spa.h revision f9af39bacaaa0f9dda3b75ff6858b9f3988a39af
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 */
24
25#ifndef _SYS_SPA_H
26#define	_SYS_SPA_H
27
28#include <sys/avl.h>
29#include <sys/zfs_context.h>
30#include <sys/nvpair.h>
31#include <sys/sysmacros.h>
32#include <sys/types.h>
33#include <sys/fs/zfs.h>
34
35#ifdef	__cplusplus
36extern "C" {
37#endif
38
39/*
40 * Forward references that lots of things need.
41 */
42typedef struct spa spa_t;
43typedef struct vdev vdev_t;
44typedef struct metaslab metaslab_t;
45typedef struct metaslab_group metaslab_group_t;
46typedef struct metaslab_class metaslab_class_t;
47typedef struct zio zio_t;
48typedef struct zilog zilog_t;
49typedef struct spa_aux_vdev spa_aux_vdev_t;
50typedef struct ddt ddt_t;
51typedef struct ddt_entry ddt_entry_t;
52struct dsl_pool;
53
54/*
55 * General-purpose 32-bit and 64-bit bitfield encodings.
56 */
57#define	BF32_DECODE(x, low, len)	P2PHASE((x) >> (low), 1U << (len))
58#define	BF64_DECODE(x, low, len)	P2PHASE((x) >> (low), 1ULL << (len))
59#define	BF32_ENCODE(x, low, len)	(P2PHASE((x), 1U << (len)) << (low))
60#define	BF64_ENCODE(x, low, len)	(P2PHASE((x), 1ULL << (len)) << (low))
61
62#define	BF32_GET(x, low, len)		BF32_DECODE(x, low, len)
63#define	BF64_GET(x, low, len)		BF64_DECODE(x, low, len)
64
65#define	BF32_SET(x, low, len, val)	\
66	((x) ^= BF32_ENCODE((x >> low) ^ (val), low, len))
67#define	BF64_SET(x, low, len, val)	\
68	((x) ^= BF64_ENCODE((x >> low) ^ (val), low, len))
69
70#define	BF32_GET_SB(x, low, len, shift, bias)	\
71	((BF32_GET(x, low, len) + (bias)) << (shift))
72#define	BF64_GET_SB(x, low, len, shift, bias)	\
73	((BF64_GET(x, low, len) + (bias)) << (shift))
74
75#define	BF32_SET_SB(x, low, len, shift, bias, val)	\
76	BF32_SET(x, low, len, ((val) >> (shift)) - (bias))
77#define	BF64_SET_SB(x, low, len, shift, bias, val)	\
78	BF64_SET(x, low, len, ((val) >> (shift)) - (bias))
79
80/*
81 * We currently support nine block sizes, from 512 bytes to 128K.
82 * We could go higher, but the benefits are near-zero and the cost
83 * of COWing a giant block to modify one byte would become excessive.
84 */
85#define	SPA_MINBLOCKSHIFT	9
86#define	SPA_MAXBLOCKSHIFT	17
87#define	SPA_MINBLOCKSIZE	(1ULL << SPA_MINBLOCKSHIFT)
88#define	SPA_MAXBLOCKSIZE	(1ULL << SPA_MAXBLOCKSHIFT)
89
90#define	SPA_BLOCKSIZES		(SPA_MAXBLOCKSHIFT - SPA_MINBLOCKSHIFT + 1)
91
92/*
93 * Size of block to hold the configuration data (a packed nvlist)
94 */
95#define	SPA_CONFIG_BLOCKSIZE	(1 << 14)
96
97/*
98 * The DVA size encodings for LSIZE and PSIZE support blocks up to 32MB.
99 * The ASIZE encoding should be at least 64 times larger (6 more bits)
100 * to support up to 4-way RAID-Z mirror mode with worst-case gang block
101 * overhead, three DVAs per bp, plus one more bit in case we do anything
102 * else that expands the ASIZE.
103 */
104#define	SPA_LSIZEBITS		16	/* LSIZE up to 32M (2^16 * 512)	*/
105#define	SPA_PSIZEBITS		16	/* PSIZE up to 32M (2^16 * 512)	*/
106#define	SPA_ASIZEBITS		24	/* ASIZE up to 64 times larger	*/
107
108/*
109 * All SPA data is represented by 128-bit data virtual addresses (DVAs).
110 * The members of the dva_t should be considered opaque outside the SPA.
111 */
112typedef struct dva {
113	uint64_t	dva_word[2];
114} dva_t;
115
116/*
117 * Each block has a 256-bit checksum -- strong enough for cryptographic hashes.
