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  */
  26 
  27 #ifndef _SYS_SPA_H
  28 #define _SYS_SPA_H
  29 
  30 #include <sys/avl.h>
  31 #include <sys/zfs_context.h>
  32 #include <sys/nvpair.h>
  33 #include <sys/sysmacros.h>
  34 #include <sys/types.h>
  35 #include <sys/fs/zfs.h>
  36 
  37 #ifdef  __cplusplus
  38 extern "C" {
  39 #endif
  40 
  41 /*
  42  * Forward references that lots of things need.
  43  */
  44 typedef struct spa spa_t;
  45 typedef struct vdev vdev_t;
  46 typedef struct metaslab metaslab_t;
  47 typedef struct metaslab_group metaslab_group_t;
  48 typedef struct metaslab_class metaslab_class_t;
  49 typedef struct zio zio_t;
  50 typedef struct zilog zilog_t;
  51 typedef struct spa_aux_vdev spa_aux_vdev_t;
  52 typedef struct ddt ddt_t;
  53 typedef struct ddt_entry ddt_entry_t;
  54 struct dsl_pool;
  55 struct dsl_dataset;
  56 
  57 /*
  58  * General-purpose 32-bit and 64-bit bitfield encodings.
  59  */
  60 #define BF32_DECODE(x, low, len)        P2PHASE((x) >> (low), 1U << (len))
  61 #define BF64_DECODE(x, low, len)        P2PHASE((x) >> (low), 1ULL << (len))
  62 #define BF32_ENCODE(x, low, len)        (P2PHASE((x), 1U << (len)) << (low))
  63 #define BF64_ENCODE(x, low, len)        (P2PHASE((x), 1ULL << (len)) << (low))
  64 
  65 #define BF32_GET(x, low, len)           BF32_DECODE(x, low, len)
  66 #define BF64_GET(x, low, len)           BF64_DECODE(x, low, len)
  67 
  68 #define BF32_SET(x, low, len, val)      \
  69         ((x) ^= BF32_ENCODE((x >> low) ^ (val), low, len))
  70 #define BF64_SET(x, low, len, val)      \
  71         ((x) ^= BF64_ENCODE((x >> low) ^ (val), low, len))
  72 
  73 #define BF32_GET_SB(x, low, len, shift, bias)   \
  74         ((BF32_GET(x, low, len) + (bias)) << (shift))
  75 #define BF64_GET_SB(x, low, len, shift, bias)   \
  76         ((BF64_GET(x, low, len) + (bias)) << (shift))
  77 
  78 #define BF32_SET_SB(x, low, len, shift, bias, val)      \
  79         BF32_SET(x, low, len, ((val) >> (shift)) - (bias))
  80 #define BF64_SET_SB(x, low, len, shift, bias, val)      \
  81         BF64_SET(x, low, len, ((val) >> (shift)) - (bias))
  82 
  83 /*
  84  * We currently support nine block sizes, from 512 bytes to 128K.
  85  * We could go higher, but the benefits are near-zero and the cost
  86  * of COWing a giant block to modify one byte would become excessive.
  87  */
  88 #define SPA_MINBLOCKSHIFT       9
  89 #define SPA_MAXBLOCKSHIFT       17
  90 #define SPA_MINBLOCKSIZE        (1ULL << SPA_MINBLOCKSHIFT)
  91 #define SPA_MAXBLOCKSIZE        (1ULL << SPA_MAXBLOCKSHIFT)
  92 
  93 #define SPA_BLOCKSIZES          (SPA_MAXBLOCKSHIFT - SPA_MINBLOCKSHIFT + 1)
  94 
  95 /*
  96  * Size of block to hold the configuration data (a packed nvlist)
  97  */
  98 #define SPA_CONFIG_BLOCKSIZE    (1ULL << 14)
  99 
 100 /*
 101  * The DVA size encodings for LSIZE and PSIZE support blocks up to 32MB.
 102  * The ASIZE encoding should be at least 64 times larger (6 more bits)
 103  * to support up to 4-way RAID-Z mirror mode with worst-case gang block
 104  * overhead, three DVAs per bp, plus one more bit in case we do anything
 105  * else that expands the ASIZE.
 106  */
 107 #define SPA_LSIZEBITS           16      /* LSIZE up to 32M (2^16 * 512) */
 108 #define SPA_PSIZEBITS           16      /* PSIZE up to 32M (2^16 * 512) */
 109 #define SPA_ASIZEBITS           24      /* ASIZE up to 64 times larger  */
 110 
 111 /*
 112  * All SPA data is represented by 128-bit data virtual addresses (DVAs).
 113  * The members of the dva_t should be considered opaque outside the SPA.
