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) 2012 by Delphix. All rights reserved.
  24  * Copyright 2011 Nexenta Systems, Inc.  All rights reserved.
  25  * Copyright 2013 Saso Kiselkov. All rights reserved.
  26  */
  27 
  28 #ifndef _SYS_SPA_H
  29 #define _SYS_SPA_H
  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 metaslab_group metaslab_group_t;
  49 typedef struct metaslab_class metaslab_class_t;
  50 typedef struct zio zio_t;
  51 typedef struct zilog zilog_t;
  52 typedef struct spa_aux_vdev spa_aux_vdev_t;
  53 typedef struct ddt ddt_t;
  54 typedef struct ddt_entry ddt_entry_t;
  55 struct dsl_pool;
  56 struct dsl_dataset;
  57 struct zfeature_info;
  58 
  59 /*
  60  * General-purpose 32-bit and 64-bit bitfield encodings.
  61  */
  62 #define BF32_DECODE(x, low, len)        P2PHASE((x) >> (low), 1U << (len))
  63 #define BF64_DECODE(x, low, len)        P2PHASE((x) >> (low), 1ULL << (len))
  64 #define BF32_ENCODE(x, low, len)        (P2PHASE((x), 1U << (len)) << (low))
  65 #define BF64_ENCODE(x, low, len)        (P2PHASE((x), 1ULL << (len)) << (low))
  66 
  67 #define BF32_GET(x, low, len)           BF32_DECODE(x, low, len)
  68 #define BF64_GET(x, low, len)           BF64_DECODE(x, low, len)
  69 
  70 #define BF32_SET(x, low, len, val)      \
  71         ((x) ^= BF32_ENCODE((x >> low) ^ (val), low, len))
  72 #define BF64_SET(x, low, len, val)      \
  73         ((x) ^= BF64_ENCODE((x >> low) ^ (val), low, len))
  74 
  75 #define BF32_GET_SB(x, low, len, shift, bias)   \
  76         ((BF32_GET(x, low, len) + (bias)) << (shift))
  77 #define BF64_GET_SB(x, low, len, shift, bias)   \
  78         ((BF64_GET(x, low, len) + (bias)) << (shift))
  79 
  80 #define BF32_SET_SB(x, low, len, shift, bias, val)      \
  81         BF32_SET(x, low, len, ((val) >> (shift)) - (bias))
  82 #define BF64_SET_SB(x, low, len, shift, bias, val)      \
  83         BF64_SET(x, low, len, ((val) >> (shift)) - (bias))
  84 
  85 /*
  86  * We currently support nine block sizes, from 512 bytes to 128K.
  87  * We could go higher, but the benefits are near-zero and the cost
  88  * of COWing a giant block to modify one byte would become excessive.
  89  */
  90 #define SPA_MINBLOCKSHIFT       9
  91 #define SPA_MAXBLOCKSHIFT       17
  92 #define SPA_MINBLOCKSIZE        (1ULL << SPA_MINBLOCKSHIFT)
  93 #define SPA_MAXBLOCKSIZE        (1ULL << SPA_MAXBLOCKSHIFT)
  94 
  95 #define SPA_BLOCKSIZES          (SPA_MAXBLOCKSHIFT - SPA_MINBLOCKSHIFT + 1)
  96 
  97 /*
  98  * Size of block to hold the configuration data (a packed nvlist)
  99  */
 100 #define SPA_CONFIG_BLOCKSIZE    (1ULL << 14)
 101 
 102 /*
 103  * The DVA size encodings for LSIZE and PSIZE support blocks up to 32MB.
 104  * The ASIZE encoding should be at least 64 times larger (6 more bits)
 105  * to support up to 4-way RAID-Z mirror mode with worst-case gang block
 106  * overhead, three DVAs per bp, plus one more bit in case we do anything
 107  * else that expands the ASIZE.
 108  */
 109 #define SPA_LSIZEBITS           16      /* LSIZE up to 32M (2^16 * 512) */
 110 #define SPA_PSIZEBITS           16      /* PSIZE up to 32M (2^16 * 512) */
 111 #define SPA_ASIZEBITS           24      /* ASIZE up to 64 times larger  */
 112 
 113 /*
 114  * All SPA data is represented by 128-bit data virtual addresses (DVAs).
 115  * The members of the dva_t should be considered opaque outside the SPA.
 116  */
 117 typedef struct dva {
 118         uint64_t        dva_word[2];
 119 } dva_t;
 120 
 121 /*
 122  * Each block has a 256-bit checksum -- strong enough for cryptographic hashes.
 123  */
 124 typedef struct zio_cksum {
 125         uint64_t        zc_word[4];
 126 } zio_cksum_t;
 127 
 128 /*
 129  * Some checksums/hashes need a 256-bit initialization salt. This salt is kept
 130  * secret and is suitable for use in MAC algorithms as the key.