118 */
119typedef struct zio_cksum {
120	uint64_t	zc_word[4];
121} zio_cksum_t;
122
123/*
124 * Each block is described by its DVAs, time of birth, checksum, etc.
125 * The word-by-word, bit-by-bit layout of the blkptr is as follows:
126 *
127 *	64	56	48	40	32	24	16	8	0
128 *	+-------+-------+-------+-------+-------+-------+-------+-------+
129 * 0	|		vdev1		| GRID  |	  ASIZE		|
130 *	+-------+-------+-------+-------+-------+-------+-------+-------+
131 * 1	|G|			 offset1				|
132 *	+-------+-------+-------+-------+-------+-------+-------+-------+
133 * 2	|		vdev2		| GRID  |	  ASIZE		|
134 *	+-------+-------+-------+-------+-------+-------+-------+-------+
135 * 3	|G|			 offset2				|
136 *	+-------+-------+-------+-------+-------+-------+-------+-------+
137 * 4	|		vdev3		| GRID  |	  ASIZE		|
138 *	+-------+-------+-------+-------+-------+-------+-------+-------+
139 * 5	|G|			 offset3				|
140 *	+-------+-------+-------+-------+-------+-------+-------+-------+
141 * 6	|BDX|lvl| type	| cksum | comp	|     PSIZE	|     LSIZE	|
142 *	+-------+-------+-------+-------+-------+-------+-------+-------+
143 * 7	|			padding					|
144 *	+-------+-------+-------+-------+-------+-------+-------+-------+
145 * 8	|			padding					|
146 *	+-------+-------+-------+-------+-------+-------+-------+-------+
147 * 9	|			physical birth txg			|
148 *	+-------+-------+-------+-------+-------+-------+-------+-------+
149 * a	|			logical birth txg			|
150 *	+-------+-------+-------+-------+-------+-------+-------+-------+
151 * b	|			fill count				|
152 *	+-------+-------+-------+-------+-------+-------+-------+-------+
153 * c	|			checksum[0]				|
154 *	+-------+-------+-------+-------+-------+-------+-------+-------+
155 * d	|			checksum[1]				|
156 *	+-------+-------+-------+-------+-------+-------+-------+-------+
157 * e	|			checksum[2]				|
158 *	+-------+-------+-------+-------+-------+-------+-------+-------+
159 * f	|			checksum[3]				|
160 *	+-------+-------+-------+-------+-------+-------+-------+-------+
161 *
162 * Legend:
163 *
164 * vdev		virtual device ID
165 * offset	offset into virtual device
166 * LSIZE	logical size
167 * PSIZE	physical size (after compression)
168 * ASIZE	allocated size (including RAID-Z parity and gang block headers)
169 * GRID		RAID-Z layout information (reserved for future use)
170 * cksum	checksum function
171 * comp		compression function
172 * G		gang block indicator
173 * B		byteorder (endianness)
174 * D		dedup
175 * X		unused
176 * lvl		level of indirection
177 * type		DMU object type
178 * phys birth	txg of block allocation; zero if same as logical birth txg
179 * log. birth	transaction group in which the block was logically born
180 * fill count	number of non-zero blocks under this bp
181 * checksum[4]	256-bit checksum of the data this bp describes
182 */
183#define	SPA_BLKPTRSHIFT	7		/* blkptr_t is 128 bytes	*/
184#define	SPA_DVAS_PER_BP	3		/* Number of DVAs in a bp	*/
185
186typedef struct blkptr {
187	dva_t		blk_dva[SPA_DVAS_PER_BP]; /* Data Virtual Addresses */
188	uint64_t	blk_prop;	/* size, compression, type, etc	    */
189	uint64_t	blk_pad[2];	/* Extra space for the future	    */
190	uint64_t	blk_phys_birth;	/* txg when block was allocated	    */
191	uint64_t	blk_birth;	/* transaction group at birth	    */
192	uint64_t	blk_fill;	/* fill count			    */
193	zio_cksum_t	blk_cksum;	/* 256-bit checksum		    */
194} blkptr_t;
195
196/*
197 * Macros to get and set fields in a bp or DVA.
