET_COMPRESS(bp) \ 407 BF64_GET((bp)->blk_prop, 32, SPA_COMPRESSBITS) 408 #define BP_SET_COMPRESS(bp, x) \ 409 BF64_SET((bp)->blk_prop, 32, SPA_COMPRESSBITS, x) 410 411 #define BP_IS_EMBEDDED(bp) BF64_GET((bp)->blk_prop, 39, 1) 412 #define BP_SET_EMBEDDED(bp, x) BF64_SET((bp)->blk_prop, 39, 1, x) 413 414 #define BP_GET_CHECKSUM(bp) \ 415 (BP_IS_EMBEDDED(bp) ? ZIO_CHECKSUM_OFF : \ 416 BF64_GET((bp)->blk_prop, 40, 8)) 417 #define BP_SET_CHECKSUM(bp, x) do { \ 418 ASSERT(!BP_IS_EMBEDDED(bp)); \ 419 BF64_SET((bp)->blk_prop, 40, 8, x); \ 420 _NOTE(CONSTCOND) } while (0) 421 422 #define BP_GET_TYPE(bp) BF64_GET((bp)->blk_prop, 48, 8) 423 #define BP_SET_TYPE(bp, x) BF64_SET((bp)->blk_prop, 48, 8, x) 424 425 #define BP_GET_LEVEL(bp) BF64_GET((bp)->blk_prop, 56, 5) 426 #define BP_SET_LEVEL(bp, x) BF64_SET((bp)->blk_prop, 56, 5, x) 427 428 #define BP_GET_DEDUP(bp) BF64_GET((bp)->blk_prop, 62, 1) 429 #define BP_SET_DEDUP(bp, x) BF64_SET((bp)->blk_prop, 62, 1, x) 430 431 #define BP_GET_BYTEORDER(bp) BF64_GET((bp)->blk_prop, 63, 1) 432 #define BP_SET_BYTEORDER(bp, x) BF64_SET((bp)->blk_prop, 63, 1, x) 433 434 #define BP_PHYSICAL_BIRTH(bp) \ 435 (BP_IS_EMBEDDED(bp) ? 0 : \ 436 (bp)->blk_phys_birth ? (bp)->blk_phys_birth : (bp)->blk_birth) 437 438 #define BP_SET_BIRTH(bp, logical, physical) \ 439 { \ 440 ASSERT(!BP_IS_EMBEDDED(bp)); \ 441 (bp)->blk_birth = (logical); \ 442 (bp)->blk_phys_birth = ((logical) == (physical) ? 0 : (physical)); \ 443 } 444 445 #define BP_GET_FILL(bp) (BP_IS_EMBEDDED(bp) ? 1 : (bp)->blk_fill) 446 447 #define BP_IS_METADATA(bp) \ 448 (BP_GET_LEVEL(bp) > 0 || DMU_OT_IS_METADATA(BP_GET_TYPE(bp))) 449 450 #define BP_GET_ASIZE(bp) \ 451 (BP_IS_EMBEDDED(bp) ? 0 : \ 452 DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \ 453 DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \ 454 DVA_GET_ASIZE(&(bp)->blk_dva[2])) 455 456 #define BP_GET_UCSIZE(bp) \ 457 (BP_IS_METADATA(bp) ? BP_GET_PSIZE(bp) : BP_GET_LSIZE(bp)) 458 459 #define BP_GET_NDVAS(bp) \ 460 (BP_IS_EMBEDDED(bp) ? 0 : \ 461 !!DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \ 462 !!DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \ 463 !!DVA_GET_ASIZE(&(bp)->blk_dva[2])) 464 465 #define BP_COUNT_GANG(bp) \ 466 (BP_IS_EMBEDDED(bp) ? 0 : \ 467 (DVA_GET_GANG(&(bp)->blk_dva[0]) + \ 468 DVA_GET_GANG(&(bp)->blk_dva[1]) + \ 469 DVA_GET_GANG(&(bp)->blk_dva[2]))) 470 471 #define DVA_EQUAL(dva1, dva2) \ 472 ((dva1)->dva_word[1] == (dva2)->dva_word[1] && \ 473 (dva1)->dva_word[0] == (dva2)->dva_word[0]) 474 475 #define BP_EQUAL(bp1, bp2) \ 476 (BP_PHYSICAL_BIRTH(bp1) == BP_PHYSICAL_BIRTH(bp2) && \ 477 (bp1)->blk_birth == (bp2)->blk_birth && \ 478 DVA_EQUAL(&(bp1)->blk_dva[0], &(bp2)->blk_dva[0]) && \ 479 DVA_EQUAL(&(bp1)->blk_dva[1], &(bp2)->blk_dva[1]) && \ 