 131  */
 132 typedef struct zio_cksum_salt {
 133         uint8_t         zcs_bytes[32];
 134 } zio_cksum_salt_t;
 135 
 136 /*
 137  * Each block is described by its DVAs, time of birth, checksum, etc.
 138  * The word-by-word, bit-by-bit layout of the blkptr is as follows:
 139  *
 140  *      64      56      48      40      32      24      16      8       0
 141  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 142  * 0    |               vdev1           | GRID  |         ASIZE         |
 143  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 144  * 1    |G|                      offset1                                |
 145  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 146  * 2    |               vdev2           | GRID  |         ASIZE         |
 147  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 148  * 3    |G|                      offset2                                |
 149  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 150  * 4    |               vdev3           | GRID  |         ASIZE         |
 151  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 152  * 5    |G|                      offset3                                |
 153  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 154  * 6    |BDX|lvl| type  | cksum | comp  |     PSIZE     |     LSIZE     |
 155  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 156  * 7    |                       padding                                 |
 157  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 158  * 8    |                       padding                                 |
 159  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 160  * 9    |                       physical birth txg                      |
 161  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 162  * a    |                       logical birth txg                       |
 163  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 164  * b    |                       fill count                              |
 165  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 166  * c    |                       checksum[0]                             |
 167  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 168  * d    |                       checksum[1]                             |
 169  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 170  * e    |                       checksum[2]                             |
 171  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 172  * f    |                       checksum[3]                             |
 173  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 174  *
 175  * Legend:
 176  *
 177  * vdev         virtual device ID
 178  * offset       offset into virtual device
 179  * LSIZE        logical size
 180  * PSIZE        physical size (after compression)
 181  * ASIZE        allocated size (including RAID-Z parity and gang block headers)
 182  * GRID         RAID-Z layout information (reserved for future use)
 183  * cksum        checksum function
 184  * comp         compression function
 185  * G            gang block indicator
 186  * B            byteorder (endianness)
 187  * D            dedup
 188  * X            unused
 189  * lvl          level of indirection
 190  * type         DMU object type
 191  * phys birth   txg of block allocation; zero if same as logical birth txg
 192  * log. birth   transaction group in which the block was logically born
 193  * fill count   number of non-zero blocks under this bp
 194  * checksum[4]  256-bit checksum of the data this bp describes
 195  */
 196 #define SPA_BLKPTRSHIFT 7               /* blkptr_t is 128 bytes        */
 197 #define SPA_DVAS_PER_BP 3               /* Number of DVAs in a bp       */
 198 
 199 typedef struct blkptr {
 200         dva_t           blk_dva[SPA_DVAS_PER_BP]; /* Data Virtual Addresses */
 201         uint64_t        blk_prop;       /* size, compression, type, etc     */
 202         uint64_t        blk_pad[2];     /* Extra space for the future       */
 203         uint64_t        blk_phys_birth; /* txg when block was allocated     */
 204         uint64_t        blk_birth;      /* transaction group at birth       */
 205         uint64_t        blk_fill;       /* fill count                       */
 206         zio_cksum_t     blk_cksum;      /* 256-bit checksum                 */
 207 } blkptr_t;
 208 
 209 /*
 210  * Macros to get and set fields in a bp or DVA.
