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
38extern "C" {
39#endif
40
41/*
42 * Forward references that lots of things need.
43 */
44typedef struct spa spa_t;
45typedef struct vdev vdev_t;
46typedef struct metaslab metaslab_t;
47typedef struct metaslab_group metaslab_group_t;
48typedef struct metaslab_class metaslab_class_t;
49typedef struct zio zio_t;
50typedef struct zilog zilog_t;
51typedef struct spa_aux_vdev spa_aux_vdev_t;
52typedef struct ddt ddt_t;
53typedef struct ddt_entry ddt_entry_t;
54struct dsl_pool;
55
56/*
57 * General-purpose 32-bit and 64-bit bitfield encodings.
58 */
59#define BF32_DECODE(x, low, len) P2PHASE((x) >> (low), 1U << (len))
60#define BF64_DECODE(x, low, len) P2PHASE((x) >> (low), 1ULL << (len))
61#define BF32_ENCODE(x, low, len) (P2PHASE((x), 1U << (len)) << (low))
62#define BF64_ENCODE(x, low, len) (P2PHASE((x), 1ULL << (len)) << (low))
63
64#define BF32_GET(x, low, len) BF32_DECODE(x, low, len)
65#define BF64_GET(x, low, len) BF64_DECODE(x, low, len)
66
67#define BF32_SET(x, low, len, val) \
68 ((x) ^= BF32_ENCODE((x >> low) ^ (val), low, len))
69#define BF64_SET(x, low, len, val) \
70 ((x) ^= BF64_ENCODE((x >> low) ^ (val), low, len))
71
72#define BF32_GET_SB(x, low, len, shift, bias) \
73 ((BF32_GET(x, low, len) + (bias)) << (shift))
74#define BF64_GET_SB(x, low, len, shift, bias) \
75 ((BF64_GET(x, low, len) + (bias)) << (shift))
76
77#define BF32_SET_SB(x, low, len, shift, bias, val) \
78 BF32_SET(x, low, len, ((val) >> (shift)) - (bias))
79#define BF64_SET_SB(x, low, len, shift, bias, val) \
80 BF64_SET(x, low, len, ((val) >> (shift)) - (bias))
81
82/*
83 * We currently support nine block sizes, from 512 bytes to 128K.
84 * We could go higher, but the benefits are near-zero and the cost
85 * of COWing a giant block to modify one byte would become excessive.
86 */
87#define SPA_MINBLOCKSHIFT 9
88#define SPA_MAXBLOCKSHIFT 17
89#define SPA_MINBLOCKSIZE (1ULL << SPA_MINBLOCKSHIFT)
90#define SPA_MAXBLOCKSIZE (1ULL << SPA_MAXBLOCKSHIFT)
91
92#define SPA_BLOCKSIZES (SPA_MAXBLOCKSHIFT - SPA_MINBLOCKSHIFT + 1)
93
94/*
95 * Size of block to hold the configuration data (a packed nvlist)
96 */
97#define SPA_CONFIG_BLOCKSIZE (1ULL << 14)
98
99/*
100 * The DVA size encodings for LSIZE and PSIZE support blocks up to 32MB.
101 * The ASIZE encoding should be at least 64 times larger (6 more bits)
102 * to support up to 4-way RAID-Z mirror mode with worst-case gang block
103 * overhead, three DVAs per bp, plus one more bit in case we do anything
104 * else that expands the ASIZE.
105 */
106#define SPA_LSIZEBITS 16 /* LSIZE up to 32M (2^16 * 512) */
107#define SPA_PSIZEBITS 16 /* PSIZE up to 32M (2^16 * 512) */
108#define SPA_ASIZEBITS 24 /* ASIZE up to 64 times larger */
109
110/*
111 * All SPA data is represented by 128-bit data virtual addresses (DVAs).
112 * The members of the dva_t should be considered opaque outside the SPA.
113 */
114typedef struct dva {
115 uint64_t dva_word[2];
116} dva_t;
117
118/*
119 * Each block has a 256-bit checksum -- strong enough for cryptographic hashes.
120 */
121typedef struct zio_cksum {
122 uint64_t zc_word[4];
123} zio_cksum_t;
124
125/*
126 * Each block is described by its DVAs, time of birth, checksum, etc.
127 * The word-by-word, bit-by-bit layout of the blkptr is as follows:
128 *
129 * 64 56 48 40 32 24 16 8 0
130 * +-------+-------+-------+-------+-------+-------+-------+-------+
131 * 0 | vdev1 | GRID | ASIZE |
132 * +-------+-------+-------+-------+-------+-------+-------+-------+
133 * 1 |G| offset1 |
134 * +-------+-------+-------+-------+-------+-------+-------+-------+
135 * 2 | vdev2 | GRID | ASIZE |
136 * +-------+-------+-------+-------+-------+-------+-------+-------+
137 * 3 |G| offset2 |
138 * +-------+-------+-------+-------+-------+-------+-------+-------+
139 * 4 | vdev3 | GRID | ASIZE |
140 * +-------+-------+-------+-------+-------+-------+-------+-------+
141 * 5 |G| offset3 |
142 * +-------+-------+-------+-------+-------+-------+-------+-------+
143 * 6 |BDX|lvl| type | cksum | comp | PSIZE | LSIZE |
144 * +-------+-------+-------+-------+-------+-------+-------+-------+
145 * 7 | padding |
146 * +-------+-------+-------+-------+-------+-------+-------+-------+
147 * 8 | padding |
148 * +-------+-------+-------+-------+-------+-------+-------+-------+
149 * 9 | physical birth txg |
150 * +-------+-------+-------+-------+-------+-------+-------+-------+
151 * a | logical birth txg |
152 * +-------+-------+-------+-------+-------+-------+-------+-------+
153 * b | fill count |
154 * +-------+-------+-------+-------+-------+-------+-------+-------+
155 * c | checksum[0] |
156 * +-------+-------+-------+-------+-------+-------+-------+-------+
157 * d | checksum[1] |
158 * +-------+-------+-------+-------+-------+-------+-------+-------+
159 * e | checksum[2] |
160 * +-------+-------+-------+-------+-------+-------+-------+-------+
161 * f | checksum[3] |
162 * +-------+-------+-------+-------+-------+-------+-------+-------+
163 *
164 * Legend:
165 *
166 * vdev virtual device ID
167 * offset offset into virtual device
168 * LSIZE logical size
169 * PSIZE physical size (after compression)
170 * ASIZE allocated size (including RAID-Z parity and gang block headers)
171 * GRID RAID-Z layout information (reserved for future use)
172 * cksum checksum function
173 * comp compression function
174 * G gang block indicator
175 * B byteorder (endianness)
176 * D dedup
177 * X unused
178 * lvl level of indirection
179 * type DMU object type
180 * phys birth txg of block allocation; zero if same as logical birth txg
181 * log. birth transaction group in which the block was logically born
182 * fill count number of non-zero blocks under this bp
183 * checksum[4] 256-bit checksum of the data this bp describes
184 */
185#define SPA_BLKPTRSHIFT 7 /* blkptr_t is 128 bytes */
186#define SPA_DVAS_PER_BP 3 /* Number of DVAs in a bp */
187
188typedef struct blkptr {
189 dva_t blk_dva[SPA_DVAS_PER_BP]; /* Data Virtual Addresses */
190 uint64_t blk_prop; /* size, compression, type, etc */
191 uint64_t blk_pad[2]; /* Extra space for the future */
192 uint64_t blk_phys_birth; /* txg when block was allocated */
193 uint64_t blk_birth; /* transaction group at birth */
194 uint64_t blk_fill; /* fill count */
195 zio_cksum_t blk_cksum; /* 256-bit checksum */
196} blkptr_t;
197
198/*
199 * Macros to get and set fields in a bp or DVA.