 114  */
 115 typedef struct dva {
 116         uint64_t        dva_word[2];
 117 } dva_t;
 118 
 119 /*
 120  * Each block has a 256-bit checksum -- strong enough for cryptographic hashes.
 121  */
 122 typedef struct zio_cksum {
 123         uint64_t        zc_word[4];
 124 } zio_cksum_t;
 125 
 126 /*
 127  * Each block is described by its DVAs, time of birth, checksum, etc.
 128  * The word-by-word, bit-by-bit layout of the blkptr is as follows:
 129  *
 130  *      64      56      48      40      32      24      16      8       0
 131  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 132  * 0    |               vdev1           | GRID  |         ASIZE         |
 133  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 134  * 1    |G|                      offset1                                |
 135  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 136  * 2    |               vdev2           | GRID  |         ASIZE         |
 137  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 138  * 3    |G|                      offset2                                |
 139  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 140  * 4    |               vdev3           | GRID  |         ASIZE         |
 141  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 142  * 5    |G|                      offset3                                |
 143  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 144  * 6    |BDX|lvl| type  | cksum | comp  |     PSIZE     |     LSIZE     |
 145  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 146  * 7    |                       padding                                 |
 147  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 148  * 8    |                       padding                                 |
 149  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 150  * 9    |                       physical birth txg                      |
 151  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 152  * a    |                       logical birth txg                       |
 153  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 154  * b    |                       fill count                              |
 155  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 156  * c    |                       checksum[0]                             |
 157  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 158  * d    |                       checksum[1]                             |
 159  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 160  * e    |                       checksum[2]                             |
 161  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 162  * f    |                       checksum[3]                             |
 163  *      +-------+-------+-------+-------+-------+-------+-------+-------+
 164  *
 165  * Legend:
 166  *
 167  * vdev         virtual device ID
 168  * offset       offset into virtual device
 169  * LSIZE        logical size
 170  * PSIZE        physical size (after compression)
 171  * ASIZE        allocated size (including RAID-Z parity and gang block headers)
 172  * GRID         RAID-Z layout information (reserved for future use)
 173  * cksum        checksum function
 174  * comp         compression function
 175  * G            gang block indicator
 176  * B            byteorder (endianness)
 177  * D            dedup
 178  * X            unused
 179  * lvl          level of indirection
 180  * type         DMU object type
 181  * phys birth   txg of block allocation; zero if same as logical birth txg
 182  * log. birth   transaction group in which the block was logically born
 183  * fill count   number of non-zero blocks under this bp
 184  * checksum[4]  256-bit checksum of the data this bp describes
 185  */
 186 #define SPA_BLKPTRSHIFT 7               /* blkptr_t is 128 bytes        */
 187 #define SPA_DVAS_PER_BP 3               /* Number of DVAs in a bp       */
 188 
 189 typedef struct blkptr {
 190         dva_t           blk_dva[SPA_DVAS_PER_BP]; /* Data Virtual Addresses */
 191         uint64_t        blk_prop;       /* size, compression, type, etc     */
 192         uint64_t        blk_pad[2];     /* Extra space for the future       */
 193         uint64_t        blk_phys_birth; /* txg when block was allocated     */
 194         uint64_t        blk_birth;      /* transaction group at birth       */
 195         uint64_t        blk_fill;       /* fill count                       */
 196         zio_cksum_t     blk_cksum;      /* 256-bit checksum                 */
 197 } blkptr_t;
 198 
 199 /*
 200  * Macros to get and set fields in a bp or DVA.