 211  */
 212 #define DVA_GET_ASIZE(dva)      \
 213         BF64_GET_SB((dva)->dva_word[0], 0, 24, SPA_MINBLOCKSHIFT, 0)
 214 #define DVA_SET_ASIZE(dva, x)   \
 215         BF64_SET_SB((dva)->dva_word[0], 0, 24, SPA_MINBLOCKSHIFT, 0, x)
 216 
 217 #define DVA_GET_GRID(dva)       BF64_GET((dva)->dva_word[0], 24, 8)
 218 #define DVA_SET_GRID(dva, x)    BF64_SET((dva)->dva_word[0], 24, 8, x)
 219 
 220 #define DVA_GET_VDEV(dva)       BF64_GET((dva)->dva_word[0], 32, 32)
 221 #define DVA_SET_VDEV(dva, x)    BF64_SET((dva)->dva_word[0], 32, 32, x)
 222 
 223 #define DVA_GET_OFFSET(dva)     \
 224         BF64_GET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0)
 225 #define DVA_SET_OFFSET(dva, x)  \
 226         BF64_SET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0, x)
 227 
 228 #define DVA_GET_GANG(dva)       BF64_GET((dva)->dva_word[1], 63, 1)
 229 #define DVA_SET_GANG(dva, x)    BF64_SET((dva)->dva_word[1], 63, 1, x)
 230 
 231 #define BP_GET_LSIZE(bp)        \
 232         BF64_GET_SB((bp)->blk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1)
 233 #define BP_SET_LSIZE(bp, x)     \
 234         BF64_SET_SB((bp)->blk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1, x)
 235 
 236 #define BP_GET_PSIZE(bp)        \
 237         BF64_GET_SB((bp)->blk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1)
 238 #define BP_SET_PSIZE(bp, x)     \
 239         BF64_SET_SB((bp)->blk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1, x)
 240 
 241 #define BP_GET_COMPRESS(bp)             BF64_GET((bp)->blk_prop, 32, 8)
 242 #define BP_SET_COMPRESS(bp, x)          BF64_SET((bp)->blk_prop, 32, 8, x)
 243 
 244 #define BP_GET_CHECKSUM(bp)             BF64_GET((bp)->blk_prop, 40, 8)
 245 #define BP_SET_CHECKSUM(bp, x)          BF64_SET((bp)->blk_prop, 40, 8, x)
 246 
 247 #define BP_GET_TYPE(bp)                 BF64_GET((bp)->blk_prop, 48, 8)
 248 #define BP_SET_TYPE(bp, x)              BF64_SET((bp)->blk_prop, 48, 8, x)
 249 
 250 #define BP_GET_LEVEL(bp)                BF64_GET((bp)->blk_prop, 56, 5)
 251 #define BP_SET_LEVEL(bp, x)             BF64_SET((bp)->blk_prop, 56, 5, x)
 252 
 253 #define BP_GET_PROP_BIT_61(bp)          BF64_GET((bp)->blk_prop, 61, 1)
 254 #define BP_SET_PROP_BIT_61(bp, x)       BF64_SET((bp)->blk_prop, 61, 1, x)
 255 
 256 #define BP_GET_DEDUP(bp)                BF64_GET((bp)->blk_prop, 62, 1)
 257 #define BP_SET_DEDUP(bp, x)             BF64_SET((bp)->blk_prop, 62, 1, x)
 258 
 259 #define BP_GET_BYTEORDER(bp)            (0 - BF64_GET((bp)->blk_prop, 63, 1))
 260 #define BP_SET_BYTEORDER(bp, x)         BF64_SET((bp)->blk_prop, 63, 1, x)
 261 
 262 #define BP_PHYSICAL_BIRTH(bp)           \
 263         ((bp)->blk_phys_birth ? (bp)->blk_phys_birth : (bp)->blk_birth)
 264 
 265 #define BP_SET_BIRTH(bp, logical, physical)     \
 266 {                                               \
 267         (bp)->blk_birth = (logical);         \
 268         (bp)->blk_phys_birth = ((logical) == (physical) ? 0 : (physical)); \
 269 }
 270 
 271 #define BP_GET_ASIZE(bp)        \
 272         (DVA_GET_ASIZE(&(bp)->blk_dva[0]) + DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
 273                 DVA_GET_ASIZE(&(bp)->blk_dva[2]))
 274 
 275 #define BP_GET_UCSIZE(bp) \
 276         ((BP_GET_LEVEL(bp) > 0 || DMU_OT_IS_METADATA(BP_GET_TYPE(bp))) ? \
 277         BP_GET_PSIZE(bp) : BP_GET_LSIZE(bp))
 278 
 279 #define BP_GET_NDVAS(bp)        \
 280         (!!DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \
 281         !!DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
 282         !!DVA_GET_ASIZE(&(bp)->blk_dva[2]))
 283 
 284 #define BP_COUNT_GANG(bp)       \
 285         (DVA_GET_GANG(&(bp)->blk_dva[0]) + \
 286         DVA_GET_GANG(&(bp)->blk_dva[1]) + \
 287         DVA_GET_GANG(&(bp)->blk_dva[2]))
 288 
 289 #define DVA_EQUAL(dva1, dva2)   \
 290         ((dva1)->dva_word[1] == (dva2)->dva_word[1] && \
 291         (dva1)->dva_word[0] == (dva2)->dva_word[0])
 292 
 293 #define BP_EQUAL(bp1, bp2)      \
 294         (BP_PHYSICAL_BIRTH(bp1) == BP_PHYSICAL_BIRTH(bp2) &&    \
 295         DVA_EQUAL(&(bp1)->blk_dva[0], &(bp2)->blk_dva[0]) &&      \
 296         DVA_EQUAL(&(bp1)->blk_dva[1], &(bp2)->blk_dva[1]) &&      \
 297         DVA_EQUAL(&(bp1)->blk_dva[2], &(bp2)->blk_dva[2]))
 298 
 299 #define ZIO_CHECKSUM_EQUAL(zc1, zc2) \