200 */
201#define DVA_GET_ASIZE(dva) \
202 BF64_GET_SB((dva)->dva_word[0], 0, 24, SPA_MINBLOCKSHIFT, 0)
203#define DVA_SET_ASIZE(dva, x) \
204 BF64_SET_SB((dva)->dva_word[0], 0, 24, SPA_MINBLOCKSHIFT, 0, x)
205
206#define DVA_GET_GRID(dva) BF64_GET((dva)->dva_word[0], 24, 8)
207#define DVA_SET_GRID(dva, x) BF64_SET((dva)->dva_word[0], 24, 8, x)
208
209#define DVA_GET_VDEV(dva) BF64_GET((dva)->dva_word[0], 32, 32)
210#define DVA_SET_VDEV(dva, x) BF64_SET((dva)->dva_word[0], 32, 32, x)
211
212#define DVA_GET_OFFSET(dva) \
213 BF64_GET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0)
214#define DVA_SET_OFFSET(dva, x) \
215 BF64_SET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0, x)
216
217#define DVA_GET_GANG(dva) BF64_GET((dva)->dva_word[1], 63, 1)
218#define DVA_SET_GANG(dva, x) BF64_SET((dva)->dva_word[1], 63, 1, x)
219
220#define BP_GET_LSIZE(bp) \
221 BF64_GET_SB((bp)->blk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1)
222#define BP_SET_LSIZE(bp, x) \
223 BF64_SET_SB((bp)->blk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1, x)
224
225#define BP_GET_PSIZE(bp) \
226 BF64_GET_SB((bp)->blk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1)
227#define BP_SET_PSIZE(bp, x) \
228 BF64_SET_SB((bp)->blk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1, x)
229
230#define BP_GET_COMPRESS(bp) BF64_GET((bp)->blk_prop, 32, 8)
231#define BP_SET_COMPRESS(bp, x) BF64_SET((bp)->blk_prop, 32, 8, x)
232
233#define BP_GET_CHECKSUM(bp) BF64_GET((bp)->blk_prop, 40, 8)
234#define BP_SET_CHECKSUM(bp, x) BF64_SET((bp)->blk_prop, 40, 8, x)
235
236#define BP_GET_TYPE(bp) BF64_GET((bp)->blk_prop, 48, 8)
237#define BP_SET_TYPE(bp, x) BF64_SET((bp)->blk_prop, 48, 8, x)
238
239#define BP_GET_LEVEL(bp) BF64_GET((bp)->blk_prop, 56, 5)
240#define BP_SET_LEVEL(bp, x) BF64_SET((bp)->blk_prop, 56, 5, x)
241
242#define BP_GET_PROP_BIT_61(bp) BF64_GET((bp)->blk_prop, 61, 1)
243#define BP_SET_PROP_BIT_61(bp, x) BF64_SET((bp)->blk_prop, 61, 1, x)
244
245#define BP_GET_DEDUP(bp) BF64_GET((bp)->blk_prop, 62, 1)
246#define BP_SET_DEDUP(bp, x) BF64_SET((bp)->blk_prop, 62, 1, x)
247
248#define BP_GET_BYTEORDER(bp) (0 - BF64_GET((bp)->blk_prop, 63, 1))
249#define BP_SET_BYTEORDER(bp, x) BF64_SET((bp)->blk_prop, 63, 1, x)
250
251#define BP_PHYSICAL_BIRTH(bp) \
252 ((bp)->blk_phys_birth ? (bp)->blk_phys_birth : (bp)->blk_birth)
253
254#define BP_SET_BIRTH(bp, logical, physical) \
255{ \
256 (bp)->blk_birth = (logical); \
257 (bp)->blk_phys_birth = ((logical) == (physical) ? 0 : (physical)); \
258}
259
260#define BP_GET_ASIZE(bp) \
261 (DVA_GET_ASIZE(&(bp)->blk_dva[0]) + DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
262 DVA_GET_ASIZE(&(bp)->blk_dva[2]))
263
264#define BP_GET_UCSIZE(bp) \
265 ((BP_GET_LEVEL(bp) > 0 || DMU_OT_IS_METADATA(BP_GET_TYPE(bp))) ? \
266 BP_GET_PSIZE(bp) : BP_GET_LSIZE(bp))
267
268#define BP_GET_NDVAS(bp) \
269 (!!DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \
270 !!DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
271 !!DVA_GET_ASIZE(&(bp)->blk_dva[2]))
272
273#define BP_COUNT_GANG(bp) \
274 (DVA_GET_GANG(&(bp)->blk_dva[0]) + \
275 DVA_GET_GANG(&(bp)->blk_dva[1]) + \
276 DVA_GET_GANG(&(bp)->blk_dva[2]))
277
278#define DVA_EQUAL(dva1, dva2) \
279 ((dva1)->dva_word[1] == (dva2)->dva_word[1] && \
280 (dva1)->dva_word[0] == (dva2)->dva_word[0])
281
282#define BP_EQUAL(bp1, bp2) \
283 (BP_PHYSICAL_BIRTH(bp1) == BP_PHYSICAL_BIRTH(bp2) && \
284 DVA_EQUAL(&(bp1)->blk_dva[0], &(bp2)->blk_dva[0]) && \
285 DVA_EQUAL(&(bp1)->blk_dva[1], &(bp2)->blk_dva[1]) && \
286 DVA_EQUAL(&(bp1)->blk_dva[2], &(bp2)->blk_dva[2]))
287
288#define ZIO_CHECKSUM_EQUAL(zc1, zc2) \
289 (0 == (((zc1).zc_word[0] - (zc2).zc_word[0]) | \
290 ((zc1).zc_word[1] - (zc2).zc_word[1]) | \
291 ((zc1).zc_word[2] - (zc2).zc_word[2]) | \
292 ((zc1).zc_word[3] - (zc2).zc_word[3])))
293
294#define DVA_IS_VALID(dva) (DVA_GET_ASIZE(dva) != 0)
295
296#define ZIO_SET_CHECKSUM(zcp, w0, w1, w2, w3) \
297{ \
298 (zcp)->zc_word[0] = w0; \
299 (zcp)->zc_word[1] = w1; \
300 (zcp)->zc_word[2] = w2; \
301 (zcp)->zc_word[3] = w3; \
302}
303
304#define BP_IDENTITY(bp) (&(bp)->blk_dva[0])
305#define BP_IS_GANG(bp) DVA_GET_GANG(BP_IDENTITY(bp))
306#define BP_IS_HOLE(bp) ((bp)->blk_birth == 0)
307
308/* BP_IS_RAIDZ(bp) assumes no block compression */
309#define BP_IS_RAIDZ(bp) (DVA_GET_ASIZE(&(bp)->blk_dva[0]) > \
310 BP_GET_PSIZE(bp))
311
312#define BP_ZERO(bp) \
313{ \
314 (bp)->blk_dva[0].dva_word[0] = 0; \
315 (bp)->blk_dva[0].dva_word[1] = 0; \
316 (bp)->blk_dva[1].dva_word[0] = 0; \
317 (bp)->blk_dva[1].dva_word[1] = 0; \
318 (bp)->blk_dva[2].dva_word[0] = 0; \
319 (bp)->blk_dva[2].dva_word[1] = 0; \
320 (bp)->blk_prop = 0; \
321 (bp)->blk_pad[0] = 0; \
322 (bp)->blk_pad[1] = 0; \
323 (bp)->blk_phys_birth = 0; \
324 (bp)->blk_birth = 0; \
325 (bp)->blk_fill = 0; \
326 ZIO_SET_CHECKSUM(&(bp)->blk_cksum, 0, 0, 0, 0); \
327}
328
329/*
330 * Note: the byteorder is either 0 or -1, both of which are palindromes.
331 * This simplifies the endianness handling a bit.
332 */
333#if BYTE_ORDER == _BIG_ENDIAN
334#define ZFS_HOST_BYTEORDER (0ULL)
335#else
336#define ZFS_HOST_BYTEORDER (-1ULL)
337#endif
338
339#define BP_SHOULD_BYTESWAP(bp) (BP_GET_BYTEORDER(bp) != ZFS_HOST_BYTEORDER)
340
341#define BP_SPRINTF_LEN 320
342
343/*
344 * This macro allows code sharing between zfs, libzpool, and mdb.
345 * 'func' is either snprintf() or mdb_snprintf().
346 * 'ws' (whitespace) can be ' ' for single-line format, '\n' for multi-line.