198 */
199#define	DVA_GET_ASIZE(dva)	\
200	BF64_GET_SB((dva)->dva_word[0], 0, 24, SPA_MINBLOCKSHIFT, 0)
201#define	DVA_SET_ASIZE(dva, x)	\
202	BF64_SET_SB((dva)->dva_word[0], 0, 24, SPA_MINBLOCKSHIFT, 0, x)
203
204#define	DVA_GET_GRID(dva)	BF64_GET((dva)->dva_word[0], 24, 8)
205#define	DVA_SET_GRID(dva, x)	BF64_SET((dva)->dva_word[0], 24, 8, x)
206
207#define	DVA_GET_VDEV(dva)	BF64_GET((dva)->dva_word[0], 32, 32)
208#define	DVA_SET_VDEV(dva, x)	BF64_SET((dva)->dva_word[0], 32, 32, x)
209
210#define	DVA_GET_OFFSET(dva)	\
211	BF64_GET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0)
212#define	DVA_SET_OFFSET(dva, x)	\
213	BF64_SET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0, x)
214
215#define	DVA_GET_GANG(dva)	BF64_GET((dva)->dva_word[1], 63, 1)
216#define	DVA_SET_GANG(dva, x)	BF64_SET((dva)->dva_word[1], 63, 1, x)
217
218#define	BP_GET_LSIZE(bp)	\
219	BF64_GET_SB((bp)->blk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1)
220#define	BP_SET_LSIZE(bp, x)	\
221	BF64_SET_SB((bp)->blk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1, x)
222
223#define	BP_GET_PSIZE(bp)	\
224	BF64_GET_SB((bp)->blk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1)
225#define	BP_SET_PSIZE(bp, x)	\
226	BF64_SET_SB((bp)->blk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1, x)
227
228#define	BP_GET_COMPRESS(bp)		BF64_GET((bp)->blk_prop, 32, 8)
229#define	BP_SET_COMPRESS(bp, x)		BF64_SET((bp)->blk_prop, 32, 8, x)
230
231#define	BP_GET_CHECKSUM(bp)		BF64_GET((bp)->blk_prop, 40, 8)
232#define	BP_SET_CHECKSUM(bp, x)		BF64_SET((bp)->blk_prop, 40, 8, x)
233
234#define	BP_GET_TYPE(bp)			BF64_GET((bp)->blk_prop, 48, 8)
235#define	BP_SET_TYPE(bp, x)		BF64_SET((bp)->blk_prop, 48, 8, x)
236
237#define	BP_GET_LEVEL(bp)		BF64_GET((bp)->blk_prop, 56, 5)
238#define	BP_SET_LEVEL(bp, x)		BF64_SET((bp)->blk_prop, 56, 5, x)
239
240#define	BP_GET_PROP_BIT_61(bp)		BF64_GET((bp)->blk_prop, 61, 1)
241#define	BP_SET_PROP_BIT_61(bp, x)	BF64_SET((bp)->blk_prop, 61, 1, x)
242
243#define	BP_GET_DEDUP(bp)		BF64_GET((bp)->blk_prop, 62, 1)
244#define	BP_SET_DEDUP(bp, x)		BF64_SET((bp)->blk_prop, 62, 1, x)
245
246#define	BP_GET_BYTEORDER(bp)		(0 - BF64_GET((bp)->blk_prop, 63, 1))
247#define	BP_SET_BYTEORDER(bp, x)		BF64_SET((bp)->blk_prop, 63, 1, x)
248
249#define	BP_PHYSICAL_BIRTH(bp)		\
250	((bp)->blk_phys_birth ? (bp)->blk_phys_birth : (bp)->blk_birth)
251
252#define	BP_SET_BIRTH(bp, logical, physical)	\
253{						\
254	(bp)->blk_birth = (logical);		\
255	(bp)->blk_phys_birth = ((logical) == (physical) ? 0 : (physical)); \
256}
257
258#define	BP_GET_ASIZE(bp)	\
259	(DVA_GET_ASIZE(&(bp)->blk_dva[0]) + DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
260		DVA_GET_ASIZE(&(bp)->blk_dva[2]))
261
262#define	BP_GET_UCSIZE(bp) \
