spa.h revision 485bbbf5450c6645352388d798251c1a89ef4c9c
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 2009 Sun Microsystems, Inc.  All rights reserved.
23 * Use is subject to license terms.
24 */
25
26#ifndef _SYS_SPA_H
27#define	_SYS_SPA_H
28
29#include <sys/avl.h>
30#include <sys/zfs_context.h>
31#include <sys/nvpair.h>
32#include <sys/sysmacros.h>
33#include <sys/types.h>
34#include <sys/fs/zfs.h>
35
36#ifdef	__cplusplus
37extern "C" {
38#endif
39
40/*
41 * Forward references that lots of things need.
42 */
43typedef struct spa spa_t;
44typedef struct vdev vdev_t;
45typedef struct metaslab metaslab_t;
46typedef struct metaslab_group metaslab_group_t;
47typedef struct metaslab_class metaslab_class_t;
48typedef struct zio zio_t;
49typedef struct zilog zilog_t;
50typedef struct spa_aux_vdev spa_aux_vdev_t;
51typedef struct ddt ddt_t;
52typedef struct ddt_entry ddt_entry_t;
53struct dsl_pool;
54
55/*
56 * General-purpose 32-bit and 64-bit bitfield encodings.
57 */
58#define	BF32_DECODE(x, low, len)	P2PHASE((x) >> (low), 1U << (len))
59#define	BF64_DECODE(x, low, len)	P2PHASE((x) >> (low), 1ULL << (len))
60#define	BF32_ENCODE(x, low, len)	(P2PHASE((x), 1U << (len)) << (low))
61#define	BF64_ENCODE(x, low, len)	(P2PHASE((x), 1ULL << (len)) << (low))
62
63#define	BF32_GET(x, low, len)		BF32_DECODE(x, low, len)
64#define	BF64_GET(x, low, len)		BF64_DECODE(x, low, len)
65
66#define	BF32_SET(x, low, len, val)	\
67	((x) ^= BF32_ENCODE((x >> low) ^ (val), low, len))
68#define	BF64_SET(x, low, len, val)	\
69	((x) ^= BF64_ENCODE((x >> low) ^ (val), low, len))
70
71#define	BF32_GET_SB(x, low, len, shift, bias)	\
72	((BF32_GET(x, low, len) + (bias)) << (shift))
73#define	BF64_GET_SB(x, low, len, shift, bias)	\
74	((BF64_GET(x, low, len) + (bias)) << (shift))
75
76#define	BF32_SET_SB(x, low, len, shift, bias, val)	\
77	BF32_SET(x, low, len, ((val) >> (shift)) - (bias))
78#define	BF64_SET_SB(x, low, len, shift, bias, val)	\
79	BF64_SET(x, low, len, ((val) >> (shift)) - (bias))
80
81/*
82 * We currently support nine block sizes, from 512 bytes to 128K.
83 * We could go higher, but the benefits are near-zero and the cost
84 * of COWing a giant block to modify one byte would become excessive.
85 */
86#define	SPA_MINBLOCKSHIFT	9
87#define	SPA_MAXBLOCKSHIFT	17
88#define	SPA_MINBLOCKSIZE	(1ULL << SPA_MINBLOCKSHIFT)
89#define	SPA_MAXBLOCKSIZE	(1ULL << SPA_MAXBLOCKSHIFT)
90
91#define	SPA_BLOCKSIZES		(SPA_MAXBLOCKSHIFT - SPA_MINBLOCKSHIFT + 1)
92
93/*
94 * Size of block to hold the configuration data (a packed nvlist)
95 */
96#define	SPA_CONFIG_BLOCKSIZE	(1 << 14)
97
98/*
99 * The DVA size encodings for LSIZE and PSIZE support blocks up to 32MB.
100 * The ASIZE encoding should be at least 64 times larger (6 more bits)
101 * to support up to 4-way RAID-Z mirror mode with worst-case gang block
102 * overhead, three DVAs per bp, plus one more bit in case we do anything
103 * else that expands the ASIZE.
104 */
105#define	SPA_LSIZEBITS		16	/* LSIZE up to 32M (2^16 * 512)	*/
106#define	SPA_PSIZEBITS		16	/* PSIZE up to 32M (2^16 * 512)	*/
107#define	SPA_ASIZEBITS		24	/* ASIZE up to 64 times larger	*/
108
109/*
110 * All SPA data is represented by 128-bit data virtual addresses (DVAs).
111 * The members of the dva_t should be considered opaque outside the SPA.