347 */
348#define SPRINTF_BLKPTR(func, ws, buf, bp, type, checksum, compress) \
349{ \
350 static const char *copyname[] = \
351 { "zero", "single", "double", "triple" }; \
352 int size = BP_SPRINTF_LEN; \
353 int len = 0; \
354 int copies = 0; \
355 \
356 if (bp == NULL) { \
357 len = func(buf + len, size - len, "<NULL>"); \
358 } else if (BP_IS_HOLE(bp)) { \
359 len = func(buf + len, size - len, "<hole>"); \
360 } else { \
361 for (int d = 0; d < BP_GET_NDVAS(bp); d++) { \
362 const dva_t *dva = &bp->blk_dva[d]; \
363 if (DVA_IS_VALID(dva)) \
364 copies++; \
365 len += func(buf + len, size - len, \
366 "DVA[%d]=<%llu:%llx:%llx>%c", d, \
367 (u_longlong_t)DVA_GET_VDEV(dva), \
368 (u_longlong_t)DVA_GET_OFFSET(dva), \
369 (u_longlong_t)DVA_GET_ASIZE(dva), \
370 ws); \
371 } \
372 if (BP_IS_GANG(bp) && \
373 DVA_GET_ASIZE(&bp->blk_dva[2]) <= \
374 DVA_GET_ASIZE(&bp->blk_dva[1]) / 2) \
375 copies--; \
376 len += func(buf + len, size - len, \
377 "[L%llu %s] %s %s %s %s %s %s%c" \
378 "size=%llxL/%llxP birth=%lluL/%lluP fill=%llu%c" \
379 "cksum=%llx:%llx:%llx:%llx", \
380 (u_longlong_t)BP_GET_LEVEL(bp), \
381 type, \
382 checksum, \
383 compress, \
384 BP_GET_BYTEORDER(bp) == 0 ? "BE" : "LE", \
385 BP_IS_GANG(bp) ? "gang" : "contiguous", \
386 BP_GET_DEDUP(bp) ? "dedup" : "unique", \
387 copyname[copies], \
388 ws, \
389 (u_longlong_t)BP_GET_LSIZE(bp), \
390 (u_longlong_t)BP_GET_PSIZE(bp), \
391 (u_longlong_t)bp->blk_birth, \
392 (u_longlong_t)BP_PHYSICAL_BIRTH(bp), \
393 (u_longlong_t)bp->blk_fill, \
394 ws, \
395 (u_longlong_t)bp->blk_cksum.zc_word[0], \
396 (u_longlong_t)bp->blk_cksum.zc_word[1], \
397 (u_longlong_t)bp->blk_cksum.zc_word[2], \
398 (u_longlong_t)bp->blk_cksum.zc_word[3]); \
399 } \
400 ASSERT(len < size); \
401}
402
403#include <sys/dmu.h>
404
405#define BP_GET_BUFC_TYPE(bp) \
406 (((BP_GET_LEVEL(bp) > 0) || (DMU_OT_IS_METADATA(BP_GET_TYPE(bp)))) ? \
407 ARC_BUFC_METADATA : ARC_BUFC_DATA)
408
409typedef enum spa_import_type {
410 SPA_IMPORT_EXISTING,
411 SPA_IMPORT_ASSEMBLE
412} spa_import_type_t;
413
414/* state manipulation functions */
415extern int spa_open(const char *pool, spa_t **, void *tag);
416extern int spa_open_rewind(const char *pool, spa_t **, void *tag,
417 nvlist_t *policy, nvlist_t **config);
418extern int spa_get_stats(const char *pool, nvlist_t **config, char *altroot,
419 size_t buflen);
420extern int spa_create(const char *pool, nvlist_t *config, nvlist_t *props,
421 const char *history_str, nvlist_t *zplprops);
| 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
38extern "C" {
39#endif
40
41/*
42 * Forward references that lots of things need.
43 */
44typedef struct spa spa_t;
45typedef struct vdev vdev_t;
46typedef struct metaslab metaslab_t;
47typedef struct metaslab_group metaslab_group_t;
48typedef struct metaslab_class metaslab_class_t;
49typedef struct zio zio_t;
50typedef struct zilog zilog_t;
51typedef struct spa_aux_vdev spa_aux_vdev_t;
52typedef struct ddt ddt_t;
53typedef struct ddt_entry ddt_entry_t;
54struct dsl_pool;
55
56/*
57 * General-purpose 32-bit and 64-bit bitfield encodings.
58 */
59#define BF32_DECODE(x, low, len) P2PHASE((x) >> (low), 1U << (len))
60#define BF64_DECODE(x, low, len) P2PHASE((x) >> (low), 1ULL << (len))
61#define BF32_ENCODE(x, low, len) (P2PHASE((x), 1U << (len)) << (low))
62#define BF64_ENCODE(x, low, len) (P2PHASE((x), 1ULL << (len)) << (low))
63
64#define BF32_GET(x, low, len) BF32_DECODE(x, low, len)
65#define BF64_GET(x, low, len) BF64_DECODE(x, low, len)
66
67#define BF32_SET(x, low, len, val) \
68 ((x) ^= BF32_ENCODE((x >> low) ^ (val), low, len))
69#define BF64_SET(x, low, len, val) \
70 ((x) ^= BF64_ENCODE((x >> low) ^ (val), low, len))
71
72#define BF32_GET_SB(x, low, len, shift, bias) \
73 ((BF32_GET(x, low, len) + (bias)) << (shift))
74#define BF64_GET_SB(x, low, len, shift, bias) \
75 ((BF64_GET(x, low, len) + (bias)) << (shift))
76
77#define BF32_SET_SB(x, low, len, shift, bias, val) \
78 BF32_SET(x, low, len, ((val) >> (shift)) - (bias))
79#define BF64_SET_SB(x, low, len, shift, bias, val) \
80 BF64_SET(x, low, len, ((val) >> (shift)) - (bias))
81
82/*
83 * We currently support nine block sizes, from 512 bytes to 128K.
84 * We could go higher, but the benefits are near-zero and the cost
85 * of COWing a giant block to modify one byte would become excessive.
86 */
87#define SPA_MINBLOCKSHIFT 9
88#define SPA_MAXBLOCKSHIFT 17
89#define SPA_MINBLOCKSIZE (1ULL << SPA_MINBLOCKSHIFT)
90#define SPA_MAXBLOCKSIZE (1ULL << SPA_MAXBLOCKSHIFT)
91
92#define SPA_BLOCKSIZES (SPA_MAXBLOCKSHIFT - SPA_MINBLOCKSHIFT + 1)
93
94/*
95 * Size of block to hold the configuration data (a packed nvlist)
96 */
97#define SPA_CONFIG_BLOCKSIZE (1ULL << 14)
98
99/*
100 * The DVA size encodings for LSIZE and PSIZE support blocks up to 32MB.
101 * The ASIZE encoding should be at least 64 times larger (6 more bits)
102 * to support up to 4-way RAID-Z mirror mode with worst-case gang block
103 * overhead, three DVAs per bp, plus one more bit in case we do anything
104 * else that expands the ASIZE.
105 */
106#define SPA_LSIZEBITS 16 /* LSIZE up to 32M (2^16 * 512) */
107#define SPA_PSIZEBITS 16 /* PSIZE up to 32M (2^16 * 512) */
108#define SPA_ASIZEBITS 24 /* ASIZE up to 64 times larger */
109
110/*
111 * All SPA data is represented by 128-bit data virtual addresses (DVAs).
112 * The members of the dva_t should be considered opaque outside the SPA.
113 */
114typedef struct dva {
115 uint64_t dva_word[2];
116} dva_t;
117
118/*
119 * Each block has a 256-bit checksum -- strong enough for cryptographic hashes.
120 */
121typedef struct zio_cksum {
122 uint64_t zc_word[4];
123} zio_cksum_t;
124
125/*
126 * Each block is described by its DVAs, time of birth, checksum, etc.
127 * The word-by-word, bit-by-bit layout of the blkptr is as follows:
128 *
129 * 64 56 48 40 32 24 16 8 0
130 * +-------+-------+-------+-------+-------+-------+-------+-------+
131 * 0 | vdev1 | GRID | ASIZE |
132 * +-------+-------+-------+-------+-------+-------+-------+-------+
133 * 1 |G| offset1 |
134 * +-------+-------+-------+-------+-------+-------+-------+-------+
135 * 2 | vdev2 | GRID | ASIZE |
136 * +-------+-------+-------+-------+-------+-------+-------+-------+
137 * 3 |G| offset2 |
138 * +-------+-------+-------+-------+-------+-------+-------+-------+
139 * 4 | vdev3 | GRID | ASIZE |
140 * +-------+-------+-------+-------+-------+-------+-------+-------+
141 * 5 |G| offset3 |
142 * +-------+-------+-------+-------+-------+-------+-------+-------+
143 * 6 |BDX|lvl| type | cksum | comp | PSIZE | LSIZE |
144 * +-------+-------+-------+-------+-------+-------+-------+-------+
145 * 7 | padding |
146 * +-------+-------+-------+-------+-------+-------+-------+-------+
147 * 8 | padding |
148 * +-------+-------+-------+-------+-------+-------+-------+-------+
149 * 9 | physical birth txg |
150 * +-------+-------+-------+-------+-------+-------+-------+-------+
151 * a | logical birth txg |
152 * +-------+-------+-------+-------+-------+-------+-------+-------+
153 * b | fill count |
154 * +-------+-------+-------+-------+-------+-------+-------+-------+
155 * c | checksum[0] |
156 * +-------+-------+-------+-------+-------+-------+-------+-------+
157 * d | checksum[1] |
158 * +-------+-------+-------+-------+-------+-------+-------+-------+
159 * e | checksum[2] |
160 * +-------+-------+-------+-------+-------+-------+-------+-------+
161 * f | checksum[3] |
162 * +-------+-------+-------+-------+-------+-------+-------+-------+
163 *
164 * Legend:
165 *
166 * vdev virtual device ID
167 * offset offset into virtual device
168 * LSIZE logical size
169 * PSIZE physical size (after compression)
170 * ASIZE allocated size (including RAID-Z parity and gang block headers)
171 * GRID RAID-Z layout information (reserved for future use)
172 * cksum checksum function
173 * comp compression function
174 * G gang block indicator
175 * B byteorder (endianness)
176 * D dedup
177 * X unused
178 * lvl level of indirection
179 * type DMU object type
180 * phys birth txg of block allocation; zero if same as logical birth txg
181 * log. birth transaction group in which the block was logically born
182 * fill count number of non-zero blocks under this bp
183 * checksum[4] 256-bit checksum of the data this bp describes
184 */
185#define SPA_BLKPTRSHIFT 7 /* blkptr_t is 128 bytes */
186#define SPA_DVAS_PER_BP 3 /* Number of DVAs in a bp */
187
188typedef struct blkptr {
189 dva_t blk_dva[SPA_DVAS_PER_BP]; /* Data Virtual Addresses */
190 uint64_t blk_prop; /* size, compression, type, etc */
191 uint64_t blk_pad[2]; /* Extra space for the future */
192 uint64_t blk_phys_birth; /* txg when block was allocated */
193 uint64_t blk_birth; /* transaction group at birth */
194 uint64_t blk_fill; /* fill count */
195 zio_cksum_t blk_cksum; /* 256-bit checksum */
196} blkptr_t;
197
198/*
199 * Macros to get and set fields in a bp or DVA.