 300         (0 == (((zc1).zc_word[0] - (zc2).zc_word[0]) | \
 301         ((zc1).zc_word[1] - (zc2).zc_word[1]) | \
 302         ((zc1).zc_word[2] - (zc2).zc_word[2]) | \
 303         ((zc1).zc_word[3] - (zc2).zc_word[3])))
 304 
 305 #define DVA_IS_VALID(dva)       (DVA_GET_ASIZE(dva) != 0)
 306 
 307 #define ZIO_SET_CHECKSUM(zcp, w0, w1, w2, w3)   \
 308 {                                               \
 309         (zcp)->zc_word[0] = w0;                      \
 310         (zcp)->zc_word[1] = w1;                      \
 311         (zcp)->zc_word[2] = w2;                      \
 312         (zcp)->zc_word[3] = w3;                      \
 313 }
 314 
 315 #define BP_IDENTITY(bp)         (&(bp)->blk_dva[0])
 316 #define BP_IS_GANG(bp)          DVA_GET_GANG(BP_IDENTITY(bp))
 317 #define BP_IS_HOLE(bp)          ((bp)->blk_birth == 0)
 318 
 319 /* BP_IS_RAIDZ(bp) assumes no block compression */
 320 #define BP_IS_RAIDZ(bp)         (DVA_GET_ASIZE(&(bp)->blk_dva[0]) > \
 321                                 BP_GET_PSIZE(bp))
 322 
 323 #define BP_ZERO(bp)                             \
 324 {                                               \
 325         (bp)->blk_dva[0].dva_word[0] = 0;    \
 326         (bp)->blk_dva[0].dva_word[1] = 0;    \
 327         (bp)->blk_dva[1].dva_word[0] = 0;    \
 328         (bp)->blk_dva[1].dva_word[1] = 0;    \
 329         (bp)->blk_dva[2].dva_word[0] = 0;    \
 330         (bp)->blk_dva[2].dva_word[1] = 0;    \
 331         (bp)->blk_prop = 0;                  \
 332         (bp)->blk_pad[0] = 0;                        \
 333         (bp)->blk_pad[1] = 0;                        \
 334         (bp)->blk_phys_birth = 0;            \
 335         (bp)->blk_birth = 0;                 \
 336         (bp)->blk_fill = 0;                  \
 337         ZIO_SET_CHECKSUM(&(bp)->blk_cksum, 0, 0, 0, 0);  \
 338 }
 339 
 340 /*
 341  * Note: the byteorder is either 0 or -1, both of which are palindromes.
 342  * This simplifies the endianness handling a bit.
 343  */
 344 #ifdef _BIG_ENDIAN
 345 #define ZFS_HOST_BYTEORDER      (0ULL)
 346 #else
 347 #define ZFS_HOST_BYTEORDER      (-1ULL)
 348 #endif
 349 
 350 #define BP_SHOULD_BYTESWAP(bp)  (BP_GET_BYTEORDER(bp) != ZFS_HOST_BYTEORDER)
 351 
 352 #define BP_SPRINTF_LEN  320
 353 
 354 /*
 355  * This macro allows code sharing between zfs, libzpool, and mdb.
 356  * 'func' is either snprintf() or mdb_snprintf().
 357  * 'ws' (whitespace) can be ' ' for single-line format, '\n' for multi-line.
 358  */
 359 #define SPRINTF_BLKPTR(func, ws, buf, bp, type, checksum, compress)     \
 360 {                                                                       \
 361         static const char *copyname[] =                                 \
 362             { "zero", "single", "double", "triple" };                   \
 363         int size = BP_SPRINTF_LEN;                                      \
 364         int len = 0;                                                    \
 365         int copies = 0;                                                 \
 366                                                                         \
 367         if (bp == NULL) {                                               \
 368                 len = func(buf + len, size - len, "<NULL>");              \
 369         } else if (BP_IS_HOLE(bp)) {                                    \
 370                 len = func(buf + len, size - len, "<hole>");              \
 371         } else {                                                        \
 372                 for (int d = 0; d < BP_GET_NDVAS(bp); d++) {         \
 373                         const dva_t *dva = &bp->blk_dva[d];              \
 374                         if (DVA_IS_VALID(dva))                          \
 375                                 copies++;                               \
 376                         len += func(buf + len, size - len,              \
 377                             "DVA[%d]=<%llu:%llx:%llx>%c", d,              \
 378                             (u_longlong_t)DVA_GET_VDEV(dva),            \
 379                             (u_longlong_t)DVA_GET_OFFSET(dva),          \
 380                             (u_longlong_t)DVA_GET_ASIZE(dva),           \
 381                             ws);                                        \
 382                 }                                                       \