112 */
113typedef struct dva {
114	uint64_t	dva_word[2];
115} dva_t;
116
117/*
118 * Each block has a 256-bit checksum -- strong enough for cryptographic hashes.
119 */
120typedef struct zio_cksum {
121	uint64_t	zc_word[4];
122} zio_cksum_t;
123
124/*
125 * Each block is described by its DVAs, time of birth, checksum, etc.
126 * The word-by-word, bit-by-bit layout of the blkptr is as follows:
127 *
128 *	64	56	48	40	32	24	16	8	0
129 *	+-------+-------+-------+-------+-------+-------+-------+-------+
130 * 0	|		vdev1		| GRID  |	  ASIZE		|
131 *	+-------+-------+-------+-------+-------+-------+-------+-------+
132 * 1	|G|			 offset1				|
133 *	+-------+-------+-------+-------+-------+-------+-------+-------+
134 * 2	|		vdev2		| GRID  |	  ASIZE		|
135 *	+-------+-------+-------+-------+-------+-------+-------+-------+
136 * 3	|G|			 offset2				|
137 *	+-------+-------+-------+-------+-------+-------+-------+-------+
138 * 4	|		vdev3		| GRID  |	  ASIZE		|
139 *	+-------+-------+-------+-------+-------+-------+-------+-------+
140 * 5	|G|			 offset3				|
141 *	+-------+-------+-------+-------+-------+-------+-------+-------+
142 * 6	|BDX|lvl| type	| cksum | comp	|     PSIZE	|     LSIZE	|
143 *	+-------+-------+-------+-------+-------+-------+-------+-------+
144 * 7	|			padding					|
145 *	+-------+-------+-------+-------+-------+-------+-------+-------+
146 * 8	|			padding					|
147 *	+-------+-------+-------+-------+-------+-------+-------+-------+
148 * 9	|			physical birth txg			|
149 *	+-------+-------+-------+-------+-------+-------+-------+-------+
150 * a	|			logical birth txg			|
151 *	+-------+-------+-------+-------+-------+-------+-------+-------+
152 * b	|			fill count				|
153 *	+-------+-------+-------+-------+-------+-------+-------+-------+
154 * c	|			checksum[0]				|
155 *	+-------+-------+-------+-------+-------+-------+-------+-------+
156 * d	|			checksum[1]				|
157 *	+-------+-------+-------+-------+-------+-------+-------+-------+
158 * e	|			checksum[2]				|
159 *	+-------+-------+-------+-------+-------+-------+-------+-------+
160 * f	|			checksum[3]				|
161 *	+-------+-------+-------+-------+-------+-------+-------+-------+
162 *
163 * Legend:
164 *
165 * vdev		virtual device ID
166 * offset	offset into virtual device
167 * LSIZE	logical size
168 * PSIZE	physical size (after compression)
169 * ASIZE	allocated size (including RAID-Z parity and gang block headers)
170 * GRID		RAID-Z layout information (reserved for future use)
171 * cksum	checksum function
172 * comp		compression function
173 * G		gang block indicator
174 * B		byteorder (endianness)
175 * D		dedup
176 * X		unused
177 * lvl		level of indirection
178 * type		DMU object type
179 * phys birth	txg of block allocation; zero if same as logical birth txg
180 * log. birth	transaction group in which the block was logically born
181 * fill count	number of non-zero blocks under this bp
182 * checksum[4]	256-bit checksum of the data this bp describes
183 */
184#define	SPA_BLKPTRSHIFT	7		/* blkptr_t is 128 bytes	*/
185#define	SPA_DVAS_PER_BP	3		/* Number of DVAs in a bp	*/
186
187typedef struct blkptr {
188	dva_t		blk_dva[SPA_DVAS_PER_BP]; /* Data Virtual Addresses */
189	uint64_t	blk_prop;	/* size, compression, type, etc	    */
190	uint64_t	blk_pad[2];	/* Extra space for the future	    */
191	uint64_t	blk_phys_birth;	/* txg when block was allocated	    */
192	uint64_t	blk_birth;	/* transaction group at birth	    */
193	uint64_t	blk_fill;	/* fill count			    */
194	zio_cksum_t	blk_cksum;	/* 256-bit checksum		    */
195} blkptr_t;
196
197/*
198 * Macros to get and set fields in a bp or DVA.