263	((BP_GET_LEVEL(bp) > 0 || dmu_ot[BP_GET_TYPE(bp)].ot_metadata) ? \
264	BP_GET_PSIZE(bp) : BP_GET_LSIZE(bp))
265
266#define	BP_GET_NDVAS(bp)	\
267	(!!DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \
268	!!DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
269	!!DVA_GET_ASIZE(&(bp)->blk_dva[2]))
270
271#define	BP_COUNT_GANG(bp)	\
272	(DVA_GET_GANG(&(bp)->blk_dva[0]) + \
273	DVA_GET_GANG(&(bp)->blk_dva[1]) + \
274	DVA_GET_GANG(&(bp)->blk_dva[2]))
275
276#define	DVA_EQUAL(dva1, dva2)	\
277	((dva1)->dva_word[1] == (dva2)->dva_word[1] && \
278	(dva1)->dva_word[0] == (dva2)->dva_word[0])
279
280#define	BP_EQUAL(bp1, bp2)	\
281	(BP_PHYSICAL_BIRTH(bp1) == BP_PHYSICAL_BIRTH(bp2) &&	\
282	DVA_EQUAL(&(bp1)->blk_dva[0], &(bp2)->blk_dva[0]) &&	\
283	DVA_EQUAL(&(bp1)->blk_dva[1], &(bp2)->blk_dva[1]) &&	\
284	DVA_EQUAL(&(bp1)->blk_dva[2], &(bp2)->blk_dva[2]))
285
286#define	ZIO_CHECKSUM_EQUAL(zc1, zc2) \
287	(0 == (((zc1).zc_word[0] - (zc2).zc_word[0]) | \
288	((zc1).zc_word[1] - (zc2).zc_word[1]) | \
289	((zc1).zc_word[2] - (zc2).zc_word[2]) | \
290	((zc1).zc_word[3] - (zc2).zc_word[3])))
291
292#define	DVA_IS_VALID(dva)	(DVA_GET_ASIZE(dva) != 0)
293
294#define	ZIO_SET_CHECKSUM(zcp, w0, w1, w2, w3)	\
295{						\
296	(zcp)->zc_word[0] = w0;			\
297	(zcp)->zc_word[1] = w1;			\
298	(zcp)->zc_word[2] = w2;			\
299	(zcp)->zc_word[3] = w3;			\
300}
301
302#define	BP_IDENTITY(bp)		(&(bp)->blk_dva[0])
303#define	BP_IS_GANG(bp)		DVA_GET_GANG(BP_IDENTITY(bp))
304#define	BP_IS_HOLE(bp)		((bp)->blk_birth == 0)
305
306/* BP_IS_RAIDZ(bp) assumes no block compression */
307#define	BP_IS_RAIDZ(bp)		(DVA_GET_ASIZE(&(bp)->blk_dva[0]) > \
308				BP_GET_PSIZE(bp))
309
310#define	BP_ZERO(bp)				\
311{						\
312	(bp)->blk_dva[0].dva_word[0] = 0;	\
313	(bp)->blk_dva[0].dva_word[1] = 0;	\
314	(bp)->blk_dva[1].dva_word[0] = 0;	\
315	(bp)->blk_dva[1].dva_word[1] = 0;	\
316	(bp)->blk_dva[2].dva_word[0] = 0;	\
317	(bp)->blk_dva[2].dva_word[1] = 0;	\
318	(bp)->blk_prop = 0;			\
319	(bp)->blk_pad[0] = 0;			\
320	(bp)->blk_pad[1] = 0;			\
321	(bp)->blk_phys_birth = 0;		\
322	(bp)->blk_birth = 0;			\
323	(bp)->blk_fill = 0;			\
324	ZIO_SET_CHECKSUM(&(bp)->blk_cksum, 0, 0, 0, 0);	\
325}
326
327/*
328 * Note: the byteorder is either 0 or -1, both of which are palindromes.
329 * This simplifies the endianness handling a bit.
330 */
331#ifdef _BIG_ENDIAN
332#define	ZFS_HOST_BYTEORDER	(0ULL)
333#else
334#define	ZFS_HOST_BYTEORDER	(-1ULL)
335#endif
336
337#define	BP_SHOULD_BYTESWAP(bp)	(BP_GET_BYTEORDER(bp) != ZFS_HOST_BYTEORDER)
338
339#define	BP_SPRINTF_LEN	320
340
341/*
342 * This macro allows code sharing between zfs, libzpool, and mdb.
343 * 'func' is either snprintf() or mdb_snprintf().
344 * 'ws' (whitespace) can be ' ' for single-line format, '\n' for multi-line.