 383                 if (BP_IS_GANG(bp) &&                                   \
 384                     DVA_GET_ASIZE(&bp->blk_dva[2]) <=                 \
 385                     DVA_GET_ASIZE(&bp->blk_dva[1]) / 2)                  \
 386                         copies--;                                       \
 387                 len += func(buf + len, size - len,                      \
 388                     "[L%llu %s] %s %s %s %s %s %s%c"                    \
 389                     "size=%llxL/%llxP birth=%lluL/%lluP fill=%llu%c"    \
 390                     "cksum=%llx:%llx:%llx:%llx",                        \
 391                     (u_longlong_t)BP_GET_LEVEL(bp),                     \
 392                     type,                                               \
 393                     checksum,                                           \
 394                     compress,                                           \
 395                     BP_GET_BYTEORDER(bp) == 0 ? "BE" : "LE",            \
 396                     BP_IS_GANG(bp) ? "gang" : "contiguous",             \
 397                     BP_GET_DEDUP(bp) ? "dedup" : "unique",              \
 398                     copyname[copies],                                   \
 399                     ws,                                                 \
 400                     (u_longlong_t)BP_GET_LSIZE(bp),                     \
 401                     (u_longlong_t)BP_GET_PSIZE(bp),                     \
 402                     (u_longlong_t)bp->blk_birth,                     \
 403                     (u_longlong_t)BP_PHYSICAL_BIRTH(bp),                \
 404                     (u_longlong_t)bp->blk_fill,                              \
 405                     ws,                                                 \
 406                     (u_longlong_t)bp->blk_cksum.zc_word[0],          \
 407                     (u_longlong_t)bp->blk_cksum.zc_word[1],          \
 408                     (u_longlong_t)bp->blk_cksum.zc_word[2],          \
 409                     (u_longlong_t)bp->blk_cksum.zc_word[3]);         \
 410         }                                                               \
 411         ASSERT(len < size);                                          \
 412 }
 413 
 414 #include <sys/dmu.h>
 415 
 416 #define BP_GET_BUFC_TYPE(bp)                                            \
 417         (((BP_GET_LEVEL(bp) > 0) || (DMU_OT_IS_METADATA(BP_GET_TYPE(bp)))) ? \
 418         ARC_BUFC_METADATA : ARC_BUFC_DATA)
 419 
 420 typedef enum spa_import_type {
 421         SPA_IMPORT_EXISTING,
 422         SPA_IMPORT_ASSEMBLE
 423 } spa_import_type_t;
 424 
 425 /* state manipulation functions */
 426 extern int spa_open(const char *pool, spa_t **, void *tag);
 427 extern int spa_open_rewind(const char *pool, spa_t **, void *tag,
 428     nvlist_t *policy, nvlist_t **config);
 429 extern int spa_get_stats(const char *pool, nvlist_t **config, char *altroot,
 430     size_t buflen);
 431 extern int spa_create(const char *pool, nvlist_t *config, nvlist_t *props,
 432     nvlist_t *zplprops);
 433 extern int spa_import_rootpool(char *devpath, char *devid);
 434 extern int spa_import(const char *pool, nvlist_t *config, nvlist_t *props,
 435     uint64_t flags);
 436 extern nvlist_t *spa_tryimport(nvlist_t *tryconfig);
 437 extern int spa_destroy(char *pool);
 438 extern int spa_export(char *pool, nvlist_t **oldconfig, boolean_t force,
 439     boolean_t hardforce);
 440 extern int spa_reset(char *pool);
 441 extern void spa_async_request(spa_t *spa, int flag);
 442 extern void spa_async_unrequest(spa_t *spa, int flag);
 443 extern void spa_async_suspend(spa_t *spa);
 444 extern void spa_async_resume(spa_t *spa);
 445 extern spa_t *spa_inject_addref(char *pool);
 446 extern void spa_inject_delref(spa_t *spa);
 447 extern void spa_scan_stat_init(spa_t *spa);
 448 extern int spa_scan_get_stats(spa_t *spa, pool_scan_stat_t *ps);
 449 
 450 #define SPA_ASYNC_CONFIG_UPDATE 0x01
 451 #define SPA_ASYNC_REMOVE        0x02
 452 #define SPA_ASYNC_PROBE         0x04
 453 #define SPA_ASYNC_RESILVER_DONE 0x08
 454 #define SPA_ASYNC_RESILVER      0x10
 455 #define SPA_ASYNC_AUTOEXPAND    0x20
 456 #define SPA_ASYNC_REMOVE_DONE   0x40
 457 #define SPA_ASYNC_REMOVE_STOP   0x80
 458 
 459 /*
 460  * Controls the behavior of spa_vdev_remove().