199 */
200#define	DVA_GET_ASIZE(dva)	\
201	BF64_GET_SB((dva)->dva_word[0], 0, 24, SPA_MINBLOCKSHIFT, 0)
202#define	DVA_SET_ASIZE(dva, x)	\
203	BF64_SET_SB((dva)->dva_word[0], 0, 24, SPA_MINBLOCKSHIFT, 0, x)
204
205#define	DVA_GET_GRID(dva)	BF64_GET((dva)->dva_word[0], 24, 8)
206#define	DVA_SET_GRID(dva, x)	BF64_SET((dva)->dva_word[0], 24, 8, x)
207
208#define	DVA_GET_VDEV(dva)	BF64_GET((dva)->dva_word[0], 32, 32)
209#define	DVA_SET_VDEV(dva, x)	BF64_SET((dva)->dva_word[0], 32, 32, x)
210
211#define	DVA_GET_OFFSET(dva)	\
212	BF64_GET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0)
213#define	DVA_SET_OFFSET(dva, x)	\
214	BF64_SET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0, x)
215
216#define	DVA_GET_GANG(dva)	BF64_GET((dva)->dva_word[1], 63, 1)
217#define	DVA_SET_GANG(dva, x)	BF64_SET((dva)->dva_word[1], 63, 1, x)
218
219#define	BP_GET_LSIZE(bp)	\
220	BF64_GET_SB((bp)->blk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1)
221#define	BP_SET_LSIZE(bp, x)	\
222	BF64_SET_SB((bp)->blk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1, x)
223
224#define	BP_GET_PSIZE(bp)	\
225	BF64_GET_SB((bp)->blk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1)
226#define	BP_SET_PSIZE(bp, x)	\
227	BF64_SET_SB((bp)->blk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1, x)
228
229#define	BP_GET_COMPRESS(bp)		BF64_GET((bp)->blk_prop, 32, 8)
230#define	BP_SET_COMPRESS(bp, x)		BF64_SET((bp)->blk_prop, 32, 8, x)
231
232#define	BP_GET_CHECKSUM(bp)		BF64_GET((bp)->blk_prop, 40, 8)
233#define	BP_SET_CHECKSUM(bp, x)		BF64_SET((bp)->blk_prop, 40, 8, x)
234
235#define	BP_GET_TYPE(bp)			BF64_GET((bp)->blk_prop, 48, 8)
236#define	BP_SET_TYPE(bp, x)		BF64_SET((bp)->blk_prop, 48, 8, x)
237
238#define	BP_GET_LEVEL(bp)		BF64_GET((bp)->blk_prop, 56, 5)
239#define	BP_SET_LEVEL(bp, x)		BF64_SET((bp)->blk_prop, 56, 5, x)
240
241#define	BP_GET_PROP_BIT_61(bp)		BF64_GET((bp)->blk_prop, 61, 1)
242#define	BP_SET_PROP_BIT_61(bp, x)	BF64_SET((bp)->blk_prop, 61, 1, x)
243
244#define	BP_GET_DEDUP(bp)		BF64_GET((bp)->blk_prop, 62, 1)
245#define	BP_SET_DEDUP(bp, x)		BF64_SET((bp)->blk_prop, 62, 1, x)
246
247#define	BP_GET_BYTEORDER(bp)		(0 - BF64_GET((bp)->blk_prop, 63, 1))
248#define	BP_SET_BYTEORDER(bp, x)		BF64_SET((bp)->blk_prop, 63, 1, x)
249
250#define	BP_PHYSICAL_BIRTH(bp)		\
251	((bp)->blk_phys_birth ? (bp)->blk_phys_birth : (bp)->blk_birth)
252
253#define	BP_SET_BIRTH(bp, logical, physical)	\
254{						\
255	(bp)->blk_birth = (logical);		\
256	(bp)->blk_phys_birth = ((logical) == (physical) ? 0 : (physical)); \
257}
258
259#define	BP_GET_ASIZE(bp)	\
260	(DVA_GET_ASIZE(&(bp)->blk_dva[0]) + DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