345 */
346#define	SPRINTF_BLKPTR(func, ws, buf, bp, type, checksum, compress)	\
347{									\
348	static const char *copyname[] =					\
349	    { "zero", "single", "double", "triple" };			\
350	int size = BP_SPRINTF_LEN;					\
351	int len = 0;							\
352	int copies = 0;							\
353									\
354	if (bp == NULL) {						\
355		len = func(buf + len, size - len, "<NULL>");		\
356	} else if (BP_IS_HOLE(bp)) {					\
357		len = func(buf + len, size - len, "<hole>");		\
358	} else {							\
359		for (int d = 0; d < BP_GET_NDVAS(bp); d++) {		\
360			const dva_t *dva = &bp->blk_dva[d];		\
361			if (DVA_IS_VALID(dva))				\
362				copies++;				\
363			len += func(buf + len, size - len,		\
364			    "DVA[%d]=<%llu:%llx:%llx>%c", d,		\
365			    (u_longlong_t)DVA_GET_VDEV(dva),		\
366			    (u_longlong_t)DVA_GET_OFFSET(dva),		\
367			    (u_longlong_t)DVA_GET_ASIZE(dva),		\
368			    ws);					\
369		}							\
370		if (BP_IS_GANG(bp) &&					\
371		    DVA_GET_ASIZE(&bp->blk_dva[2]) <=			\
372		    DVA_GET_ASIZE(&bp->blk_dva[1]) / 2)			\
373			copies--;					\
374		len += func(buf + len, size - len,			\
375		    "[L%llu %s] %s %s %s %s %s %s%c"			\
376		    "size=%llxL/%llxP birth=%lluL/%lluP fill=%llu%c"	\
377		    "cksum=%llx:%llx:%llx:%llx",			\
378		    (u_longlong_t)BP_GET_LEVEL(bp),			\
379		    type,						\
380		    checksum,						\
381		    compress,						\
382		    BP_GET_BYTEORDER(bp) == 0 ? "BE" : "LE",		\
383		    BP_IS_GANG(bp) ? "gang" : "contiguous",		\
384		    BP_GET_DEDUP(bp) ? "dedup" : "unique",		\
385		    copyname[copies],					\
386		    ws,							\
387		    (u_longlong_t)BP_GET_LSIZE(bp),			\
388		    (u_longlong_t)BP_GET_PSIZE(bp),			\
389		    (u_longlong_t)bp->blk_birth,			\
390		    (u_longlong_t)BP_PHYSICAL_BIRTH(bp),		\
391		    (u_longlong_t)bp->blk_fill,				\
392		    ws,							\
393		    (u_longlong_t)bp->blk_cksum.zc_word[0],		\
394		    (u_longlong_t)bp->blk_cksum.zc_word[1],		\
395		    (u_longlong_t)bp->blk_cksum.zc_word[2],		\
396		    (u_longlong_t)bp->blk_cksum.zc_word[3]);		\
397	}								\
398	ASSERT(len < size);						\
399}
400
401#include <sys/dmu.h>
402
403#define	BP_GET_BUFC_TYPE(bp)						\
404	(((BP_GET_LEVEL(bp) > 0) || (dmu_ot[BP_GET_TYPE(bp)].ot_metadata)) ? \
405	ARC_BUFC_METADATA : ARC_BUFC_DATA);
406
407typedef enum spa_import_type {
408	SPA_IMPORT_EXISTING,
409	SPA_IMPORT_ASSEMBLE
410} spa_import_type_t;
411
412/* state manipulation functions */
413extern int spa_open(const char *pool, spa_t **, void *tag);
414extern int spa_open_rewind(const char *pool, spa_t **, void *tag,
415    nvlist_t *policy, nvlist_t **config);
416extern int spa_get_stats(const char *pool, nvlist_t **config,
417    char *altroot, size_t buflen);
418extern int spa_create(const char *pool, nvlist_t *config, nvlist_t *props,
419    const char *history_str, nvlist_t *zplprops);
420extern int spa_import_rootpool(char *devpath, char *devid);
421extern int spa_import(const char *pool, nvlist_t *config, nvlist_t *props,
422    uint64_t flags);
423extern nvlist_t *spa_tryimport(nvlist_t *tryconfig);
424extern int spa_destroy(char *pool);
425extern int spa_export(char *pool, nvlist_t **oldconfig, boolean_t force,
426    boolean_t hardforce);
427extern int spa_reset(char *pool);
428extern void spa_async_request(spa_t *spa, int flag);
429extern void spa_async_unrequest(spa_t *spa, int flag);
430extern void spa_async_suspend(spa_t *spa);
431extern void spa_async_resume(spa_t *spa);
432extern spa_t *spa_inject_addref(char *pool);
433extern void spa_inject_delref(spa_t *spa);
434extern void spa_scan_stat_init(spa_t *spa);
435extern int spa_scan_get_stats(spa_t *spa, pool_scan_stat_t *ps);
436
437#define	SPA_ASYNC_CONFIG_UPDATE	0x01
438#define	SPA_ASYNC_REMOVE	0x02
439#define	SPA_ASYNC_PROBE		0x04
440#define	SPA_ASYNC_RESILVER_DONE	0x08
441#define	SPA_ASYNC_RESILVER	0x10
442#define	SPA_ASYNC_AUTOEXPAND	0x20
443#define	SPA_ASYNC_REMOVE_DONE	0x40
444#define	SPA_ASYNC_REMOVE_STOP	0x80
445
446/*
447 * Controls the behavior of spa_vdev_remove().