200 */
201#define DVA_GET_ASIZE(dva) \
202 BF64_GET_SB((dva)->dva_word[0], 0, 24, SPA_MINBLOCKSHIFT, 0)
203#define DVA_SET_ASIZE(dva, x) \
204 BF64_SET_SB((dva)->dva_word[0], 0, 24, SPA_MINBLOCKSHIFT, 0, x)
205
206#define DVA_GET_GRID(dva) BF64_GET((dva)->dva_word[0], 24, 8)
207#define DVA_SET_GRID(dva, x) BF64_SET((dva)->dva_word[0], 24, 8, x)
208
209#define DVA_GET_VDEV(dva) BF64_GET((dva)->dva_word[0], 32, 32)
210#define DVA_SET_VDEV(dva, x) BF64_SET((dva)->dva_word[0], 32, 32, x)
211
212#define DVA_GET_OFFSET(dva) \
213 BF64_GET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0)
214#define DVA_SET_OFFSET(dva, x) \
215 BF64_SET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0, x)
216
217#define DVA_GET_GANG(dva) BF64_GET((dva)->dva_word[1], 63, 1)
218#define DVA_SET_GANG(dva, x) BF64_SET((dva)->dva_word[1], 63, 1, x)
219
220#define BP_GET_LSIZE(bp) \
221 BF64_GET_SB((bp)->blk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1)
222#define BP_SET_LSIZE(bp, x) \
223 BF64_SET_SB((bp)->blk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1, x)
224
225#define BP_GET_PSIZE(bp) \
226 BF64_GET_SB((bp)->blk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1)
227#define BP_SET_PSIZE(bp, x) \
228 BF64_SET_SB((bp)->blk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1, x)
229
230#define BP_GET_COMPRESS(bp) BF64_GET((bp)->blk_prop, 32, 8)
231#define BP_SET_COMPRESS(bp, x) BF64_SET((bp)->blk_prop, 32, 8, x)
232
233#define BP_GET_CHECKSUM(bp) BF64_GET((bp)->blk_prop, 40, 8)
234#define BP_SET_CHECKSUM(bp, x) BF64_SET((bp)->blk_prop, 40, 8, x)
235
236#define BP_GET_TYPE(bp) BF64_GET((bp)->blk_prop, 48, 8)
237#define BP_SET_TYPE(bp, x) BF64_SET((bp)->blk_prop, 48, 8, x)
238
239#define BP_GET_LEVEL(bp) BF64_GET((bp)->blk_prop, 56, 5)
240#define BP_SET_LEVEL(bp, x) BF64_SET((bp)->blk_prop, 56, 5, x)
241
242#define BP_GET_PROP_BIT_61(bp) BF64_GET((bp)->blk_prop, 61, 1)
243#define BP_SET_PROP_BIT_61(bp, x) BF64_SET((bp)->blk_prop, 61, 1, x)
244
245#define BP_GET_DEDUP(bp) BF64_GET((bp)->blk_prop, 62, 1)
246#define BP_SET_DEDUP(bp, x) BF64_SET((bp)->blk_prop, 62, 1, x)
247
248#define BP_GET_BYTEORDER(bp) (0 - BF64_GET((bp)->blk_prop, 63, 1))
249#define BP_SET_BYTEORDER(bp, x) BF64_SET((bp)->blk_prop, 63, 1, x)
250
251#define BP_PHYSICAL_BIRTH(bp) \
252 ((bp)->blk_phys_birth ? (bp)->blk_phys_birth : (bp)->blk_birth)
253
254#define BP_SET_BIRTH(bp, logical, physical) \
255{ \
256 (bp)->blk_birth = (logical); \
257 (bp)->blk_phys_birth = ((logical) == (physical) ? 0 : (physical)); \
258}
259
260#define BP_GET_ASIZE(bp) \
261 (DVA_GET_ASIZE(&(bp)->blk_dva[0]) + DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
262 DVA_GET_ASIZE(&(bp)->blk_dva[2]))
263
264#define BP_GET_UCSIZE(bp) \
265 ((BP_GET_LEVEL(bp) > 0 || DMU_OT_IS_METADATA(BP_GET_TYPE(bp))) ? \
266 BP_GET_PSIZE(bp) : BP_GET_LSIZE(bp))
267
268#define BP_GET_NDVAS(bp) \
269 (!!DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \
270 !!DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
271 !!DVA_GET_ASIZE(&(bp)->blk_dva[2]))
272
273#define BP_COUNT_GANG(bp) \
274 (DVA_GET_GANG(&(bp)->blk_dva[0]) + \
275 DVA_GET_GANG(&(bp)->blk_dva[1]) + \
276 DVA_GET_GANG(&(bp)->blk_dva[2]))
277
278#define DVA_EQUAL(dva1, dva2) \
279 ((dva1)->dva_word[1] == (dva2)->dva_word[1] && \
280 (dva1)->dva_word[0] == (dva2)->dva_word[0])
281
282#define BP_EQUAL(bp1, bp2) \
283 (BP_PHYSICAL_BIRTH(bp1) == BP_PHYSICAL_BIRTH(bp2) && \
284 DVA_EQUAL(&(bp1)->blk_dva[0], &(bp2)->blk_dva[0]) && \
285 DVA_EQUAL(&(bp1)->blk_dva[1], &(bp2)->blk_dva[1]) && \
286 DVA_EQUAL(&(bp1)->blk_dva[2], &(bp2)->blk_dva[2]))
287
288#define ZIO_CHECKSUM_EQUAL(zc1, zc2) \
289 (0 == (((zc1).zc_word[0] - (zc2).zc_word[0]) | \
290 ((zc1).zc_word[1] - (zc2).zc_word[1]) | \
291 ((zc1).zc_word[2] - (zc2).zc_word[2]) | \
292 ((zc1).zc_word[3] - (zc2).zc_word[3])))
293
294#define DVA_IS_VALID(dva) (DVA_GET_ASIZE(dva) != 0)
295
296#define ZIO_SET_CHECKSUM(zcp, w0, w1, w2, w3) \
297{ \
298 (zcp)->zc_word[0] = w0; \
299 (zcp)->zc_word[1] = w1; \
300 (zcp)->zc_word[2] = w2; \
301 (zcp)->zc_word[3] = w3; \
302}
303
304#define BP_IDENTITY(bp) (&(bp)->blk_dva[0])
305#define BP_IS_GANG(bp) DVA_GET_GANG(BP_IDENTITY(bp))
306#define BP_IS_HOLE(bp) ((bp)->blk_birth == 0)
307
308/* BP_IS_RAIDZ(bp) assumes no block compression */
309#define BP_IS_RAIDZ(bp) (DVA_GET_ASIZE(&(bp)->blk_dva[0]) > \
310 BP_GET_PSIZE(bp))
311
312#define BP_ZERO(bp) \
313{ \
314 (bp)->blk_dva[0].dva_word[0] = 0; \
315 (bp)->blk_dva[0].dva_word[1] = 0; \
316 (bp)->blk_dva[1].dva_word[0] = 0; \
317 (bp)->blk_dva[1].dva_word[1] = 0; \
318 (bp)->blk_dva[2].dva_word[0] = 0; \
319 (bp)->blk_dva[2].dva_word[1] = 0; \
320 (bp)->blk_prop = 0; \
321 (bp)->blk_pad[0] = 0; \
322 (bp)->blk_pad[1] = 0; \
323 (bp)->blk_phys_birth = 0; \
324 (bp)->blk_birth = 0; \
325 (bp)->blk_fill = 0; \
326 ZIO_SET_CHECKSUM(&(bp)->blk_cksum, 0, 0, 0, 0); \
327}
328
329/*
330 * Note: the byteorder is either 0 or -1, both of which are palindromes.