 201  */
 202 #define DVA_GET_ASIZE(dva)      \
 203         BF64_GET_SB((dva)->dva_word[0], 0, 24, SPA_MINBLOCKSHIFT, 0)
 204 #define DVA_SET_ASIZE(dva, x)   \
 205         BF64_SET_SB((dva)->dva_word[0], 0, 24, SPA_MINBLOCKSHIFT, 0, x)
 206 
 207 #define DVA_GET_GRID(dva)       BF64_GET((dva)->dva_word[0], 24, 8)
 208 #define DVA_SET_GRID(dva, x)    BF64_SET((dva)->dva_word[0], 24, 8, x)
 209 
 210 #define DVA_GET_VDEV(dva)       BF64_GET((dva)->dva_word[0], 32, 32)
 211 #define DVA_SET_VDEV(dva, x)    BF64_SET((dva)->dva_word[0], 32, 32, x)
 212 
 213 #define DVA_GET_OFFSET(dva)     \
 214         BF64_GET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0)
 215 #define DVA_SET_OFFSET(dva, x)  \
 216         BF64_SET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0, x)
 217 
 218 #define DVA_GET_GANG(dva)       BF64_GET((dva)->dva_word[1], 63, 1)
 219 #define DVA_SET_GANG(dva, x)    BF64_SET((dva)->dva_word[1], 63, 1, x)
 220 
 221 #define BP_GET_LSIZE(bp)        \
 222         BF64_GET_SB((bp)->blk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1)
 223 #define BP_SET_LSIZE(bp, x)     \
 224         BF64_SET_SB((bp)->blk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1, x)
 225 
 226 #define BP_GET_PSIZE(bp)        \
 227         BF64_GET_SB((bp)->blk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1)
 228 #define BP_SET_PSIZE(bp, x)     \
 229         BF64_SET_SB((bp)->blk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1, x)
 230 
 231 #define BP_GET_COMPRESS(bp)             BF64_GET((bp)->blk_prop, 32, 8)
 232 #define BP_SET_COMPRESS(bp, x)          BF64_SET((bp)->blk_prop, 32, 8, x)
 233 
 234 #define BP_GET_CHECKSUM(bp)             BF64_GET((bp)->blk_prop, 40, 8)
 235 #define BP_SET_CHECKSUM(bp, x)          BF64_SET((bp)->blk_prop, 40, 8, x)
 236 
 237 #define BP_GET_TYPE(bp)                 BF64_GET((bp)->blk_prop, 48, 8)
 238 #define BP_SET_TYPE(bp, x)              BF64_SET((bp)->blk_prop, 48, 8, x)
 239 
 240 #define BP_GET_LEVEL(bp)                BF64_GET((bp)->blk_prop, 56, 5)
 241 #define BP_SET_LEVEL(bp, x)             BF64_SET((bp)->blk_prop, 56, 5, x)
 242 
 243 #define BP_GET_PROP_BIT_61(bp)          BF64_GET((bp)->blk_prop, 61, 1)
 244 #define BP_SET_PROP_BIT_61(bp, x)       BF64_SET((bp)->blk_prop, 61, 1, x)
 245 
 246 #define BP_GET_DEDUP(bp)                BF64_GET((bp)->blk_prop, 62, 1)
 247 #define BP_SET_DEDUP(bp, x)             BF64_SET((bp)->blk_prop, 62, 1, x)
 248 
 249 #define BP_GET_BYTEORDER(bp)            (0 - BF64_GET((bp)->blk_prop, 63, 1))
 250 #define BP_SET_BYTEORDER(bp, x)         BF64_SET((bp)->blk_prop, 63, 1, x)
 251 
 252 #define BP_PHYSICAL_BIRTH(bp)           \
 253         ((bp)->blk_phys_birth ? (bp)->blk_phys_birth : (bp)->blk_birth)
 254 
 255 #define BP_SET_BIRTH(bp, logical, physical)     \
 256 {                                               \
 257         (bp)->blk_birth = (logical);         \
 258         (bp)->blk_phys_birth = ((logical) == (physical) ? 0 : (physical)); \
 259 }
 260 
 261 #define BP_GET_ASIZE(bp)        \
 262         (DVA_GET_ASIZE(&(bp)->blk_dva[0]) + DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
 263                 DVA_GET_ASIZE(&(bp)->blk_dva[2]))
 264 
 265 #define BP_GET_UCSIZE(bp) \
 266         ((BP_GET_LEVEL(bp) > 0 || DMU_OT_IS_METADATA(BP_GET_TYPE(bp))) ? \
 267         BP_GET_PSIZE(bp) : BP_GET_LSIZE(bp))
 268 
 269 #define BP_GET_NDVAS(bp)        \
 270         (!!DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \
 271         !!DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
 272         !!DVA_GET_ASIZE(&(bp)->blk_dva[2]))
 273 
 274 #define BP_COUNT_GANG(bp)       \
 275         (DVA_GET_GANG(&(bp)->blk_dva[0]) + \
 276         DVA_GET_GANG(&(bp)->blk_dva[1]) + \
 277         DVA_GET_GANG(&(bp)->blk_dva[2]))
 278 
 279 #define DVA_EQUAL(dva1, dva2)   \
 280         ((dva1)->dva_word[1] == (dva2)->dva_word[1] && \
 281         (dva1)->dva_word[0] == (dva2)->dva_word[0])
 282 
 283 #define BP_EQUAL(bp1, bp2)      \
 284         (BP_PHYSICAL_BIRTH(bp1) == BP_PHYSICAL_BIRTH(bp2) &&    \
 285         DVA_EQUAL(&(bp1)->blk_dva[0], &(bp2)->blk_dva[0]) &&      \
 286         DVA_EQUAL(&(bp1)->blk_dva[1], &(bp2)->blk_dva[1]) &&      \
 287         DVA_EQUAL(&(bp1)->blk_dva[2], &(bp2)->blk_dva[2]))
 288 
 289 #define ZIO_CHECKSUM_EQUAL(zc1, zc2) \