480 DVA_EQUAL(&(bp1)->blk_dva[2], &(bp2)->blk_dva[2])) 481 482 #define ZIO_CHECKSUM_EQUAL(zc1, zc2) \ 483 (0 == (((zc1).zc_word[0] - (zc2).zc_word[0]) | \ 484 ((zc1).zc_word[1] - (zc2).zc_word[1]) | \ 485 ((zc1).zc_word[2] - (zc2).zc_word[2]) | \ 486 ((zc1).zc_word[3] - (zc2).zc_word[3]))) 487 488 #define ZIO_CHECKSUM_IS_ZERO(zc) \ 489 (0 == ((zc)->zc_word[0] | (zc)->zc_word[1] | \ 490 (zc)->zc_word[2] | (zc)->zc_word[3])) 491 492 #define ZIO_CHECKSUM_BSWAP(zcp) \ 493 { \ 494 (zcp)->zc_word[0] = BSWAP_64((zcp)->zc_word[0]); \ 495 (zcp)->zc_word[1] = BSWAP_64((zcp)->zc_word[1]); \ 496 (zcp)->zc_word[2] = BSWAP_64((zcp)->zc_word[2]); \ 497 (zcp)->zc_word[3] = BSWAP_64((zcp)->zc_word[3]); \ 498 } 499 500 501 #define DVA_IS_VALID(dva) (DVA_GET_ASIZE(dva) != 0) 502 503 #define ZIO_SET_CHECKSUM(zcp, w0, w1, w2, w3) \ 504 { \ 505 (zcp)->zc_word[0] = w0; \ 506 (zcp)->zc_word[1] = w1; \ 507 (zcp)->zc_word[2] = w2; \ 508 (zcp)->zc_word[3] = w3; \ 509 } 510 511 #define BP_IDENTITY(bp) (ASSERT(!BP_IS_EMBEDDED(bp)), &(bp)->blk_dva[0]) 512 #define BP_IS_GANG(bp) \ 513 (BP_IS_EMBEDDED(bp) ? B_FALSE : DVA_GET_GANG(BP_IDENTITY(bp))) 514 #define DVA_IS_EMPTY(dva) ((dva)->dva_word[0] == 0ULL && \ 515 (dva)->dva_word[1] == 0ULL) 516 #define BP_IS_HOLE(bp) \ 517 (!BP_IS_EMBEDDED(bp) && DVA_IS_EMPTY(BP_IDENTITY(bp))) 518 519 /* BP_IS_RAIDZ(bp) assumes no block compression */ 520 #define BP_IS_RAIDZ(bp) (DVA_GET_ASIZE(&(bp)->blk_dva[0]) > \ 521 BP_GET_PSIZE(bp)) 522 523 #define BP_ZERO(bp) \ 524 { \ 525 (bp)->blk_dva[0].dva_word[0] = 0; \ 526 (bp)->blk_dva[0].dva_word[1] = 0; \ 527 (bp)->blk_dva[1].dva_word[0] = 0; \ 528 (bp)->blk_dva[1].dva_word[1] = 0; \ 529 (bp)->blk_dva[2].dva_word[0] = 0; \ 530 (bp)->blk_dva[2].dva_word[1] = 0; \ 531 (bp)->blk_prop = 0; \ 532 (bp)->blk_pad[0] = 0; \ 533 (bp)->blk_pad[1] = 0; \ 534 (bp)->blk_phys_birth = 0; \ 535 (bp)->blk_birth = 0; \ 536 (bp)->blk_fill = 0; \ 537 ZIO_SET_CHECKSUM(&(bp)->blk_cksum, 0, 0, 0, 0); \ 538 } 539 540 #if BYTE_ORDER == _BIG_ENDIAN 541 #define ZFS_HOST_BYTEORDER (0ULL) 542 #else 543 #define ZFS_HOST_BYTEORDER (1ULL) 544 #endif 545 546 #define BP_SHOULD_BYTESWAP(bp) (BP_GET_BYTEORDER(bp) != ZFS_HOST_BYTEORDER) 547 548 #define BP_SPRINTF_LEN 320 549 550 /* 551 * This macro allows code sharing between zfs, libzpool, and mdb. 552 * 'func' is either snprintf() or mdb_snprintf(). 