 461  */
 462 #define SPA_REMOVE_UNSPARE      0x01
 463 #define SPA_REMOVE_DONE         0x02
 464 
 465 /* device manipulation */
 466 extern int spa_vdev_add(spa_t *spa, nvlist_t *nvroot);
 467 extern int spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot,
 468     int replacing);
 469 extern int spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid,
 470     int replace_done);
 471 extern int spa_vdev_remove(spa_t *spa, uint64_t guid, boolean_t unspare);
 472 extern boolean_t spa_vdev_remove_active(spa_t *spa);
 473 extern int spa_vdev_setpath(spa_t *spa, uint64_t guid, const char *newpath);
 474 extern int spa_vdev_setfru(spa_t *spa, uint64_t guid, const char *newfru);
 475 extern int spa_vdev_split_mirror(spa_t *spa, char *newname, nvlist_t *config,
 476     nvlist_t *props, boolean_t exp);
 477 
 478 /* spare state (which is global across all pools) */
 479 extern void spa_spare_add(vdev_t *vd);
 480 extern void spa_spare_remove(vdev_t *vd);
 481 extern boolean_t spa_spare_exists(uint64_t guid, uint64_t *pool, int *refcnt);
 482 extern void spa_spare_activate(vdev_t *vd);
 483 
 484 /* L2ARC state (which is global across all pools) */
 485 extern void spa_l2cache_add(vdev_t *vd);
 486 extern void spa_l2cache_remove(vdev_t *vd);
 487 extern boolean_t spa_l2cache_exists(uint64_t guid, uint64_t *pool);
 488 extern void spa_l2cache_activate(vdev_t *vd);
 489 extern void spa_l2cache_drop(spa_t *spa);
 490 
 491 /* scanning */
 492 extern int spa_scan(spa_t *spa, pool_scan_func_t func);
 493 extern int spa_scan_stop(spa_t *spa);
 494 
 495 /* spa syncing */
 496 extern void spa_sync(spa_t *spa, uint64_t txg); /* only for DMU use */
 497 extern void spa_sync_allpools(void);
 498 
 499 /* spa namespace global mutex */
 500 extern kmutex_t spa_namespace_lock;
 501 
 502 /*
 503  * SPA configuration functions in spa_config.c
 504  */
 505 
 506 #define SPA_CONFIG_UPDATE_POOL  0
 507 #define SPA_CONFIG_UPDATE_VDEVS 1
 508 
 509 extern void spa_config_sync(spa_t *, boolean_t, boolean_t);
 510 extern void spa_config_load(void);
 511 extern nvlist_t *spa_all_configs(uint64_t *);
 512 extern void spa_config_set(spa_t *spa, nvlist_t *config);
 513 extern nvlist_t *spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg,
 514     int getstats);
 515 extern void spa_config_update(spa_t *spa, int what);
 516 
 517 /*
 518  * Miscellaneous SPA routines in spa_misc.c
 519  */
 520 
 521 /* Namespace manipulation */
 522 extern spa_t *spa_lookup(const char *name);
 523 extern spa_t *spa_add(const char *name, nvlist_t *config, const char *altroot);
 524 extern void spa_remove(spa_t *spa);
 525 extern spa_t *spa_next(spa_t *prev);
 526 
 527 /* Refcount functions */
 528 extern void spa_open_ref(spa_t *spa, void *tag);
 529 extern void spa_close(spa_t *spa, void *tag);
 530 extern boolean_t spa_refcount_zero(spa_t *spa);
 531 
 532 #define SCL_NONE        0x00
 533 #define SCL_CONFIG      0x01
 534 #define SCL_STATE       0x02
 535 #define SCL_L2ARC       0x04            /* hack until L2ARC 2.0 */
 536 #define SCL_ALLOC       0x08
 537 #define SCL_ZIO         0x10
 538 #define SCL_FREE        0x20
 539 #define SCL_VDEV        0x40
 540 #define SCL_LOCKS       7
 541 #define SCL_ALL         ((1 << SCL_LOCKS) - 1)
 542 #define SCL_STATE_ALL   (SCL_STATE | SCL_L2ARC | SCL_ZIO)
 543 
 544 /* Pool configuration locks */
 545 extern int spa_config_tryenter(spa_t *spa, int locks, void *tag, krw_t rw);
 546 extern void spa_config_enter(spa_t *spa, int locks, void *tag, krw_t rw);
 547 extern void spa_config_exit(spa_t *spa, int locks, void *tag);
 548 extern int spa_config_held(spa_t *spa, int locks, krw_t rw);
 549 
 550 /* Pool vdev add/remove lock */
 551 extern uint64_t spa_vdev_enter(spa_t *spa);
 552 extern uint64_t spa_vdev_config_enter(spa_t *spa);