 290         (0 == (((zc1).zc_word[0] - (zc2).zc_word[0]) | \
 291         ((zc1).zc_word[1] - (zc2).zc_word[1]) | \
 292         ((zc1).zc_word[2] - (zc2).zc_word[2]) | \
 293         ((zc1).zc_word[3] - (zc2).zc_word[3])))
 294 
 295 #define DVA_IS_VALID(dva)       (DVA_GET_ASIZE(dva) != 0)
 296 
 297 #define ZIO_SET_CHECKSUM(zcp, w0, w1, w2, w3)   \
 298 {                                               \
 299         (zcp)->zc_word[0] = w0;                      \
 300         (zcp)->zc_word[1] = w1;                      \
 301         (zcp)->zc_word[2] = w2;                      \
 302         (zcp)->zc_word[3] = w3;                      \
 303 }
 304 
 305 #define BP_IDENTITY(bp)         (&(bp)->blk_dva[0])
 306 #define BP_IS_GANG(bp)          DVA_GET_GANG(BP_IDENTITY(bp))
 307 #define BP_IS_HOLE(bp)          ((bp)->blk_birth == 0)
 308 
 309 /* BP_IS_RAIDZ(bp) assumes no block compression */
 310 #define BP_IS_RAIDZ(bp)         (DVA_GET_ASIZE(&(bp)->blk_dva[0]) > \
 311                                 BP_GET_PSIZE(bp))
 312 
 313 #define BP_ZERO(bp)                             \
 314 {                                               \
 315         (bp)->blk_dva[0].dva_word[0] = 0;    \
 316         (bp)->blk_dva[0].dva_word[1] = 0;    \
 317         (bp)->blk_dva[1].dva_word[0] = 0;    \
 318         (bp)->blk_dva[1].dva_word[1] = 0;    \
 319         (bp)->blk_dva[2].dva_word[0] = 0;    \
 320         (bp)->blk_dva[2].dva_word[1] = 0;    \
 321         (bp)->blk_prop = 0;                  \
 322         (bp)->blk_pad[0] = 0;                        \
 323         (bp)->blk_pad[1] = 0;                        \
 324         (bp)->blk_phys_birth = 0;            \
 325         (bp)->blk_birth = 0;                 \
 326         (bp)->blk_fill = 0;                  \
 327         ZIO_SET_CHECKSUM(&(bp)->blk_cksum, 0, 0, 0, 0);  \
 328 }
 329 
 330 /*
 331  * Note: the byteorder is either 0 or -1, both of which are palindromes.
 332  * This simplifies the endianness handling a bit.
 333  */
 334 #ifdef _BIG_ENDIAN
 335 #define ZFS_HOST_BYTEORDER      (0ULL)
 336 #else
 337 #define ZFS_HOST_BYTEORDER      (-1ULL)
 338 #endif
 339 
 340 #define BP_SHOULD_BYTESWAP(bp)  (BP_GET_BYTEORDER(bp) != ZFS_HOST_BYTEORDER)
 341 
 342 #define BP_SPRINTF_LEN  320
 343 
 344 /*
 345  * This macro allows code sharing between zfs, libzpool, and mdb.
 346  * 'func' is either snprintf() or mdb_snprintf().
 347  * 'ws' (whitespace) can be ' ' for single-line format, '\n' for multi-line.
 348  */
 349 #define SPRINTF_BLKPTR(func, ws, buf, bp, type, checksum, compress)     \
 350 {                                                                       \
 351         static const char *copyname[] =                                 \
 352             { "zero", "single", "double", "triple" };                   \
 353         int size = BP_SPRINTF_LEN;                                      \
 354         int len = 0;                                                    \
 355         int copies = 0;                                                 \
 356                                                                         \
 357         if (bp == NULL) {                                               \
 358                 len = func(buf + len, size - len, "<NULL>");              \
 359         } else if (BP_IS_HOLE(bp)) {                                    \
 360                 len = func(buf + len, size - len, "<hole>");              \
 361         } else {                                                        \
 362                 for (int d = 0; d < BP_GET_NDVAS(bp); d++) {         \
 363                         const dva_t *dva = &bp->blk_dva[d];              \
 364                         if (DVA_IS_VALID(dva))                          \
 365                                 copies++;                               \
 366                         len += func(buf + len, size - len,              \
 367                             "DVA[%d]=<%llu:%llx:%llx>%c", d,              \
 368                             (u_longlong_t)DVA_GET_VDEV(dva),            \
 369                             (u_longlong_t)DVA_GET_OFFSET(dva),          \
 370                             (u_longlong_t)DVA_GET_ASIZE(dva),           \
 371                             ws);                                        \
 372                 }                                                       \