331 * This simplifies the endianness handling a bit.
332 */
333#if BYTE_ORDER == _BIG_ENDIAN
334#define ZFS_HOST_BYTEORDER (0ULL)
335#else
336#define ZFS_HOST_BYTEORDER (-1ULL)
337#endif
338
339#define BP_SHOULD_BYTESWAP(bp) (BP_GET_BYTEORDER(bp) != ZFS_HOST_BYTEORDER)
340
341#define BP_SPRINTF_LEN 320
342
343/*
344 * This macro allows code sharing between zfs, libzpool, and mdb.
345 * 'func' is either snprintf() or mdb_snprintf().
346 * 'ws' (whitespace) can be ' ' for single-line format, '\n' for multi-line.
347 */
348#define SPRINTF_BLKPTR(func, ws, buf, bp, type, checksum, compress) \
349{ \
350 static const char *copyname[] = \
351 { "zero", "single", "double", "triple" }; \
352 int size = BP_SPRINTF_LEN; \
353 int len = 0; \
354 int copies = 0; \
355 \
356 if (bp == NULL) { \
357 len = func(buf + len, size - len, "<NULL>"); \
358 } else if (BP_IS_HOLE(bp)) { \
359 len = func(buf + len, size - len, "<hole>"); \
360 } else { \
361 for (int d = 0; d < BP_GET_NDVAS(bp); d++) { \
362 const dva_t *dva = &bp->blk_dva[d]; \
363 if (DVA_IS_VALID(dva)) \
364 copies++; \
365 len += func(buf + len, size - len, \
366 "DVA[%d]=<%llu:%llx:%llx>%c", d, \
367 (u_longlong_t)DVA_GET_VDEV(dva), \
368 (u_longlong_t)DVA_GET_OFFSET(dva), \
369 (u_longlong_t)DVA_GET_ASIZE(dva), \
370 ws); \
371 } \
372 if (BP_IS_GANG(bp) && \
373 DVA_GET_ASIZE(&bp->blk_dva[2]) <= \
374 DVA_GET_ASIZE(&bp->blk_dva[1]) / 2) \
375 copies--; \
376 len += func(buf + len, size - len, \
377 "[L%llu %s] %s %s %s %s %s %s%c" \
378 "size=%llxL/%llxP birth=%lluL/%lluP fill=%llu%c" \
379 "cksum=%llx:%llx:%llx:%llx", \
380 (u_longlong_t)BP_GET_LEVEL(bp), \
381 type, \
382 checksum, \
383 compress, \
384 BP_GET_BYTEORDER(bp) == 0 ? "BE" : "LE", \
385 BP_IS_GANG(bp) ? "gang" : "contiguous", \
386 BP_GET_DEDUP(bp) ? "dedup" : "unique", \
387 copyname[copies], \
388 ws, \
389 (u_longlong_t)BP_GET_LSIZE(bp), \
390 (u_longlong_t)BP_GET_PSIZE(bp), \
391 (u_longlong_t)bp->blk_birth, \
392 (u_longlong_t)BP_PHYSICAL_BIRTH(bp), \
393 (u_longlong_t)bp->blk_fill, \
394 ws, \
395 (u_longlong_t)bp->blk_cksum.zc_word[0], \
396 (u_longlong_t)bp->blk_cksum.zc_word[1], \
397 (u_longlong_t)bp->blk_cksum.zc_word[2], \
398 (u_longlong_t)bp->blk_cksum.zc_word[3]); \
399 } \
400 ASSERT(len < size); \
401}
402
403#include <sys/dmu.h>
404
405#define BP_GET_BUFC_TYPE(bp) \
406 (((BP_GET_LEVEL(bp) > 0) || (DMU_OT_IS_METADATA(BP_GET_TYPE(bp)))) ? \
407 ARC_BUFC_METADATA : ARC_BUFC_DATA)
408
409typedef enum spa_import_type {
410 SPA_IMPORT_EXISTING,
411 SPA_IMPORT_ASSEMBLE
412} spa_import_type_t;
413
414/* state manipulation functions */
415extern int spa_open(const char *pool, spa_t **, void *tag);
416extern int spa_open_rewind(const char *pool, spa_t **, void *tag,
417 nvlist_t *policy, nvlist_t **config);
418extern int spa_get_stats(const char *pool, nvlist_t **config, char *altroot,
419 size_t buflen);
420extern int spa_create(const char *pool, nvlist_t *config, nvlist_t *props,
421 const char *history_str, nvlist_t *zplprops);
|
423extern int spa_import(const char *pool, nvlist_t *config, nvlist_t *props,
424 uint64_t flags);
425extern nvlist_t *spa_tryimport(nvlist_t *tryconfig);
426extern int spa_destroy(char *pool);
427extern int spa_export(char *pool, nvlist_t **oldconfig, boolean_t force,
428 boolean_t hardforce);
429extern int spa_reset(char *pool);
430extern void spa_async_request(spa_t *spa, int flag);
431extern void spa_async_unrequest(spa_t *spa, int flag);
432extern void spa_async_suspend(spa_t *spa);
433extern void spa_async_resume(spa_t *spa);
434extern spa_t *spa_inject_addref(char *pool);
435extern void spa_inject_delref(spa_t *spa);
436extern void spa_scan_stat_init(spa_t *spa);
437extern int spa_scan_get_stats(spa_t *spa, pool_scan_stat_t *ps);
438
439#define SPA_ASYNC_CONFIG_UPDATE 0x01
440#define SPA_ASYNC_REMOVE 0x02
441#define SPA_ASYNC_PROBE 0x04
442#define SPA_ASYNC_RESILVER_DONE 0x08
443#define SPA_ASYNC_RESILVER 0x10
444#define SPA_ASYNC_AUTOEXPAND 0x20
445#define SPA_ASYNC_REMOVE_DONE 0x40
446#define SPA_ASYNC_REMOVE_STOP 0x80
447
448/*
449 * Controls the behavior of spa_vdev_remove().
450 */
451#define SPA_REMOVE_UNSPARE 0x01
452#define SPA_REMOVE_DONE 0x02
453
454/* device manipulation */
455extern int spa_vdev_add(spa_t *spa, nvlist_t *nvroot);
456extern int spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot,
457 int replacing);
458extern int spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid,
459 int replace_done);
460extern int spa_vdev_remove(spa_t *spa, uint64_t guid, boolean_t unspare);
461extern boolean_t spa_vdev_remove_active(spa_t *spa);
462extern int spa_vdev_setpath(spa_t *spa, uint64_t guid, const char *newpath);
463extern int spa_vdev_setfru(spa_t *spa, uint64_t guid, const char *newfru);
464extern int spa_vdev_split_mirror(spa_t *spa, char *newname, nvlist_t *config,
465 nvlist_t *props, boolean_t exp);
466
467/* spare state (which is global across all pools) */
468extern void spa_spare_add(vdev_t *vd);
469extern void spa_spare_remove(vdev_t *vd);
470extern boolean_t spa_spare_exists(uint64_t guid, uint64_t *pool, int *refcnt);
471extern void spa_spare_activate(vdev_t *vd);
472
473/* L2ARC state (which is global across all pools) */
474extern void spa_l2cache_add(vdev_t *vd);
475extern void spa_l2cache_remove(vdev_t *vd);
476extern boolean_t spa_l2cache_exists(uint64_t guid, uint64_t *pool);
477extern void spa_l2cache_activate(vdev_t *vd);
478extern void spa_l2cache_drop(spa_t *spa);
479
480/* scanning */
481extern int spa_scan(spa_t *spa, pool_scan_func_t func);
482extern int spa_scan_stop(spa_t *spa);
483
484/* spa syncing */
485extern void spa_sync(spa_t *spa, uint64_t txg); /* only for DMU use */
486extern void spa_sync_allpools(void);
487
488/*
489 * DEFERRED_FREE must be large enough that regular blocks are not
490 * deferred. XXX so can't we change it back to 1?