 553 extern void spa_vdev_config_exit(spa_t *spa, vdev_t *vd, uint64_t txg,
 554     int error, char *tag);
 555 extern int spa_vdev_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error);
 556 
 557 /* Pool vdev state change lock */
 558 extern void spa_vdev_state_enter(spa_t *spa, int oplock);
 559 extern int spa_vdev_state_exit(spa_t *spa, vdev_t *vd, int error);
 560 
 561 /* Log state */
 562 typedef enum spa_log_state {
 563         SPA_LOG_UNKNOWN = 0,    /* unknown log state */
 564         SPA_LOG_MISSING,        /* missing log(s) */
 565         SPA_LOG_CLEAR,          /* clear the log(s) */
 566         SPA_LOG_GOOD,           /* log(s) are good */
 567 } spa_log_state_t;
 568 
 569 extern spa_log_state_t spa_get_log_state(spa_t *spa);
 570 extern void spa_set_log_state(spa_t *spa, spa_log_state_t state);
 571 extern int spa_offline_log(spa_t *spa);
 572 
 573 /* Log claim callback */
 574 extern void spa_claim_notify(zio_t *zio);
 575 
 576 /* Accessor functions */
 577 extern boolean_t spa_shutting_down(spa_t *spa);
 578 extern struct dsl_pool *spa_get_dsl(spa_t *spa);
 579 extern boolean_t spa_is_initializing(spa_t *spa);
 580 extern blkptr_t *spa_get_rootblkptr(spa_t *spa);
 581 extern void spa_set_rootblkptr(spa_t *spa, const blkptr_t *bp);
 582 extern void spa_altroot(spa_t *, char *, size_t);
 583 extern int spa_sync_pass(spa_t *spa);
 584 extern char *spa_name(spa_t *spa);
 585 extern uint64_t spa_guid(spa_t *spa);
 586 extern uint64_t spa_load_guid(spa_t *spa);
 587 extern uint64_t spa_last_synced_txg(spa_t *spa);
 588 extern uint64_t spa_first_txg(spa_t *spa);
 589 extern uint64_t spa_syncing_txg(spa_t *spa);
 590 extern uint64_t spa_version(spa_t *spa);
 591 extern pool_state_t spa_state(spa_t *spa);
 592 extern spa_load_state_t spa_load_state(spa_t *spa);
 593 extern uint64_t spa_freeze_txg(spa_t *spa);
 594 extern uint64_t spa_get_asize(spa_t *spa, uint64_t lsize);
 595 extern uint64_t spa_get_dspace(spa_t *spa);
 596 extern void spa_update_dspace(spa_t *spa);
 597 extern uint64_t spa_version(spa_t *spa);
 598 extern boolean_t spa_deflate(spa_t *spa);
 599 extern metaslab_class_t *spa_normal_class(spa_t *spa);
 600 extern metaslab_class_t *spa_log_class(spa_t *spa);
 601 extern int spa_max_replication(spa_t *spa);
 602 extern int spa_prev_software_version(spa_t *spa);
 603 extern int spa_busy(void);
 604 extern uint8_t spa_get_failmode(spa_t *spa);
 605 extern boolean_t spa_suspended(spa_t *spa);
 606 extern uint64_t spa_bootfs(spa_t *spa);
 607 extern uint64_t spa_delegation(spa_t *spa);
 608 extern objset_t *spa_meta_objset(spa_t *spa);
 609 extern uint64_t spa_deadman_synctime(spa_t *spa);
 610 
 611 /* Miscellaneous support routines */
 612 extern void spa_activate_mos_feature(spa_t *spa, const char *feature);
 613 extern void spa_deactivate_mos_feature(spa_t *spa, const char *feature);
 614 extern int spa_rename(const char *oldname, const char *newname);
 615 extern spa_t *spa_by_guid(uint64_t pool_guid, uint64_t device_guid);
 616 extern boolean_t spa_guid_exists(uint64_t pool_guid, uint64_t device_guid);
 617 extern char *spa_strdup(const char *);
 618 extern void spa_strfree(char *);
 619 extern uint64_t spa_get_random(uint64_t range);
 620 extern uint64_t spa_generate_guid(spa_t *spa);
 621 extern void sprintf_blkptr(char *buf, const blkptr_t *bp);
 622 extern void spa_freeze(spa_t *spa);
 623 extern int spa_change_guid(spa_t *spa);
 624 extern void spa_upgrade(spa_t *spa, uint64_t version);
 625 extern void spa_evict_all(void);
 626 extern vdev_t *spa_lookup_by_guid(spa_t *spa, uint64_t guid,
 627     boolean_t l2cache);
 628 extern boolean_t spa_has_spare(spa_t *, uint64_t guid);
 629 extern uint64_t dva_get_dsize_sync(spa_t *spa, const dva_t *dva);
 630 extern uint64_t bp_get_dsize_sync(spa_t *spa, const blkptr_t *bp);