 373                 if (BP_IS_GANG(bp) &&                                   \
 374                     DVA_GET_ASIZE(&bp->blk_dva[2]) <=                 \
 375                     DVA_GET_ASIZE(&bp->blk_dva[1]) / 2)                  \
 376                         copies--;                                       \
 377                 len += func(buf + len, size - len,                      \
 378                     "[L%llu %s] %s %s %s %s %s %s%c"                    \
 379                     "size=%llxL/%llxP birth=%lluL/%lluP fill=%llu%c"    \
 380                     "cksum=%llx:%llx:%llx:%llx",                        \
 381                     (u_longlong_t)BP_GET_LEVEL(bp),                     \
 382                     type,                                               \
 383                     checksum,                                           \
 384                     compress,                                           \
 385                     BP_GET_BYTEORDER(bp) == 0 ? "BE" : "LE",            \
 386                     BP_IS_GANG(bp) ? "gang" : "contiguous",             \
 387                     BP_GET_DEDUP(bp) ? "dedup" : "unique",              \
 388                     copyname[copies],                                   \
 389                     ws,                                                 \
 390                     (u_longlong_t)BP_GET_LSIZE(bp),                     \
 391                     (u_longlong_t)BP_GET_PSIZE(bp),                     \
 392                     (u_longlong_t)bp->blk_birth,                     \
 393                     (u_longlong_t)BP_PHYSICAL_BIRTH(bp),                \
 394                     (u_longlong_t)bp->blk_fill,                              \
 395                     ws,                                                 \
 396                     (u_longlong_t)bp->blk_cksum.zc_word[0],          \
 397                     (u_longlong_t)bp->blk_cksum.zc_word[1],          \
 398                     (u_longlong_t)bp->blk_cksum.zc_word[2],          \
 399                     (u_longlong_t)bp->blk_cksum.zc_word[3]);         \
 400         }                                                               \
 401         ASSERT(len < size);                                          \
 402 }
 403 
 404 #include <sys/dmu.h>
 405 
 406 #define BP_GET_BUFC_TYPE(bp)                                            \
 407         (((BP_GET_LEVEL(bp) > 0) || (DMU_OT_IS_METADATA(BP_GET_TYPE(bp)))) ? \
 408         ARC_BUFC_METADATA : ARC_BUFC_DATA)
 409 
 410 typedef enum spa_import_type {
 411         SPA_IMPORT_EXISTING,
 412         SPA_IMPORT_ASSEMBLE
 413 } spa_import_type_t;
 414 
 415 /* state manipulation functions */
 416 extern int spa_open(const char *pool, spa_t **, void *tag);
 417 extern int spa_open_rewind(const char *pool, spa_t **, void *tag,
 418     nvlist_t *policy, nvlist_t **config);
 419 extern int spa_get_stats(const char *pool, nvlist_t **config, char *altroot,
 420     size_t buflen);
 421 extern int spa_create(const char *pool, nvlist_t *config, nvlist_t *props,
 422     nvlist_t *zplprops);
 423 extern int spa_import_rootpool(char *devpath, char *devid);
 424 extern int spa_import(const char *pool, nvlist_t *config, nvlist_t *props,
 425     uint64_t flags);
 426 extern nvlist_t *spa_tryimport(nvlist_t *tryconfig);
 427 extern int spa_destroy(char *pool);
 428 extern int spa_export(char *pool, nvlist_t **oldconfig, boolean_t force,
 429     boolean_t hardforce);
 430 extern int spa_reset(char *pool);
 431 extern void spa_async_request(spa_t *spa, int flag);
 432 extern void spa_async_unrequest(spa_t *spa, int flag);
 433 extern void spa_async_suspend(spa_t *spa);
 434 extern void spa_async_resume(spa_t *spa);
 435 extern spa_t *spa_inject_addref(char *pool);
 436 extern void spa_inject_delref(spa_t *spa);
 437 extern void spa_scan_stat_init(spa_t *spa);
 438 extern int spa_scan_get_stats(spa_t *spa, pool_scan_stat_t *ps);
 439 
 440 #define SPA_ASYNC_CONFIG_UPDATE 0x01
 441 #define SPA_ASYNC_REMOVE        0x02
 442 #define SPA_ASYNC_PROBE         0x04
 443 #define SPA_ASYNC_RESILVER_DONE 0x08
 444 #define SPA_ASYNC_RESILVER      0x10
 445 #define SPA_ASYNC_AUTOEXPAND    0x20
 446 #define SPA_ASYNC_REMOVE_DONE   0x40
 447 #define SPA_ASYNC_REMOVE_STOP   0x80
 448 
 449 /*
 450  * Controls the behavior of spa_vdev_remove().