491 */
492#define SYNC_PASS_DEFERRED_FREE 2 /* defer frees after this pass */
493#define SYNC_PASS_DONT_COMPRESS 4 /* don't compress after this pass */
494#define SYNC_PASS_REWRITE 1 /* rewrite new bps after this pass */
495
496/* spa namespace global mutex */
497extern kmutex_t spa_namespace_lock;
498
499/*
500 * SPA configuration functions in spa_config.c
501 */
502
503#define SPA_CONFIG_UPDATE_POOL 0
504#define SPA_CONFIG_UPDATE_VDEVS 1
505
506extern void spa_config_sync(spa_t *, boolean_t, boolean_t);
507extern void spa_config_load(void);
508extern nvlist_t *spa_all_configs(uint64_t *);
509extern void spa_config_set(spa_t *spa, nvlist_t *config);
510extern nvlist_t *spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg,
511 int getstats);
512extern void spa_config_update(spa_t *spa, int what);
513
514/*
515 * Miscellaneous SPA routines in spa_misc.c
516 */
517
518/* Namespace manipulation */
519extern spa_t *spa_lookup(const char *name);
520extern spa_t *spa_add(const char *name, nvlist_t *config, const char *altroot);
521extern void spa_remove(spa_t *spa);
522extern spa_t *spa_next(spa_t *prev);
523
524/* Refcount functions */
525extern void spa_open_ref(spa_t *spa, void *tag);
526extern void spa_close(spa_t *spa, void *tag);
527extern boolean_t spa_refcount_zero(spa_t *spa);
528
529#define SCL_NONE 0x00
530#define SCL_CONFIG 0x01
531#define SCL_STATE 0x02
532#define SCL_L2ARC 0x04 /* hack until L2ARC 2.0 */
533#define SCL_ALLOC 0x08
534#define SCL_ZIO 0x10
535#define SCL_FREE 0x20
536#define SCL_VDEV 0x40
537#define SCL_LOCKS 7
538#define SCL_ALL ((1 << SCL_LOCKS) - 1)
539#define SCL_STATE_ALL (SCL_STATE | SCL_L2ARC | SCL_ZIO)
540
541/* Pool configuration locks */
542extern int spa_config_tryenter(spa_t *spa, int locks, void *tag, krw_t rw);
543extern void spa_config_enter(spa_t *spa, int locks, void *tag, krw_t rw);
544extern void spa_config_exit(spa_t *spa, int locks, void *tag);
545extern int spa_config_held(spa_t *spa, int locks, krw_t rw);
546
547/* Pool vdev add/remove lock */
548extern uint64_t spa_vdev_enter(spa_t *spa);
549extern uint64_t spa_vdev_config_enter(spa_t *spa);
550extern void spa_vdev_config_exit(spa_t *spa, vdev_t *vd, uint64_t txg,
551 int error, char *tag);
552extern int spa_vdev_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error);
553
554/* Pool vdev state change lock */
555extern void spa_vdev_state_enter(spa_t *spa, int oplock);
556extern int spa_vdev_state_exit(spa_t *spa, vdev_t *vd, int error);
557
558/* Log state */
559typedef enum spa_log_state {
560 SPA_LOG_UNKNOWN = 0, /* unknown log state */
561 SPA_LOG_MISSING, /* missing log(s) */
562 SPA_LOG_CLEAR, /* clear the log(s) */
563 SPA_LOG_GOOD, /* log(s) are good */
564} spa_log_state_t;
565
566extern spa_log_state_t spa_get_log_state(spa_t *spa);
567extern void spa_set_log_state(spa_t *spa, spa_log_state_t state);
568extern int spa_offline_log(spa_t *spa);
569
570/* Log claim callback */
571extern void spa_claim_notify(zio_t *zio);
572
573/* Accessor functions */
574extern boolean_t spa_shutting_down(spa_t *spa);
575extern struct dsl_pool *spa_get_dsl(spa_t *spa);
576extern boolean_t spa_is_initializing(spa_t *spa);
577extern blkptr_t *spa_get_rootblkptr(spa_t *spa);
578extern void spa_set_rootblkptr(spa_t *spa, const blkptr_t *bp);
579extern void spa_altroot(spa_t *, char *, size_t);
580extern int spa_sync_pass(spa_t *spa);
581extern char *spa_name(spa_t *spa);
582extern uint64_t spa_guid(spa_t *spa);
583extern uint64_t spa_load_guid(spa_t *spa);
584extern uint64_t spa_last_synced_txg(spa_t *spa);
585extern uint64_t spa_first_txg(spa_t *spa);
586extern uint64_t spa_syncing_txg(spa_t *spa);
587extern uint64_t spa_version(spa_t *spa);
588extern pool_state_t spa_state(spa_t *spa);
589extern spa_load_state_t spa_load_state(spa_t *spa);
590extern uint64_t spa_freeze_txg(spa_t *spa);
591extern uint64_t spa_get_asize(spa_t *spa, uint64_t lsize);
592extern uint64_t spa_get_dspace(spa_t *spa);
593extern void spa_update_dspace(spa_t *spa);
594extern uint64_t spa_version(spa_t *spa);
595extern boolean_t spa_deflate(spa_t *spa);
596extern metaslab_class_t *spa_normal_class(spa_t *spa);
597extern metaslab_class_t *spa_log_class(spa_t *spa);
598extern int spa_max_replication(spa_t *spa);
599extern int spa_prev_software_version(spa_t *spa);
600extern int spa_busy(void);
601extern uint8_t spa_get_failmode(spa_t *spa);
602extern boolean_t spa_suspended(spa_t *spa);
603extern uint64_t spa_bootfs(spa_t *spa);
604extern uint64_t spa_delegation(spa_t *spa);
605extern objset_t *spa_meta_objset(spa_t *spa);
606
607/* Miscellaneous support routines */
608extern void spa_activate_mos_feature(spa_t *spa, const char *feature);
609extern void spa_deactivate_mos_feature(spa_t *spa, const char *feature);
610extern int spa_rename(const char *oldname, const char *newname);
611extern spa_t *spa_by_guid(uint64_t pool_guid, uint64_t device_guid);
612extern boolean_t spa_guid_exists(uint64_t pool_guid, uint64_t device_guid);
613extern char *spa_strdup(const char *);
614extern void spa_strfree(char *);
615extern uint64_t spa_get_random(uint64_t range);
616extern uint64_t spa_generate_guid(spa_t *spa);
617extern void sprintf_blkptr(char *buf, const blkptr_t *bp);
618extern void spa_freeze(spa_t *spa);
619extern int spa_change_guid(spa_t *spa);
620extern void spa_upgrade(spa_t *spa, uint64_t version);
621extern void spa_evict_all(void);
622extern vdev_t *spa_lookup_by_guid(spa_t *spa, uint64_t guid,
623 boolean_t l2cache);
624extern boolean_t spa_has_spare(spa_t *, uint64_t guid);
625extern uint64_t dva_get_dsize_sync(spa_t *spa, const dva_t *dva);
626extern uint64_t bp_get_dsize_sync(spa_t *spa, const blkptr_t *bp);
627extern uint64_t bp_get_dsize(spa_t *spa, const blkptr_t *bp);
628extern boolean_t spa_has_slogs(spa_t *spa);
629extern boolean_t spa_is_root(spa_t *spa);
630extern boolean_t spa_writeable(spa_t *spa);
631
632extern int spa_mode(spa_t *spa);
633extern uint64_t zfs_strtonum(const char *str, char **nptr);
634#define strtonum(str, nptr) zfs_strtonum((str), (nptr))
635
636/* history logging */
637typedef enum history_log_type {
638 LOG_CMD_POOL_CREATE,
639 LOG_CMD_NORMAL,
640 LOG_INTERNAL
641} history_log_type_t;
642
643typedef struct history_arg {
644 char *ha_history_str;
645 history_log_type_t ha_log_type;
646 history_internal_events_t ha_event;
647 char *ha_zone;
648 uid_t ha_uid;
649} history_arg_t;
650
651extern char *spa_his_ievent_table[];
652
653extern void spa_history_create_obj(spa_t *spa, dmu_tx_t *tx);
654extern int spa_history_get(spa_t *spa, uint64_t *offset, uint64_t *len_read,
655 char *his_buf);
656extern int spa_history_log(spa_t *spa, const char *his_buf,
657 history_log_type_t what);
658extern void spa_history_log_internal(history_internal_events_t event,
659 spa_t *spa, dmu_tx_t *tx, const char *fmt, ...);
660extern void spa_history_log_version(spa_t *spa, history_internal_events_t evt);
661
662/* error handling */
663struct zbookmark;
664extern void spa_log_error(spa_t *spa, zio_t *zio);
665extern void zfs_ereport_post(const char *cls, spa_t *spa, vdev_t *vd,
666 zio_t *zio, uint64_t stateoroffset, uint64_t length);
667extern void zfs_post_remove(spa_t *spa, vdev_t *vd);
668extern void zfs_post_state_change(spa_t *spa, vdev_t *vd);
669extern void zfs_post_autoreplace(spa_t *spa, vdev_t *vd);
670extern uint64_t spa_get_errlog_size(spa_t *spa);
671extern int spa_get_errlog(spa_t *spa, void *uaddr, size_t *count);
672extern void spa_errlog_rotate(spa_t *spa);
673extern void spa_errlog_drain(spa_t *spa);
674extern void spa_errlog_sync(spa_t *spa, uint64_t txg);
675extern void spa_get_errlists(spa_t *spa, avl_tree_t *last, avl_tree_t *scrub);
676
677/* vdev cache */
678extern void vdev_cache_stat_init(void);
679extern void vdev_cache_stat_fini(void);
680
681/* Initialization and termination */
682extern void spa_init(int flags);
683extern void spa_fini(void);
684extern void spa_boot_init();
685
686/* properties */
687extern int spa_prop_set(spa_t *spa, nvlist_t *nvp);
688extern int spa_prop_get(spa_t *spa, nvlist_t **nvp);
689extern void spa_prop_clear_bootfs(spa_t *spa, uint64_t obj, dmu_tx_t *tx);
690extern void spa_configfile_set(spa_t *, nvlist_t *, boolean_t);
691
692/* asynchronous event notification */
693extern void spa_event_notify(spa_t *spa, vdev_t *vdev, const char *name);
694
695#ifdef ZFS_DEBUG
696#define dprintf_bp(bp, fmt, ...) do { \
697 if (zfs_flags & ZFS_DEBUG_DPRINTF) { \
698 char *__blkbuf = kmem_alloc(BP_SPRINTF_LEN, KM_SLEEP); \
699 sprintf_blkptr(__blkbuf, (bp)); \
700 dprintf(fmt " %s\n", __VA_ARGS__, __blkbuf); \
701 kmem_free(__blkbuf, BP_SPRINTF_LEN); \
702 } \
703_NOTE(CONSTCOND) } while (0)
704#else
705#define dprintf_bp(bp, fmt, ...)