448 */
449#define	SPA_REMOVE_UNSPARE	0x01
450#define	SPA_REMOVE_DONE		0x02
451
452/* device manipulation */
453extern int spa_vdev_add(spa_t *spa, nvlist_t *nvroot);
454extern int spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot,
455    int replacing);
456extern int spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid,
457    int replace_done);
458extern int spa_vdev_remove(spa_t *spa, uint64_t guid, boolean_t unspare);
459extern boolean_t spa_vdev_remove_active(spa_t *spa);
460extern int spa_vdev_setpath(spa_t *spa, uint64_t guid, const char *newpath);
461extern int spa_vdev_setfru(spa_t *spa, uint64_t guid, const char *newfru);
462extern int spa_vdev_split_mirror(spa_t *spa, char *newname, nvlist_t *config,
463    nvlist_t *props, boolean_t exp);
464
465/* spare state (which is global across all pools) */
466extern void spa_spare_add(vdev_t *vd);
467extern void spa_spare_remove(vdev_t *vd);
468extern boolean_t spa_spare_exists(uint64_t guid, uint64_t *pool, int *refcnt);
469extern void spa_spare_activate(vdev_t *vd);
470
471/* L2ARC state (which is global across all pools) */
472extern void spa_l2cache_add(vdev_t *vd);
473extern void spa_l2cache_remove(vdev_t *vd);
474extern boolean_t spa_l2cache_exists(uint64_t guid, uint64_t *pool);
475extern void spa_l2cache_activate(vdev_t *vd);
476extern void spa_l2cache_drop(spa_t *spa);
477
478/* scanning */
479extern int spa_scan(spa_t *spa, pool_scan_func_t func);
480extern int spa_scan_stop(spa_t *spa);
481
482/* spa syncing */
483extern void spa_sync(spa_t *spa, uint64_t txg); /* only for DMU use */
484extern void spa_sync_allpools(void);
485
486/*
487 * DEFERRED_FREE must be large enough that regular blocks are not
488 * deferred.  XXX so can't we change it back to 1?
489 */
490#define	SYNC_PASS_DEFERRED_FREE	2	/* defer frees after this pass */
491#define	SYNC_PASS_DONT_COMPRESS	4	/* don't compress after this pass */
492#define	SYNC_PASS_REWRITE	1	/* rewrite new bps after this pass */
493
494/* spa namespace global mutex */
495extern kmutex_t spa_namespace_lock;
496
497/*
498 * SPA configuration functions in spa_config.c
499 */
500
501#define	SPA_CONFIG_UPDATE_POOL	0
502#define	SPA_CONFIG_UPDATE_VDEVS	1
503
504extern void spa_config_sync(spa_t *, boolean_t, boolean_t);
505extern void spa_config_load(void);
506extern nvlist_t *spa_all_configs(uint64_t *);
507extern void spa_config_set(spa_t *spa, nvlist_t *config);
508extern nvlist_t *spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg,
509    int getstats);
510extern void spa_config_update(spa_t *spa, int what);
511
512/*
513 * Miscellaneous SPA routines in spa_misc.c
514 */
515
516/* Namespace manipulation */
517extern spa_t *spa_lookup(const char *name);
518extern spa_t *spa_add(const char *name, nvlist_t *config, const char *altroot);
519extern void spa_remove(spa_t *spa);
520extern spa_t *spa_next(spa_t *prev);
521
522/* Refcount functions */
523extern void spa_open_ref(spa_t *spa, void *tag);
524extern void spa_close(spa_t *spa, void *tag);
525extern boolean_t spa_refcount_zero(spa_t *spa);
526
527#define	SCL_NONE	0x00
528#define	SCL_CONFIG	0x01
529#define	SCL_STATE	0x02
530#define	SCL_L2ARC	0x04		/* hack until L2ARC 2.0 */
531#define	SCL_ALLOC	0x08
532#define	SCL_ZIO		0x10
533#define	SCL_FREE	0x20
534#define	SCL_VDEV	0x40
535#define	SCL_LOCKS	7
536#define	SCL_ALL		((1 << SCL_LOCKS) - 1)
537#define	SCL_STATE_ALL	(SCL_STATE | SCL_L2ARC | SCL_ZIO)
538