 451  */
 452 #define SPA_REMOVE_UNSPARE      0x01
 453 #define SPA_REMOVE_DONE         0x02
 454 
 455 /* device manipulation */
 456 extern int spa_vdev_add(spa_t *spa, nvlist_t *nvroot);
 457 extern int spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot,
 458     int replacing);
 459 extern int spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid,
 460     int replace_done);
 461 extern int spa_vdev_remove(spa_t *spa, uint64_t guid, boolean_t unspare);
 462 extern boolean_t spa_vdev_remove_active(spa_t *spa);
 463 extern int spa_vdev_setpath(spa_t *spa, uint64_t guid, const char *newpath);
 464 extern int spa_vdev_setfru(spa_t *spa, uint64_t guid, const char *newfru);
 465 extern int spa_vdev_split_mirror(spa_t *spa, char *newname, nvlist_t *config,
 466     nvlist_t *props, boolean_t exp);
 467 
 468 /* spare state (which is global across all pools) */
 469 extern void spa_spare_add(vdev_t *vd);
 470 extern void spa_spare_remove(vdev_t *vd);
 471 extern boolean_t spa_spare_exists(uint64_t guid, uint64_t *pool, int *refcnt);
 472 extern void spa_spare_activate(vdev_t *vd);
 473 
 474 /* L2ARC state (which is global across all pools) */
 475 extern void spa_l2cache_add(vdev_t *vd);
 476 extern void spa_l2cache_remove(vdev_t *vd);
 477 extern boolean_t spa_l2cache_exists(uint64_t guid, uint64_t *pool);
 478 extern void spa_l2cache_activate(vdev_t *vd);
 479 extern void spa_l2cache_drop(spa_t *spa);
 480 
 481 /* scanning */
 482 extern int spa_scan(spa_t *spa, pool_scan_func_t func);
 483 extern int spa_scan_stop(spa_t *spa);
 484 
 485 /* spa syncing */
 486 extern void spa_sync(spa_t *spa, uint64_t txg); /* only for DMU use */
 487 extern void spa_sync_allpools(void);
 488 
 489 /* spa namespace global mutex */
 490 extern kmutex_t spa_namespace_lock;
 491 
 492 /*
 493  * SPA configuration functions in spa_config.c
 494  */
 495 
 496 #define SPA_CONFIG_UPDATE_POOL  0
 497 #define SPA_CONFIG_UPDATE_VDEVS 1
 498 
 499 extern void spa_config_sync(spa_t *, boolean_t, boolean_t);
 500 extern void spa_config_load(void);
 501 extern nvlist_t *spa_all_configs(uint64_t *);
 502 extern void spa_config_set(spa_t *spa, nvlist_t *config);
 503 extern nvlist_t *spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg,
 504     int getstats);
 505 extern void spa_config_update(spa_t *spa, int what);
 506 
 507 /*
 508  * Miscellaneous SPA routines in spa_misc.c
 509  */
 510 
 511 /* Namespace manipulation */
 512 extern spa_t *spa_lookup(const char *name);
 513 extern spa_t *spa_add(const char *name, nvlist_t *config, const char *altroot);
 514 extern void spa_remove(spa_t *spa);
 515 extern spa_t *spa_next(spa_t *prev);
 516 
 517 /* Refcount functions */
 518 extern void spa_open_ref(spa_t *spa, void *tag);
 519 extern void spa_close(spa_t *spa, void *tag);
 520 extern boolean_t spa_refcount_zero(spa_t *spa);
 521 
 522 #define SCL_NONE        0x00
 523 #define SCL_CONFIG      0x01
 524 #define SCL_STATE       0x02
 525 #define SCL_L2ARC       0x04            /* hack until L2ARC 2.0 */
 526 #define SCL_ALLOC       0x08
 527 #define SCL_ZIO         0x10
 528 #define SCL_FREE        0x20
 529 #define SCL_VDEV        0x40
 530 #define SCL_LOCKS       7
 531 #define SCL_ALL         ((1 << SCL_LOCKS) - 1)
 532 #define SCL_STATE_ALL   (SCL_STATE | SCL_L2ARC | SCL_ZIO)
 533 
 534 /* Pool configuration locks */
 535 extern int spa_config_tryenter(spa_t *spa, int locks, void *tag, krw_t rw);
 536 extern void spa_config_enter(spa_t *spa, int locks, void *tag, krw_t rw);
 537 extern void spa_config_exit(spa_t *spa, int locks, void *tag);
 538 extern int spa_config_held(spa_t *spa, int locks, krw_t rw);
 539 
 540 /* Pool vdev add/remove lock */
 541 extern uint64_t spa_vdev_enter(spa_t *spa);