553 * 'ws' (whitespace) can be ' ' for single-line format, '\n' for multi-line. 554 */ 555 #define SNPRINTF_BLKPTR(func, ws, buf, size, bp, type, checksum, compress) \ 556 { \ 557 static const char *copyname[] = \ 558 { "zero", "single", "double", "triple" }; \ 559 int len = 0; \ 560 int copies = 0; \ 561 \ 562 if (bp == NULL) { \ 563 len += func(buf + len, size - len, "<NULL>"); \ 564 } else if (BP_IS_HOLE(bp)) { \ 565 len += func(buf + len, size - len, \ 566 "HOLE [L%llu %s] " \ 567 "size=%llxL birth=%lluL", \ 568 (u_longlong_t)BP_GET_LEVEL(bp), \ 569 type, \ 570 (u_longlong_t)BP_GET_LSIZE(bp), \ 571 (u_longlong_t)bp->blk_birth); \ 572 } else if (BP_IS_EMBEDDED(bp)) { \ 573 len = func(buf + len, size - len, \ 574 "EMBEDDED [L%llu %s] et=%u %s " \ 575 "size=%llxL/%llxP birth=%lluL", \ 576 (u_longlong_t)BP_GET_LEVEL(bp), \ 577 type, \ 578 (int)BPE_GET_ETYPE(bp), \ 579 compress, \ 580 (u_longlong_t)BPE_GET_LSIZE(bp), \ 581 (u_longlong_t)BPE_GET_PSIZE(bp), \ 582 (u_longlong_t)bp->blk_birth); \ 583 } else { \ 584 for (int d = 0; d < BP_GET_NDVAS(bp); d++) { \ 585 const dva_t *dva = &bp->blk_dva[d]; \ 586 if (DVA_IS_VALID(dva)) \ 587 copies++; \ 588 len += func(buf + len, size - len, \ 589 "DVA[%d]=<%llu:%llx:%llx>%c", d, \ 590 (u_longlong_t)DVA_GET_VDEV(dva), \ 591 (u_longlong_t)DVA_GET_OFFSET(dva), \ 592 (u_longlong_t)DVA_GET_ASIZE(dva), \ 593 ws); \ 594 } \ 595 if (BP_IS_GANG(bp) && \ 596 DVA_GET_ASIZE(&bp->blk_dva[2]) <= \ 597 DVA_GET_ASIZE(&bp->blk_dva[1]) / 2) \ 598 copies--; \ 599 len += func(buf + len, size - len, \ 600 "[L%llu %s] %s %s %s %s %s %s%c" \ 601 "size=%llxL/%llxP birth=%lluL/%lluP fill=%llu%c" \ 602 "cksum=%llx:%llx:%llx:%llx", \ 603 (u_longlong_t)BP_GET_LEVEL(bp), \ 604 type, \ 605 checksum, \ 606 compress, \ 607 BP_GET_BYTEORDER(bp) == 0 ? "BE" : "LE", \ 608 BP_IS_GANG(bp) ? "gang" : "contiguous", \ 609 BP_GET_DEDUP(bp) ? "dedup" : "unique", \ 610 copyname[copies], \ 611 ws, \ 612 (u_longlong_t)BP_GET_LSIZE(bp), \ 613 (u_longlong_t)BP_GET_PSIZE(bp), \ 614 (u_longlong_t)bp->blk_birth, \ 615 (u_longlong_t)BP_PHYSICAL_BIRTH(bp), \ 616 (u_longlong_t)BP_GET_FILL(bp), \ 617 ws, \ 618 (u_longlong_t)bp->blk_cksum.zc_word[0], \ 619 (u_longlong_t)bp->blk_cksum.zc_word[1], \ 620 (u_longlong_t)bp->blk_cksum.zc_word[2], \ 621 (u_longlong_t)bp->blk_cksum.zc_word[3]); \ 622 } \ 623 ASSERT(len < size); \ 624 } 625 626 #define BP_GET_BUFC_TYPE(bp) \ 627 (BP_IS_METADATA(bp) ? ARC_BUFC_METADATA : ARC_BUFC_DATA) 628 629 typedef enum spa_import_type { 630 SPA_IMPORT_EXISTING, 631 SPA_IMPORT_ASSEMBLE 632 } spa_import_type_t; 633 634 /* state manipulation functions */ 635 extern int spa_open(const char *pool, spa_t **, void *tag); 636 extern int spa_open_rewind(const char *pool, spa_t **, void *tag, 637 nvlist_t *policy, nvlist_t **config); 638 extern int spa_get_stats(const char *pool, nvlist_t **config, char *altroot, 639 size_t buflen); 640 extern int spa_create(const char *pool, nvlist_t *config, nvlist_t *props, 641 nvlist_t *zplprops); 642 #ifdef illumos 643 extern int spa_import_rootpool(char *devpath, char *devid); 644 #else 645 extern int spa_import_rootpool(const char *name); 646 #endif 647 extern int spa_import(const char *pool, nvlist_t *config, nvlist_t *props, 648 uint64_t flags); 649 extern nvlist_t *spa_tryimport(nvlist_t *tryconfig); 650 extern int spa_destroy(char *pool); 651 extern int spa_checkpoint(const char *pool); 652 extern int spa_checkpoint_discard(const char *pool); 653 extern int spa_export(char *pool, nvlist_t **oldconfig, boolean_t force, 654 boolean_t hardforce); 655 extern int spa_reset(char *pool); 656 extern void spa_async_request(spa_t *spa, int flag); 657 extern void spa_async_unrequest(spa_t *spa, int flag); 658 extern void spa_async_suspend(spa_t *spa); 659 extern void spa_async_resume(spa_t *spa); 660 extern spa_t *spa_inject_addref(char *pool); 661 extern void spa_inject_delref(spa_t *spa); 662 extern void spa_scan_stat_init(spa_t *spa); 663 extern int spa_scan_get_stats(spa_t *spa, pool_scan_stat_t *ps); 664 665 #define SPA_ASYNC_CONFIG_UPDATE 0x01 666 #define SPA_ASYNC_REMOVE 0x02 667 #define SPA_ASYNC_PROBE 0x04 668 #define SPA_ASYNC_RESILVER_DONE 0x08 669 #define SPA_ASYNC_RESILVER 0x10 670 #define SPA_ASYNC_AUTOEXPAND 0x20 671 #define SPA_ASYNC_REMOVE_DONE 0x40 672 #define SPA_ASYNC_REMOVE_STOP 0x80 673 #define SPA_ASYNC_INITIALIZE_RESTART 0x100 674 675 /* 676 * Controls the behavior of spa_vdev_remove(). 677 */ 678 #define SPA_REMOVE_UNSPARE 0x01 679 #define SPA_REMOVE_DONE 0x02 680 681 /* device manipulation */ 682 extern int spa_vdev_add(spa_t *spa, nvlist_t *nvroot); 683 extern int spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot, 684 int replacing); 685 extern int spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid, 686 int replace_done); 687 extern int spa_vdev_remove(spa_t *spa, uint64_t guid, boolean_t unspare); 688 extern boolean_t spa_vdev_remove_active(spa_t *spa); 689 extern int spa_vdev_initialize(spa_t *spa, uint64_t guid, uint64_t cmd_type); 690 extern int spa_vdev_setpath(spa_t *spa, uint64_t guid, const char *newpath); 691 extern int spa_vdev_setfru(spa_t *spa, uint64_t guid, const char *newfru); 692 extern int spa_vdev_split_mirror(spa_t *spa, char *newname, nvlist_t *config, 693 nvlist_t *props, boolean_t exp); 694 695 /* spare state (which is global across all pools) */ 696 extern void spa_spare_add(vdev_t *vd); 697 extern void spa_spare_remove(vdev_t *vd); 698 extern boolean_t spa_spare_exists(uint64_t guid, uint64_t *pool, int *refcnt); 699 extern void spa_spare_activate(vdev_t *vd); 700 701 /* L2ARC state (which is global across all pools) */ 702 extern void spa_l2cache_add(vdev_t *vd); 703 extern void spa_l2cache_remove(vdev_t *vd); 704 extern boolean_t spa_l2cache_exists(uint64_t guid, uint64_t *pool); 705 extern void spa_l2cache_activate(vdev_t *vd); 706 extern void spa_l2cache_drop(spa_t *spa); 707 708 /* scanning */ 709 extern int spa_scan(spa_t *spa, pool_scan_func_t func); 710 extern int spa_scan_stop(spa_t *spa); 711 extern int spa_scrub_pause_resume(spa_t *spa, pool_scrub_cmd_t flag); 712 713 /* spa syncing */ 714 extern void spa_sync(spa_t *spa, uint64_t txg); /* only for DMU use */ 715 extern void spa_sync_allpools(void); 716 717 /* spa namespace global mutex */ 718 extern kmutex_t spa_namespace_lock; 719 720 /* 721 * SPA configuration functions in spa_config.c 722 */ 723 724 #define SPA_CONFIG_UPDATE_POOL 0 725 #define SPA_CONFIG_UPDATE_VDEVS 1 726 727 extern void spa_write_cachefile(spa_t *, boolean_t, boolean_t); 728 extern void spa_config_load(void); 729 extern nvlist_t *spa_all_configs(uint64_t *); 730 extern void spa_config_set(spa_t *spa, nvlist_t *config); 731 extern nvlist_t *spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg, 732 int getstats); 733 extern void spa_config_update(spa_t *spa, int what); 734 735 /* 736 * Miscellaneous SPA routines in spa_misc.c 737 */ 738 739 /* Namespace manipulation */ 740 extern spa_t *spa_lookup(const char *name); 741 extern spa_t *spa_add(const char *name, nvlist_t *config, const char *altroot); 742 extern void spa_remove(spa_t *spa); 743 extern spa_t *spa_next(spa_t *prev); 744 745 /* Refcount functions */ 746 extern void spa_open_ref(spa_t *spa, void *tag); 747 extern void spa_close(spa_t *spa, void *tag); 748 extern void spa_async_close(spa_t *spa, void *tag); 749 extern boolean_t spa_refcount_zero(spa_t *spa); 750 751 #define SCL_NONE 0x00 752 #define SCL_CONFIG 0x01 753 #define SCL_STATE 0x02 754 #define SCL_L2ARC 0x04 /* hack until L2ARC 2.0 */ 755 #define SCL_ALLOC 0x08 756 #define SCL_ZIO 0x10 757 #define SCL_FREE 0x20 758 #define SCL_VDEV 0x40 759 #define SCL_LOCKS 7 760 #define SCL_ALL ((1 << SCL_LOCKS) - 1) 761 #define SCL_STATE_ALL (SCL_STATE | SCL_L2ARC | SCL_ZIO) 762 763 /* Pool configuration locks */ 764 extern int spa_config_tryenter(spa_t *spa, int locks, void *tag, krw_t rw); 765 extern void spa_config_enter(spa_t *spa, int locks, void *tag, krw_t rw); 766 extern void spa_config_exit(spa_t *spa, int locks, void *tag); 767 extern int spa_config_held(spa_t *spa, int locks, krw_t rw); 768 769 /* Pool vdev add/remove lock */ 770 extern uint64_t spa_vdev_enter(spa_t *spa); 771 extern uint64_t spa_vdev_config_enter(spa_t *spa); 772 extern void spa_vdev_config_exit(spa_t *spa, vdev_t *vd, uint64_t txg, 773 int error, char *tag); 774 extern int spa_vdev_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error); 775 776 /* Pool vdev state change lock */ 777 extern void spa_vdev_state_enter(spa_t *spa, int oplock); 778 extern int spa_vdev_state_exit(spa_t *spa, vdev_t *vd, int error); 779 780 /* Log state */ 781 typedef