539/* Pool configuration locks */
540extern int spa_config_tryenter(spa_t *spa, int locks, void *tag, krw_t rw);
541extern void spa_config_enter(spa_t *spa, int locks, void *tag, krw_t rw);
542extern void spa_config_exit(spa_t *spa, int locks, void *tag);
543extern int spa_config_held(spa_t *spa, int locks, krw_t rw);
544
545/* Pool vdev add/remove lock */
546extern uint64_t spa_vdev_enter(spa_t *spa);
547extern uint64_t spa_vdev_config_enter(spa_t *spa);
548extern void spa_vdev_config_exit(spa_t *spa, vdev_t *vd, uint64_t txg,
549    int error, char *tag);
550extern int spa_vdev_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error);
551
552/* Pool vdev state change lock */
553extern void spa_vdev_state_enter(spa_t *spa, int oplock);
554extern int spa_vdev_state_exit(spa_t *spa, vdev_t *vd, int error);
555
556/* Log state */
557typedef enum spa_log_state {
558	SPA_LOG_UNKNOWN = 0,	/* unknown log state */
559	SPA_LOG_MISSING,	/* missing log(s) */
560	SPA_LOG_CLEAR,		/* clear the log(s) */
561	SPA_LOG_GOOD,		/* log(s) are good */
562} spa_log_state_t;
563
564extern spa_log_state_t spa_get_log_state(spa_t *spa);
565extern void spa_set_log_state(spa_t *spa, spa_log_state_t state);
566extern int spa_offline_log(spa_t *spa);
567
568/* Log claim callback */
569extern void spa_claim_notify(zio_t *zio);
570
571/* Accessor functions */
572extern boolean_t spa_shutting_down(spa_t *spa);
573extern struct dsl_pool *spa_get_dsl(spa_t *spa);
574extern blkptr_t *spa_get_rootblkptr(spa_t *spa);
575extern void spa_set_rootblkptr(spa_t *spa, const blkptr_t *bp);
576extern void spa_altroot(spa_t *, char *, size_t);
577extern int spa_sync_pass(spa_t *spa);
578extern char *spa_name(spa_t *spa);
579extern uint64_t spa_guid(spa_t *spa);
580extern uint64_t spa_last_synced_txg(spa_t *spa);
581extern uint64_t spa_first_txg(spa_t *spa);
582extern uint64_t spa_syncing_txg(spa_t *spa);
583extern uint64_t spa_version(spa_t *spa);
584extern pool_state_t spa_state(spa_t *spa);
585extern spa_load_state_t spa_load_state(spa_t *spa);
586extern uint64_t spa_freeze_txg(spa_t *spa);
587extern uint64_t spa_get_asize(spa_t *spa, uint64_t lsize);
588extern uint64_t spa_get_dspace(spa_t *spa);
589extern void spa_update_dspace(spa_t *spa);
590extern uint64_t spa_version(spa_t *spa);
591extern boolean_t spa_deflate(spa_t *spa);
592extern metaslab_class_t *spa_normal_class(spa_t *spa);
593extern metaslab_class_t *spa_log_class(spa_t *spa);
594extern int spa_max_replication(spa_t *spa);
595extern int spa_prev_software_version(spa_t *spa);
596extern int spa_busy(void);
597extern uint8_t spa_get_failmode(spa_t *spa);
598extern boolean_t spa_suspended(spa_t *spa);
599extern uint64_t spa_bootfs(spa_t *spa);
600extern uint64_t spa_delegation(spa_t *spa);
601extern objset_t *spa_meta_objset(spa_t *spa);
602
603/* Miscellaneous support routines */
604extern int spa_rename(const char *oldname, const char *newname);
605extern spa_t *spa_by_guid(uint64_t pool_guid, uint64_t device_guid);
606extern boolean_t spa_guid_exists(uint64_t pool_guid, uint64_t device_guid);
607extern char *spa_strdup(const char *);
608extern void spa_strfree(char *);
609extern uint64_t spa_get_random(uint64_t range);
610extern uint64_t spa_generate_guid(spa_t *spa);
611extern void sprintf_blkptr(char *buf, const blkptr_t *bp);
612extern void spa_freeze(spa_t *spa);
613extern void spa_upgrade(spa_t *spa, uint64_t version);
614extern void spa_evict_all(void);