 542 extern uint64_t spa_vdev_config_enter(spa_t *spa);
 543 extern void spa_vdev_config_exit(spa_t *spa, vdev_t *vd, uint64_t txg,
 544     int error, char *tag);
 545 extern int spa_vdev_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error);
 546 
 547 /* Pool vdev state change lock */
 548 extern void spa_vdev_state_enter(spa_t *spa, int oplock);
 549 extern int spa_vdev_state_exit(spa_t *spa, vdev_t *vd, int error);
 550 
 551 /* Log state */
 552 typedef enum spa_log_state {
 553         SPA_LOG_UNKNOWN = 0,    /* unknown log state */
 554         SPA_LOG_MISSING,        /* missing log(s) */
 555         SPA_LOG_CLEAR,          /* clear the log(s) */
 556         SPA_LOG_GOOD,           /* log(s) are good */
 557 } spa_log_state_t;
 558 
 559 extern spa_log_state_t spa_get_log_state(spa_t *spa);
 560 extern void spa_set_log_state(spa_t *spa, spa_log_state_t state);
 561 extern int spa_offline_log(spa_t *spa);
 562 
 563 /* Log claim callback */
 564 extern void spa_claim_notify(zio_t *zio);
 565 
 566 /* Accessor functions */
 567 extern boolean_t spa_shutting_down(spa_t *spa);
 568 extern struct dsl_pool *spa_get_dsl(spa_t *spa);
 569 extern boolean_t spa_is_initializing(spa_t *spa);
 570 extern blkptr_t *spa_get_rootblkptr(spa_t *spa);
 571 extern void spa_set_rootblkptr(spa_t *spa, const blkptr_t *bp);
 572 extern void spa_altroot(spa_t *, char *, size_t);
 573 extern int spa_sync_pass(spa_t *spa);
 574 extern char *spa_name(spa_t *spa);
 575 extern uint64_t spa_guid(spa_t *spa);
 576 extern uint64_t spa_load_guid(spa_t *spa);
 577 extern uint64_t spa_last_synced_txg(spa_t *spa);
 578 extern uint64_t spa_first_txg(spa_t *spa);
 579 extern uint64_t spa_syncing_txg(spa_t *spa);
 580 extern uint64_t spa_version(spa_t *spa);
 581 extern pool_state_t spa_state(spa_t *spa);
 582 extern spa_load_state_t spa_load_state(spa_t *spa);
 583 extern uint64_t spa_freeze_txg(spa_t *spa);
 584 extern uint64_t spa_get_asize(spa_t *spa, uint64_t lsize);
 585 extern uint64_t spa_get_dspace(spa_t *spa);
 586 extern void spa_update_dspace(spa_t *spa);
 587 extern uint64_t spa_version(spa_t *spa);
 588 extern boolean_t spa_deflate(spa_t *spa);
 589 extern metaslab_class_t *spa_normal_class(spa_t *spa);
 590 extern metaslab_class_t *spa_log_class(spa_t *spa);
 591 extern int spa_max_replication(spa_t *spa);
 592 extern int spa_prev_software_version(spa_t *spa);
 593 extern int spa_busy(void);
 594 extern uint8_t spa_get_failmode(spa_t *spa);
 595 extern boolean_t spa_suspended(spa_t *spa);
 596 extern uint64_t spa_bootfs(spa_t *spa);
 597 extern uint64_t spa_delegation(spa_t *spa);
 598 extern objset_t *spa_meta_objset(spa_t *spa);
 599 extern uint64_t spa_deadman_synctime(spa_t *spa);
 600 
 601 /* Miscellaneous support routines */
 602 extern void spa_activate_mos_feature(spa_t *spa, const char *feature);
 603 extern void spa_deactivate_mos_feature(spa_t *spa, const char *feature);
 604 extern int spa_rename(const char *oldname, const char *newname);
 605 extern spa_t *spa_by_guid(uint64_t pool_guid, uint64_t device_guid);
 606 extern boolean_t spa_guid_exists(uint64_t pool_guid, uint64_t device_guid);
 607 extern char *spa_strdup(const char *);
 608 extern void spa_strfree(char *);
 609 extern uint64_t spa_get_random(uint64_t range);
 610 extern uint64_t spa_generate_guid(spa_t *spa);
 611 extern void sprintf_blkptr(char *buf, const blkptr_t *bp);
 612 extern void spa_freeze(spa_t *spa);
 613 extern int spa_change_guid(spa_t *spa);
 614 extern void spa_upgrade(spa_t *spa, uint64_t version);
 615 extern void spa_evict_all(void);
 616 extern vdev_t *spa_lookup_by_guid(spa_t *spa, uint64_t guid,
 617     boolean_t l2cache);
 618 extern boolean_t spa_has_spare(spa_t *, uint64_t guid);