enum spa_log_state { 782 SPA_LOG_UNKNOWN = 0, /* unknown log state */ 783 SPA_LOG_MISSING, /* missing log(s) */ 784 SPA_LOG_CLEAR, /* clear the log(s) */ 785 SPA_LOG_GOOD, /* log(s) are good */ 786 } spa_log_state_t; 787 788 extern spa_log_state_t spa_get_log_state(spa_t *spa); 789 extern void spa_set_log_state(spa_t *spa, spa_log_state_t state); 790 extern int spa_reset_logs(spa_t *spa); 791 792 /* Log claim callback */ 793 extern void spa_claim_notify(zio_t *zio); 794 795 /* Accessor functions */ 796 extern boolean_t spa_shutting_down(spa_t *spa); 797 extern struct dsl_pool *spa_get_dsl(spa_t *spa); 798 extern boolean_t spa_is_initializing(spa_t *spa); 799 extern boolean_t spa_indirect_vdevs_loaded(spa_t *spa); 800 extern blkptr_t *spa_get_rootblkptr(spa_t *spa); 801 extern void spa_set_rootblkptr(spa_t *spa, const blkptr_t *bp); 802 extern void spa_altroot(spa_t *, char *, size_t); 803 extern int spa_sync_pass(spa_t *spa); 804 extern char *spa_name(spa_t *spa); 805 extern uint64_t spa_guid(spa_t *spa); 806 extern uint64_t spa_load_guid(spa_t *spa); 807 extern uint64_t spa_last_synced_txg(spa_t *spa); 808 extern uint64_t spa_first_txg(spa_t *spa); 809 extern uint64_t spa_syncing_txg(spa_t *spa); 810 extern uint64_t spa_final_dirty_txg(spa_t *spa); 811 extern uint64_t spa_version(spa_t *spa); 812 extern pool_state_t spa_state(spa_t *spa); 813 extern spa_load_state_t spa_load_state(spa_t *spa); 814 extern uint64_t spa_freeze_txg(spa_t *spa); 815 extern uint64_t spa_get_worst_case_asize(spa_t *spa, uint64_t lsize); 816 extern uint64_t spa_get_dspace(spa_t *spa); 817 extern uint64_t spa_get_checkpoint_space(spa_t *spa); 818 extern uint64_t spa_get_slop_space(spa_t *spa); 819 extern void spa_update_dspace(spa_t *spa); 820 extern uint64_t spa_version(spa_t *spa); 821 extern boolean_t spa_deflate(spa_t *spa); 822 extern metaslab_class_t *spa_normal_class(spa_t *spa); 823 extern metaslab_class_t *spa_log_class(spa_t *spa); 824 extern void spa_evicting_os_register(spa_t *, objset_t *os); 825 extern void spa_evicting_os_deregister(spa_t *, objset_t *os); 826 extern void spa_evicting_os_wait(spa_t *spa); 827 extern int spa_max_replication(spa_t *spa); 828 extern int spa_prev_software_version(spa_t *spa); 829 extern int spa_busy(void); 830 extern uint8_t spa_get_failmode(spa_t *spa); 831 extern boolean_t spa_suspended(spa_t *spa); 832 extern uint64_t spa_bootfs(spa_t *spa); 833 extern uint64_t spa_delegation(spa_t *spa); 834 extern objset_t *spa_meta_objset(spa_t *spa); 835 extern uint64_t spa_deadman_synctime(spa_t *spa); 836 extern uint64_t spa_dirty_data(spa_t *spa); 837 838 /* Miscellaneous support routines */ 839 extern void spa_load_failed(spa_t *spa, const char *fmt, ...); 840 extern void spa_load_note(spa_t *spa, const char *fmt, ...); 841 extern void spa_activate_mos_feature(spa_t *spa, const char *feature, 842 dmu_tx_t *tx); 843 extern void spa_deactivate_mos_feature(spa_t *spa, const char *feature); 844 extern int spa_rename(const char *oldname, const char *newname); 845 extern spa_t *spa_by_guid(uint64_t pool_guid, uint64_t device_guid); 846 extern boolean_t spa_guid_exists(uint64_t pool_guid, uint64_t device_guid); 847 extern char *spa_strdup(const char *); 848 extern void spa_strfree(char *); 849 extern uint64_t spa_get_random(uint64_t range); 850 extern uint64_t spa_generate_guid(spa_t *spa); 851 extern void snprintf_blkptr(char *buf, size_t buflen, const blkptr_t *bp); 852 extern void spa_freeze(spa_t *spa); 853 extern int spa_change_guid(spa_t *spa); 854 extern void spa_upgrade(spa_t *spa, uint64_t version); 855 extern void spa_evict_all(void); 856 extern vdev_t *spa_lookup_by_guid(spa_t *spa, uint64_t guid, 857 boolean_t l2cache); 858 extern boolean_t spa_has_spare(spa_t *, uint64_t guid); 859 extern uint64_t dva_get_dsize_sync(spa_t *spa, const dva_t *dva); 860 extern uint64_t bp_get_dsize_sync(spa_t *spa, const blkptr_t *bp); 861 extern uint64_t bp_get_dsize(spa_t *spa, const blkptr_t *bp); 862 extern boolean_t spa_has_slogs(spa_t *spa); 863 extern boolean_t spa_is_root(spa_t *spa); 864 extern boolean_t spa_writeable(spa_t *spa); 865 extern boolean_t spa_has_pending_synctask(spa_t *spa); 866 extern int spa_maxblocksize(spa_t *spa); 867 extern boolean_t spa_has_checkpoint(spa_t *spa); 868 extern boolean_t spa_importing_readonly_checkpoint(spa_t *spa); 869 extern boolean_t spa_suspend_async_destroy(spa_t *spa); 870 extern uint64_t spa_min_claim_txg(spa_t *spa); 871 extern void zfs_blkptr_verify(spa_t *spa, const blkptr_t *bp); 872 extern boolean_t zfs_dva_valid(spa_t *spa, const dva_t *dva, 873 const blkptr_t *bp); 874 typedef void (*spa_remap_cb_t)(uint64_t vdev, uint64_t offset, uint64_t size, 875 void *arg); 876 extern boolean_t spa_remap_blkptr(spa_t *spa, blkptr_t *bp, 877 spa_remap_cb_t callback, void *arg); 878 extern uint64_t spa_get_last_removal_txg(spa_t *spa); 879 extern boolean_t spa_trust_config(spa_t *spa); 880 extern uint64_t spa_missing_tvds_allowed(spa_t *spa); 881 extern void spa_set_missing_tvds(spa_t *spa, uint64_t missing); 882 extern boolean_t spa_top_vdevs_spacemap_addressable(spa_t *spa); 883 884 extern int spa_mode(spa_t *spa); 885 extern uint64_t zfs_strtonum(const char *str, char **nptr); 886 887 extern char *spa_his_ievent_table[]; 888 889 extern void spa_history_create_obj(spa_t *spa, dmu_tx_t *tx); 890 extern int spa_history_get(spa_t *spa, uint64_t *offset, uint64_t *len_read, 891 char *his_buf); 892 extern int spa_history_log(spa_t *spa, const char *his_buf); 893 