615extern vdev_t *spa_lookup_by_guid(spa_t *spa, uint64_t guid,
616    boolean_t l2cache);
617extern boolean_t spa_has_spare(spa_t *, uint64_t guid);
618extern uint64_t dva_get_dsize_sync(spa_t *spa, const dva_t *dva);
619extern uint64_t bp_get_dsize_sync(spa_t *spa, const blkptr_t *bp);
620extern uint64_t bp_get_dsize(spa_t *spa, const blkptr_t *bp);
621extern boolean_t spa_has_slogs(spa_t *spa);
622extern boolean_t spa_is_root(spa_t *spa);
623extern boolean_t spa_writeable(spa_t *spa);
624
625extern int spa_mode(spa_t *spa);
626extern uint64_t strtonum(const char *str, char **nptr);
627
628/* history logging */
629typedef enum history_log_type {
630	LOG_CMD_POOL_CREATE,
631	LOG_CMD_NORMAL,
632	LOG_INTERNAL
633} history_log_type_t;
634
635typedef struct history_arg {
636	char *ha_history_str;
637	history_log_type_t ha_log_type;
638	history_internal_events_t ha_event;
639	char *ha_zone;
640	uid_t ha_uid;
641} history_arg_t;
642
643extern char *spa_his_ievent_table[];
644
645extern void spa_history_create_obj(spa_t *spa, dmu_tx_t *tx);
646extern int spa_history_get(spa_t *spa, uint64_t *offset, uint64_t *len_read,
647    char *his_buf);
648extern int spa_history_log(spa_t *spa, const char *his_buf,
649    history_log_type_t what);
650extern void spa_history_log_internal(history_internal_events_t event,
651    spa_t *spa, dmu_tx_t *tx, const char *fmt, ...);
652extern void spa_history_log_version(spa_t *spa, history_internal_events_t evt);
653
654/* error handling */
655struct zbookmark;
656extern void spa_log_error(spa_t *spa, zio_t *zio);
657extern void zfs_ereport_post(const char *class, spa_t *spa, vdev_t *vd,
658    zio_t *zio, uint64_t stateoroffset, uint64_t length);
659extern void zfs_post_remove(spa_t *spa, vdev_t *vd);
660extern void zfs_post_state_change(spa_t *spa, vdev_t *vd);
661extern void zfs_post_autoreplace(spa_t *spa, vdev_t *vd);
662extern uint64_t spa_get_errlog_size(spa_t *spa);
663extern int spa_get_errlog(spa_t *spa, void *uaddr, size_t *count);
664extern void spa_errlog_rotate(spa_t *spa);
665extern void spa_errlog_drain(spa_t *spa);
666extern void spa_errlog_sync(spa_t *spa, uint64_t txg);
667extern void spa_get_errlists(spa_t *spa, avl_tree_t *last, avl_tree_t *scrub);
668
669/* vdev cache */
670extern void vdev_cache_stat_init(void);
671extern void vdev_cache_stat_fini(void);
672
673/* Initialization and termination */
674extern void spa_init(int flags);
675extern void spa_fini(void);
676extern void spa_boot_init();
677
678/* properties */
679extern int spa_prop_set(spa_t *spa, nvlist_t *nvp);
680extern int spa_prop_get(spa_t *spa, nvlist_t **nvp);
681extern void spa_prop_clear_bootfs(spa_t *spa, uint64_t obj, dmu_tx_t *tx);
682extern void spa_configfile_set(spa_t *, nvlist_t *, boolean_t);
683
684/* asynchronous event notification */
685extern void spa_event_notify(spa_t *spa, vdev_t *vdev, const char *name);
686
687#ifdef ZFS_DEBUG
688#define	dprintf_bp(bp, fmt, ...) do {				\
689	if (zfs_flags & ZFS_DEBUG_DPRINTF) { 			\
690	char *__blkbuf = kmem_alloc(BP_SPRINTF_LEN, KM_SLEEP);	\
691	sprintf_blkptr(__blkbuf, (bp));				\
692	dprintf(fmt " %s\n", __VA_ARGS__, __blkbuf);		\
693	kmem_free(__blkbuf, BP_SPRINTF_LEN);			\
694	} \
695_NOTE(CONSTCOND) } while (0)
696#else
697#define	dprintf_bp(bp, fmt, ...)
698#endif
699
700extern int spa_mode_global;			/* mode, e.g. FREAD | FWRITE */
701
702#ifdef	__cplusplus
703}
704#endif
705
706#endif	/* _SYS_SPA_H */
707