 619 extern uint64_t dva_get_dsize_sync(spa_t *spa, const dva_t *dva);
 620 extern uint64_t bp_get_dsize_sync(spa_t *spa, const blkptr_t *bp);
 621 extern uint64_t bp_get_dsize(spa_t *spa, const blkptr_t *bp);
 622 extern boolean_t spa_has_slogs(spa_t *spa);
 623 extern boolean_t spa_is_root(spa_t *spa);
 624 extern boolean_t spa_writeable(spa_t *spa);
 625 
 626 extern int spa_mode(spa_t *spa);
 627 extern uint64_t strtonum(const char *str, char **nptr);
 628 
 629 extern char *spa_his_ievent_table[];
 630 
 631 extern void spa_history_create_obj(spa_t *spa, dmu_tx_t *tx);
 632 extern int spa_history_get(spa_t *spa, uint64_t *offset, uint64_t *len_read,
 633     char *his_buf);
 634 extern int spa_history_log(spa_t *spa, const char *his_buf);
 635 extern int spa_history_log_nvl(spa_t *spa, nvlist_t *nvl);
 636 extern void spa_history_log_version(spa_t *spa, const char *operation);
 637 extern void spa_history_log_internal(spa_t *spa, const char *operation,
 638     dmu_tx_t *tx, const char *fmt, ...);
 639 extern void spa_history_log_internal_ds(struct dsl_dataset *ds, const char *op,
 640     dmu_tx_t *tx, const char *fmt, ...);
 641 extern void spa_history_log_internal_dd(dsl_dir_t *dd, const char *operation,
 642     dmu_tx_t *tx, const char *fmt, ...);
 643 
 644 /* error handling */
 645 struct zbookmark;
 646 extern void spa_log_error(spa_t *spa, zio_t *zio);
 647 extern void zfs_ereport_post(const char *class, spa_t *spa, vdev_t *vd,
 648     zio_t *zio, uint64_t stateoroffset, uint64_t length);
 649 extern void zfs_post_remove(spa_t *spa, vdev_t *vd);
 650 extern void zfs_post_state_change(spa_t *spa, vdev_t *vd);
 651 extern void zfs_post_autoreplace(spa_t *spa, vdev_t *vd);
 652 extern uint64_t spa_get_errlog_size(spa_t *spa);
 653 extern int spa_get_errlog(spa_t *spa, void *uaddr, size_t *count);
 654 extern void spa_errlog_rotate(spa_t *spa);
 655 extern void spa_errlog_drain(spa_t *spa);
 656 extern void spa_errlog_sync(spa_t *spa, uint64_t txg);
 657 extern void spa_get_errlists(spa_t *spa, avl_tree_t *last, avl_tree_t *scrub);
 658 
 659 /* vdev cache */
 660 extern void vdev_cache_stat_init(void);
 661 extern void vdev_cache_stat_fini(void);
 662 
 663 /* Initialization and termination */
 664 extern void spa_init(int flags);
 665 extern void spa_fini(void);
 666 extern void spa_boot_init();
 667 
 668 /* properties */
 669 extern int spa_prop_set(spa_t *spa, nvlist_t *nvp);
 670 extern int spa_prop_get(spa_t *spa, nvlist_t **nvp);
 671 extern void spa_prop_clear_bootfs(spa_t *spa, uint64_t obj, dmu_tx_t *tx);
 672 extern void spa_configfile_set(spa_t *, nvlist_t *, boolean_t);
 673 
 674 /* asynchronous event notification */
 675 extern void spa_event_notify(spa_t *spa, vdev_t *vdev, const char *name);
 676 
 677 #ifdef ZFS_DEBUG
 678 #define dprintf_bp(bp, fmt, ...) do {                           \
 679         if (zfs_flags & ZFS_DEBUG_DPRINTF) {                        \
 680         char *__blkbuf = kmem_alloc(BP_SPRINTF_LEN, KM_SLEEP);  \
 681         sprintf_blkptr(__blkbuf, (bp));                         \
 682         dprintf(fmt " %s\n", __VA_ARGS__, __blkbuf);            \
 683         kmem_free(__blkbuf, BP_SPRINTF_LEN);                    \
 684         } \
 685 _NOTE(CONSTCOND) } while (0)
 686 #else
 687 #define dprintf_bp(bp, fmt, ...)
 688 #endif
 689 
 690 extern boolean_t spa_debug_enabled(spa_t *spa);
 691 #define spa_dbgmsg(spa, ...)                    \
 692 {                                               \
 693         if (spa_debug_enabled(spa))             \
 694                 zfs_dbgmsg(__VA_ARGS__);        \
 695 }
 696 
 697 extern int spa_mode_global;                     /* mode, e.g. FREAD | FWRITE */
 698 
 699 #ifdef  __cplusplus
 700 }
 701 #endif
 702 
 703 #endif  /* _SYS_SPA_H */