extern int spa_history_log_nvl(spa_t *spa, nvlist_t *nvl); 894 extern void spa_history_log_version(spa_t *spa, const char *operation); 895 extern void spa_history_log_internal(spa_t *spa, const char *operation, 896 dmu_tx_t *tx, const char *fmt, ...); 897 extern void spa_history_log_internal_ds(struct dsl_dataset *ds, const char *op, 898 dmu_tx_t *tx, const char *fmt, ...); 899 extern void spa_history_log_internal_dd(dsl_dir_t *dd, const char *operation, 900 dmu_tx_t *tx, const char *fmt, ...); 901 902 /* error handling */ 903 struct zbookmark_phys; 904 extern void spa_log_error(spa_t *spa, zio_t *zio); 905 extern void zfs_ereport_post(const char *cls, spa_t *spa, vdev_t *vd, 906 zio_t *zio, uint64_t stateoroffset, uint64_t length); 907 extern void zfs_post_remove(spa_t *spa, vdev_t *vd); 908 extern void zfs_post_state_change(spa_t *spa, vdev_t *vd); 909 extern void zfs_post_autoreplace(spa_t *spa, vdev_t *vd); 910 extern uint64_t spa_get_errlog_size(spa_t *spa); 911 extern int spa_get_errlog(spa_t *spa, void *uaddr, size_t *count); 912 extern void spa_errlog_rotate(spa_t *spa); 913 extern void spa_errlog_drain(spa_t *spa); 914 extern void spa_errlog_sync(spa_t *spa, uint64_t txg); 915 extern void spa_get_errlists(spa_t *spa, avl_tree_t *last, avl_tree_t *scrub); 916 917 /* vdev cache */ 918 extern void vdev_cache_stat_init(void); 919 extern void vdev_cache_stat_fini(void); 920 921 /* Initialization and termination */ 922 extern void spa_init(int flags); 923 extern void spa_fini(void); 924 extern void spa_boot_init(void); 925 926 /* properties */ 927 extern int spa_prop_set(spa_t *spa, nvlist_t *nvp); 928 extern int spa_prop_get(spa_t *spa, nvlist_t **nvp); 929 extern void spa_prop_clear_bootfs(spa_t *spa, uint64_t obj, dmu_tx_t *tx); 930 extern void spa_configfile_set(spa_t *, nvlist_t *, boolean_t); 931 932 /* asynchronous event notification */ 933 extern void spa_event_notify(spa_t *spa, vdev_t *vdev, nvlist_t *hist_nvl, 934 const char *name); 935 extern sysevent_t *spa_event_create(spa_t *spa, vdev_t *vd, nvlist_t *hist_nvl, 936 const char *name); 937 extern void spa_event_post(sysevent_t *ev); 938 extern void spa_event_discard(sysevent_t *ev); 939 940 #ifdef ZFS_DEBUG 941 #define dprintf_bp(bp, fmt, ...) do { \ 942 if (zfs_flags & ZFS_DEBUG_DPRINTF) { \ 943 char *__blkbuf = kmem_alloc(BP_SPRINTF_LEN, KM_SLEEP); \ 944 snprintf_blkptr(__blkbuf, BP_SPRINTF_LEN, (bp)); \ 945 dprintf(fmt " %s\n", __VA_ARGS__, __blkbuf); \ 946 kmem_free(__blkbuf, BP_SPRINTF_LEN); \ 947 } \ 948 _NOTE(CONSTCOND) } while (0) 949 #else 950 #define dprintf_bp(bp, fmt, ...) 951 #endif 952 953 extern int spa_mode_global; /* mode, e.g. FREAD | FWRITE */ 954 955 #ifdef __cplusplus 956 } 957 #endif 958 959 #endif /* _SYS_SPA_H */ 960