706#endif
707
708extern boolean_t spa_debug_enabled(spa_t *spa);
709#define spa_dbgmsg(spa, ...) \
710{ \
711 if (spa_debug_enabled(spa)) \
712 zfs_dbgmsg(__VA_ARGS__); \
713}
714
715extern int spa_mode_global; /* mode, e.g. FREAD | FWRITE */
716
717#ifdef __cplusplus
718}
719#endif
720
721#endif /* _SYS_SPA_H */
| 427extern int spa_import(const char *pool, nvlist_t *config, nvlist_t *props,
428 uint64_t flags);
429extern nvlist_t *spa_tryimport(nvlist_t *tryconfig);
430extern int spa_destroy(char *pool);
431extern int spa_export(char *pool, nvlist_t **oldconfig, boolean_t force,
432 boolean_t hardforce);
433extern int spa_reset(char *pool);
434extern void spa_async_request(spa_t *spa, int flag);
435extern void spa_async_unrequest(spa_t *spa, int flag);
436extern void spa_async_suspend(spa_t *spa);
437extern void spa_async_resume(spa_t *spa);
438extern spa_t *spa_inject_addref(char *pool);
439extern void spa_inject_delref(spa_t *spa);
440extern void spa_scan_stat_init(spa_t *spa);
441extern int spa_scan_get_stats(spa_t *spa, pool_scan_stat_t *ps);
442
443#define SPA_ASYNC_CONFIG_UPDATE 0x01
444#define SPA_ASYNC_REMOVE 0x02
445#define SPA_ASYNC_PROBE 0x04
446#define SPA_ASYNC_RESILVER_DONE 0x08
447#define SPA_ASYNC_RESILVER 0x10
448#define SPA_ASYNC_AUTOEXPAND 0x20
449#define SPA_ASYNC_REMOVE_DONE 0x40
450#define SPA_ASYNC_REMOVE_STOP 0x80
451
452/*
453 * Controls the behavior of spa_vdev_remove().
454 */
455#define SPA_REMOVE_UNSPARE 0x01
456#define SPA_REMOVE_DONE 0x02
457
458/* device manipulation */
459extern int spa_vdev_add(spa_t *spa, nvlist_t *nvroot);
460extern int spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot,
461 int replacing);
462extern int spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid,
463 int replace_done);
464extern int spa_vdev_remove(spa_t *spa, uint64_t guid, boolean_t unspare);
465extern boolean_t spa_vdev_remove_active(spa_t *spa);
466extern int spa_vdev_setpath(spa_t *spa, uint64_t guid, const char *newpath);
467extern int spa_vdev_setfru(spa_t *spa, uint64_t guid, const char *newfru);
468extern int spa_vdev_split_mirror(spa_t *spa, char *newname, nvlist_t *config,
469 nvlist_t *props, boolean_t exp);
470
471/* spare state (which is global across all pools) */
472extern void spa_spare_add(vdev_t *vd);
473extern void spa_spare_remove(vdev_t *vd);
474extern boolean_t spa_spare_exists(uint64_t guid, uint64_t *pool, int *refcnt);
475extern void spa_spare_activate(vdev_t *vd);
476
477/* L2ARC state (which is global across all pools) */
478extern void spa_l2cache_add(vdev_t *vd);
479extern void spa_l2cache_remove(vdev_t *vd);
480extern boolean_t spa_l2cache_exists(uint64_t guid, uint64_t *pool);
481extern void spa_l2cache_activate(vdev_t *vd);
482extern void spa_l2cache_drop(spa_t *spa);
483
484/* scanning */
485extern int spa_scan(spa_t *spa, pool_scan_func_t func);
486extern int spa_scan_stop(spa_t *spa);
487
488/* spa syncing */
489extern void spa_sync(spa_t *spa, uint64_t txg); /* only for DMU use */
490extern void spa_sync_allpools(void);
491
492/*
493 * DEFERRED_FREE must be large enough that regular blocks are not
494 * deferred. XXX so can't we change it back to 1?
495 */
496#define SYNC_PASS_DEFERRED_FREE 2 /* defer frees after this pass */
497#define SYNC_PASS_DONT_COMPRESS 4 /* don't compress after this pass */
498#define SYNC_PASS_REWRITE 1 /* rewrite new bps after this pass */
499
500/* spa namespace global mutex */
501extern kmutex_t spa_namespace_lock;
502
503/*
504 * SPA configuration functions in spa_config.c
505 */
506
507#define SPA_CONFIG_UPDATE_POOL 0
508#define SPA_CONFIG_UPDATE_VDEVS 1
509
510extern void spa_config_sync(spa_t *, boolean_t, boolean_t);
511extern void spa_config_load(void);
512extern nvlist_t *spa_all_configs(uint64_t *);
513extern void spa_config_set(spa_t *spa, nvlist_t *config);
514extern nvlist_t *spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg,
515 int getstats);
516extern void spa_config_update(spa_t *spa, int what);
517
518/*
519 * Miscellaneous SPA routines in spa_misc.c
520 */
521
522/* Namespace manipulation */
523extern spa_t *spa_lookup(const char *name);
524extern spa_t *spa_add(const char *name, nvlist_t *config, const char *altroot);
525extern void spa_remove(spa_t *spa);
526extern spa_t *spa_next(spa_t *prev);
527
528/* Refcount functions */
529extern void spa_open_ref(spa_t *spa, void *tag);
530extern void spa_close(spa_t *spa, void *tag);
531extern boolean_t spa_refcount_zero(spa_t *spa);
532
533#define SCL_NONE 0x00
534#define SCL_CONFIG 0x01
535#define SCL_STATE 0x02
536#define SCL_L2ARC 0x04 /* hack until L2ARC 2.0 */
537#define SCL_ALLOC 0x08
538#define SCL_ZIO 0x10
539#define SCL_FREE 0x20
540#define SCL_VDEV 0x40
541#define SCL_LOCKS 7
542#define SCL_ALL ((1 << SCL_LOCKS) - 1)
543#define SCL_STATE_ALL (SCL_STATE | SCL_L2ARC | SCL_ZIO)
544
545/* Pool configuration locks */
546extern int spa_config_tryenter(spa_t *spa, int locks, void *tag, krw_t rw);
547extern void spa_config_enter(spa_t *spa, int locks, void *tag, krw_t rw);
548extern void spa_config_exit(spa_t *spa, int locks, void *tag);
549extern int spa_config_held(spa_t *spa, int locks, krw_t rw);
550
551/* Pool vdev add/remove lock */
552extern uint64_t spa_vdev_enter(spa_t *spa);
553extern uint64_t spa_vdev_config_enter(spa_t *spa);
554extern void spa_vdev_config_exit(spa_t *spa, vdev_t *vd, uint64_t txg,
555 int error, char *tag);
556extern int spa_vdev_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error);
557
558/* Pool vdev state change lock */
559extern void spa_vdev_state_enter(spa_t *spa, int oplock);
560extern int spa_vdev_state_exit(spa_t *spa, vdev_t *vd, int error);
561
562/* Log state */
563typedef enum spa_log_state {
564 SPA_LOG_UNKNOWN = 0, /* unknown log state */
565 SPA_LOG_MISSING, /* missing log(s) */
566 SPA_LOG_CLEAR, /* clear the log(s) */
567 SPA_LOG_GOOD, /* log(s) are good */
568} spa_log_state_t;
569
570extern spa_log_state_t spa_get_log_state(spa_t *spa);
571extern void spa_set_log_state(spa_t *spa, spa_log_state_t state);
572extern int spa_offline_log(spa_t *spa);
573
574/* Log claim callback */