 631 extern uint64_t bp_get_dsize(spa_t *spa, const blkptr_t *bp);
 632 extern boolean_t spa_has_slogs(spa_t *spa);
 633 extern boolean_t spa_is_root(spa_t *spa);
 634 extern boolean_t spa_writeable(spa_t *spa);
 635 
 636 /* Salted checksum handling */
 637 extern int spa_activate_salted_cksum(spa_t *spa, struct zfeature_info *feature);
 638 
 639 extern int spa_mode(spa_t *spa);
 640 extern uint64_t strtonum(const char *str, char **nptr);
 641 
 642 extern char *spa_his_ievent_table[];
 643 
 644 extern void spa_history_create_obj(spa_t *spa, dmu_tx_t *tx);
 645 extern int spa_history_get(spa_t *spa, uint64_t *offset, uint64_t *len_read,
 646     char *his_buf);
 647 extern int spa_history_log(spa_t *spa, const char *his_buf);
 648 extern int spa_history_log_nvl(spa_t *spa, nvlist_t *nvl);
 649 extern void spa_history_log_version(spa_t *spa, const char *operation);
 650 extern void spa_history_log_internal(spa_t *spa, const char *operation,
 651     dmu_tx_t *tx, const char *fmt, ...);
 652 extern void spa_history_log_internal_ds(struct dsl_dataset *ds, const char *op,
 653     dmu_tx_t *tx, const char *fmt, ...);
 654 extern void spa_history_log_internal_dd(dsl_dir_t *dd, const char *operation,
 655     dmu_tx_t *tx, const char *fmt, ...);
 656 
 657 /* error handling */
 658 struct zbookmark;
 659 extern void spa_log_error(spa_t *spa, zio_t *zio);
 660 extern void zfs_ereport_post(const char *class, spa_t *spa, vdev_t *vd,
 661     zio_t *zio, uint64_t stateoroffset, uint64_t length);
 662 extern void zfs_post_remove(spa_t *spa, vdev_t *vd);
 663 extern void zfs_post_state_change(spa_t *spa, vdev_t *vd);
 664 extern void zfs_post_autoreplace(spa_t *spa, vdev_t *vd);
 665 extern uint64_t spa_get_errlog_size(spa_t *spa);
 666 extern int spa_get_errlog(spa_t *spa, void *uaddr, size_t *count);
 667 extern void spa_errlog_rotate(spa_t *spa);
 668 extern void spa_errlog_drain(spa_t *spa);
 669 extern void spa_errlog_sync(spa_t *spa, uint64_t txg);
 670 extern void spa_get_errlists(spa_t *spa, avl_tree_t *last, avl_tree_t *scrub);
 671 
 672 /* vdev cache */
 673 extern void vdev_cache_stat_init(void);
 674 extern void vdev_cache_stat_fini(void);
 675 
 676 /* Initialization and termination */
 677 extern void spa_init(int flags);
 678 extern void spa_fini(void);
 679 extern void spa_boot_init();
 680 
 681 /* properties */
 682 extern int spa_prop_set(spa_t *spa, nvlist_t *nvp);
 683 extern int spa_prop_get(spa_t *spa, nvlist_t **nvp);
 684 extern void spa_prop_clear_bootfs(spa_t *spa, uint64_t obj, dmu_tx_t *tx);
 685 extern void spa_configfile_set(spa_t *, nvlist_t *, boolean_t);
 686 
 687 /* asynchronous event notification */
 688 extern void spa_event_notify(spa_t *spa, vdev_t *vdev, const char *name);
 689 
 690 #ifdef ZFS_DEBUG
 691 #define dprintf_bp(bp, fmt, ...) do {                           \
 692         if (zfs_flags & ZFS_DEBUG_DPRINTF) {                        \
 693         char *__blkbuf = kmem_alloc(BP_SPRINTF_LEN, KM_SLEEP);  \
 694         sprintf_blkptr(__blkbuf, (bp));                         \
 695         dprintf(fmt " %s\n", __VA_ARGS__, __blkbuf);            \
 696         kmem_free(__blkbuf, BP_SPRINTF_LEN);                    \
 697         } \
 698 _NOTE(CONSTCOND) } while (0)
 699 #else
 700 #define dprintf_bp(bp, fmt, ...)
 701 #endif
 702 
 703 extern boolean_t spa_debug_enabled(spa_t *spa);
 704 #define spa_dbgmsg(spa, ...)                    \
 705 {                                               \
 706         if (spa_debug_enabled(spa))             \
 707                 zfs_dbgmsg(__VA_ARGS__);        \
 708 }
 709 
 710 extern int spa_mode_global;                     /* mode, e.g. FREAD | FWRITE */
 711 
 712 #ifdef  __cplusplus
 713 }
 714 #endif
 715 
 716 #endif  /* _SYS_SPA_H */