261		DVA_GET_ASIZE(&(bp)->blk_dva[2]))
262
263#define	BP_GET_UCSIZE(bp) \
264	((BP_GET_LEVEL(bp) > 0 || dmu_ot[BP_GET_TYPE(bp)].ot_metadata) ? \
265	BP_GET_PSIZE(bp) : BP_GET_LSIZE(bp));
266
267#define	BP_GET_NDVAS(bp)	\
268	(!!DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \
269	!!DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
270	!!DVA_GET_ASIZE(&(bp)->blk_dva[2]))
271
272#define	BP_COUNT_GANG(bp)	\
273	(DVA_GET_GANG(&(bp)->blk_dva[0]) + \
274	DVA_GET_GANG(&(bp)->blk_dva[1]) + \
275	DVA_GET_GANG(&(bp)->blk_dva[2]))
276
277#define	DVA_EQUAL(dva1, dva2)	\
278	((dva1)->dva_word[1] == (dva2)->dva_word[1] && \
279	(dva1)->dva_word[0] == (dva2)->dva_word[0])
280
281#define	BP_EQUAL(bp1, bp2)	\
282	(BP_PHYSICAL_BIRTH(bp1) == BP_PHYSICAL_BIRTH(bp2) &&	\
283	DVA_EQUAL(&(bp1)->blk_dva[0], &(bp2)->blk_dva[0]) &&	\
284	DVA_EQUAL(&(bp1)->blk_dva[1], &(bp2)->blk_dva[1]) &&	\
285	DVA_EQUAL(&(bp1)->blk_dva[2], &(bp2)->blk_dva[2]))
286
287#define	ZIO_CHECKSUM_EQUAL(zc1, zc2) \
288	(0 == (((zc1).zc_word[0] - (zc2).zc_word[0]) | \
289	((zc1).zc_word[1] - (zc2).zc_word[1]) | \
290	((zc1).zc_word[2] - (zc2).zc_word[2]) | \
291	((zc1).zc_word[3] - (zc2).zc_word[3])))
292
293#define	DVA_IS_VALID(dva)	(DVA_GET_ASIZE(dva) != 0)
294
295#define	ZIO_SET_CHECKSUM(zcp, w0, w1, w2, w3)	\
296{						\
297	(zcp)->zc_word[0] = w0;			\
298	(zcp)->zc_word[1] = w1;			\
299	(zcp)->zc_word[2] = w2;			\
300	(zcp)->zc_word[3] = w3;			\
301}
302
303#define	BP_IDENTITY(bp)		(&(bp)->blk_dva[0])
304#define	BP_IS_GANG(bp)		DVA_GET_GANG(BP_IDENTITY(bp))
305#define	BP_IS_HOLE(bp)		((bp)->blk_birth == 0)
306
307#define	BP_ZERO(bp)				\
308{						\
309	(bp)->blk_dva[0].dva_word[0] = 0;	\
310	(bp)->blk_dva[0].dva_word[1] = 0;	\
311	(bp)->blk_dva[1].dva_word[0] = 0;	\
312	(bp)->blk_dva[1].dva_word[1] = 0;	\
313	(bp)->blk_dva[2].dva_word[0] = 0;	\
314	(bp)->blk_dva[2].dva_word[1] = 0;	\
315	(bp)->blk_prop = 0;			\
316	(bp)->blk_pad[0] = 0;			\
317	(bp)->blk_pad[1] = 0;			\
318	(bp)->blk_phys_birth = 0;		\
319	(bp)->blk_birth = 0;			\
320	(bp)->blk_fill = 0;			\
321	ZIO_SET_CHECKSUM(&(bp)->blk_cksum, 0, 0, 0, 0);	\
322}
323
324/*
325 * Note: the byteorder is either 0 or -1, both of which are palindromes.
326 * This simplifies the endianness handling a bit.
327 */
328#ifdef _BIG_ENDIAN
329#define	ZFS_HOST_BYTEORDER	(0ULL)
330#else
331#define	ZFS_HOST_BYTEORDER	(-1ULL)
332#endif
333
334#define	BP_SHOULD_BYTESWAP(bp)	(BP_GET_BYTEORDER(bp) != ZFS_HOST_BYTEORDER)
335
336#define	BP_SPRINTF_LEN	320
337
338/*
339 * This macro allows code sharing between zfs, libzpool, and mdb.
340 * 'func' is either snprintf() or mdb_snprintf().
341 * 'ws' (whitespace) can be ' ' for single-line format, '\n' for multi-line.