575extern void spa_claim_notify(zio_t *zio);
576
577/* Accessor functions */
578extern boolean_t spa_shutting_down(spa_t *spa);
579extern struct dsl_pool *spa_get_dsl(spa_t *spa);
580extern boolean_t spa_is_initializing(spa_t *spa);
581extern blkptr_t *spa_get_rootblkptr(spa_t *spa);
582extern void spa_set_rootblkptr(spa_t *spa, const blkptr_t *bp);
583extern void spa_altroot(spa_t *, char *, size_t);
584extern int spa_sync_pass(spa_t *spa);
585extern char *spa_name(spa_t *spa);
586extern uint64_t spa_guid(spa_t *spa);
587extern uint64_t spa_load_guid(spa_t *spa);
588extern uint64_t spa_last_synced_txg(spa_t *spa);
589extern uint64_t spa_first_txg(spa_t *spa);
590extern uint64_t spa_syncing_txg(spa_t *spa);
591extern uint64_t spa_version(spa_t *spa);
592extern pool_state_t spa_state(spa_t *spa);
593extern spa_load_state_t spa_load_state(spa_t *spa);
594extern uint64_t spa_freeze_txg(spa_t *spa);
595extern uint64_t spa_get_asize(spa_t *spa, uint64_t lsize);
596extern uint64_t spa_get_dspace(spa_t *spa);
597extern void spa_update_dspace(spa_t *spa);
598extern uint64_t spa_version(spa_t *spa);
599extern boolean_t spa_deflate(spa_t *spa);
600extern metaslab_class_t *spa_normal_class(spa_t *spa);
601extern metaslab_class_t *spa_log_class(spa_t *spa);
602extern int spa_max_replication(spa_t *spa);
603extern int spa_prev_software_version(spa_t *spa);
604extern int spa_busy(void);
605extern uint8_t spa_get_failmode(spa_t *spa);
606extern boolean_t spa_suspended(spa_t *spa);
607extern uint64_t spa_bootfs(spa_t *spa);
608extern uint64_t spa_delegation(spa_t *spa);
609extern objset_t *spa_meta_objset(spa_t *spa);
610
611/* Miscellaneous support routines */
612extern void spa_activate_mos_feature(spa_t *spa, const char *feature);
613extern void spa_deactivate_mos_feature(spa_t *spa, const char *feature);
614extern int spa_rename(const char *oldname, const char *newname);
615extern spa_t *spa_by_guid(uint64_t pool_guid, uint64_t device_guid);
616extern boolean_t spa_guid_exists(uint64_t pool_guid, uint64_t device_guid);
617extern char *spa_strdup(const char *);
618extern void spa_strfree(char *);
619extern uint64_t spa_get_random(uint64_t range);
620extern uint64_t spa_generate_guid(spa_t *spa);
621extern void sprintf_blkptr(char *buf, const blkptr_t *bp);
622extern void spa_freeze(spa_t *spa);
623extern int spa_change_guid(spa_t *spa);
624extern void spa_upgrade(spa_t *spa, uint64_t version);
625extern void spa_evict_all(void);
626extern vdev_t *spa_lookup_by_guid(spa_t *spa, uint64_t guid,
627 boolean_t l2cache);
628extern boolean_t spa_has_spare(spa_t *, uint64_t guid);
629extern uint64_t dva_get_dsize_sync(spa_t *spa, const dva_t *dva);
630extern uint64_t bp_get_dsize_sync(spa_t *spa, const blkptr_t *bp);
631extern uint64_t bp_get_dsize(spa_t *spa, const blkptr_t *bp);
632extern boolean_t spa_has_slogs(spa_t *spa);
633extern boolean_t spa_is_root(spa_t *spa);
634extern boolean_t spa_writeable(spa_t *spa);
635
636extern int spa_mode(spa_t *spa);
637extern uint64_t zfs_strtonum(const char *str, char **nptr);
638#define strtonum(str, nptr) zfs_strtonum((str), (nptr))
639
640/* history logging */
641typedef enum history_log_type {
642 LOG_CMD_POOL_CREATE,
643 LOG_CMD_NORMAL,
644 LOG_INTERNAL
645} history_log_type_t;
646
647typedef struct history_arg {
648 char *ha_history_str;
649 history_log_type_t ha_log_type;
650 history_internal_events_t ha_event;
651 char *ha_zone;
652 uid_t ha_uid;
653} history_arg_t;
654
655extern char *spa_his_ievent_table[];
656
657extern void spa_history_create_obj(spa_t *spa, dmu_tx_t *tx);
658extern int spa_history_get(spa_t *spa, uint64_t *offset, uint64_t *len_read,
659 char *his_buf);
660extern int spa_history_log(spa_t *spa, const char *his_buf,
661 history_log_type_t what);
662extern void spa_history_log_internal(history_internal_events_t event,
663 spa_t *spa, dmu_tx_t *tx, const char *fmt, ...);
664extern void spa_history_log_version(spa_t *spa, history_internal_events_t evt);
665
666/* error handling */
667struct zbookmark;
668extern void spa_log_error(spa_t *spa, zio_t *zio);
669extern void zfs_ereport_post(const char *cls, spa_t *spa, vdev_t *vd,
670 zio_t *zio, uint64_t stateoroffset, uint64_t length);
671extern void zfs_post_remove(spa_t *spa, vdev_t *vd);
672extern void zfs_post_state_change(spa_t *spa, vdev_t *vd);
673extern void zfs_post_autoreplace(spa_t *spa, vdev_t *vd);
674extern uint64_t spa_get_errlog_size(spa_t *spa);
675extern int spa_get_errlog(spa_t *spa, void *uaddr, size_t *count);
676extern void spa_errlog_rotate(spa_t *spa);
677extern void spa_errlog_drain(spa_t *spa);
678extern void spa_errlog_sync(spa_t *spa, uint64_t txg);
679extern void spa_get_errlists(spa_t *spa, avl_tree_t *last, avl_tree_t *scrub);
680
681/* vdev cache */
682extern void vdev_cache_stat_init(void);
683extern void vdev_cache_stat_fini(void);
684
685/* Initialization and termination */
686extern void spa_init(int flags);
687extern void spa_fini(void);
688extern void spa_boot_init();
689
690/* properties */
691extern int spa_prop_set(spa_t *spa, nvlist_t *nvp);
692extern int spa_prop_get(spa_t *spa, nvlist_t **nvp);
693extern void spa_prop_clear_bootfs(spa_t *spa, uint64_t obj, dmu_tx_t *tx);
694extern void spa_configfile_set(spa_t *, nvlist_t *, boolean_t);
695
696/* asynchronous event notification */
697extern void spa_event_notify(spa_t *spa, vdev_t *vdev, const char *name);
698
699#ifdef ZFS_DEBUG
700#define dprintf_bp(bp, fmt, ...) do { \
701 if (zfs_flags & ZFS_DEBUG_DPRINTF) { \
702 char *__blkbuf = kmem_alloc(BP_SPRINTF_LEN, KM_SLEEP); \
703 sprintf_blkptr(__blkbuf, (bp)); \
704 dprintf(fmt " %s\n", __VA_ARGS__, __blkbuf); \
705 kmem_free(__blkbuf, BP_SPRINTF_LEN); \
706 } \
707_NOTE(CONSTCOND) } while (0)
708#else
709#define dprintf_bp(bp, fmt, ...)
710#endif
711
712extern boolean_t spa_debug_enabled(spa_t *spa);
713#define spa_dbgmsg(spa, ...) \
714{ \
715 if (spa_debug_enabled(spa)) \
716 zfs_dbgmsg(__VA_ARGS__); \
717}
718
719extern int spa_mode_global; /* mode, e.g. FREAD | FWRITE */
720
721#ifdef __cplusplus
722}
723#endif
724
725#endif /* _SYS_SPA_H */
|