342 */
343#define	SPRINTF_BLKPTR(func, ws, buf, bp, type, checksum, compress)	\
344{									\
345	static const char *copyname[] =					\
346	    { "zero", "single", "double", "triple" };			\
347	int size = BP_SPRINTF_LEN;					\
348	int len = 0;							\
349	int copies = 0;							\
350									\
351	if (bp == NULL) {						\
352		len = func(buf + len, size - len, "<NULL>");		\
353	} else if (BP_IS_HOLE(bp)) {					\
354		len = func(buf + len, size - len, "<hole>");		\
355	} else {							\
356		for (int d = 0; d < BP_GET_NDVAS(bp); d++) {		\
357			const dva_t *dva = &bp->blk_dva[d];		\
358			if (DVA_IS_VALID(dva))				\
359				copies++;				\
360			len += func(buf + len, size - len,		\
361			    "DVA[%d]=<%llu:%llx:%llx>%c", d,		\
362			    (u_longlong_t)DVA_GET_VDEV(dva),		\
363			    (u_longlong_t)DVA_GET_OFFSET(dva),		\
364			    (u_longlong_t)DVA_GET_ASIZE(dva),		\
365			    ws);					\
366		}							\
367		if (BP_IS_GANG(bp) &&					\
368		    DVA_GET_ASIZE(&bp->blk_dva[2]) <=			\
369		    DVA_GET_ASIZE(&bp->blk_dva[1]) / 2)			\
370			copies--;					\
371		len += func(buf + len, size - len,			\
372		    "[L%llu %s] %s %s %s %s %s %s%c"			\
373		    "size=%llxL/%llxP birth=%lluL/%lluP fill=%llu%c"	\
374		    "cksum=%llx:%llx:%llx:%llx",			\
375		    (u_longlong_t)BP_GET_LEVEL(bp),			\
376		    type,						\
377		    checksum,						\
378		    compress,						\
379		    BP_GET_BYTEORDER(bp) == 0 ? "BE" : "LE",		\
380		    BP_IS_GANG(bp) ? "gang" : "contiguous",		\
381		    BP_GET_DEDUP(bp) ? "dedup" : "unique",		\
382		    copyname[copies],					\
383		    ws,							\
384		    (u_longlong_t)BP_GET_LSIZE(bp),			\
385		    (u_longlong_t)BP_GET_PSIZE(bp),			\
386		    (u_longlong_t)bp->blk_birth,			\
387		    (u_longlong_t)BP_PHYSICAL_BIRTH(bp),		\
388		    (u_longlong_t)bp->blk_fill,				\
389		    ws,							\
390		    (u_longlong_t)bp->blk_cksum.zc_word[0],		\
391		    (u_longlong_t)bp->blk_cksum.zc_word[1],		\
392		    (u_longlong_t)bp->blk_cksum.zc_word[2],		\
393		    (u_longlong_t)bp->blk_cksum.zc_word[3]);		\
394	}								\
395	ASSERT(len < size);						\
396}
397
398#include <sys/dmu.h>
399
400#define	BP_GET_BUFC_TYPE(bp)						\
401	(((BP_GET_LEVEL(bp) > 0) || (dmu_ot[BP_GET_TYPE(bp)].ot_metadata)) ? \
402	ARC_BUFC_METADATA : ARC_BUFC_DATA);
403
404/* state manipulation functions */
405extern int spa_open(const char *pool, spa_t **, void *tag);
406extern int spa_open_rewind(const char *pool, spa_t **, void *tag,
407    nvlist_t *policy, nvlist_t **config);
408extern int spa_get_stats(const char *pool, nvlist_t **config,
409    char *altroot, size_t buflen);
410extern int spa_create(const char *pool, nvlist_t *config, nvlist_t *props,
411    const char *history_str, nvlist_t *zplprops);
412extern int spa_import_rootpool(char *devpath, char *devid);
413extern int spa_import(const char *pool, nvlist_t *config, nvlist_t *props);
414extern int spa_import_verbatim(const char *, nvlist_t *, nvlist_t *);
415extern nvlist_t *spa_tryimport(nvlist_t *tryconfig);
416extern int spa_destroy(char *pool);
417extern int spa_export(char *pool, nvlist_t **oldconfig, boolean_t force,
418    boolean_t hardforce);
419extern int spa_reset(char *pool);
420extern void spa_async_request(spa_t *spa, int flag);
421extern void spa_async_unrequest(spa_t *spa, int flag);
422extern void spa_async_suspend(spa_t *spa);
423extern void spa_async_resume(spa_t *spa);
424extern spa_t *spa_inject_addref(char *pool);
425extern void spa_inject_delref(spa_t *spa);
426
427#define	SPA_ASYNC_CONFIG_UPDATE	0x01
428#define	SPA_ASYNC_REMOVE	0x02
429#define	SPA_ASYNC_PROBE		0x04
430#define	SPA_ASYNC_RESILVER_DONE	0x08
431#define	SPA_ASYNC_RESILVER	0x10
432#define	SPA_ASYNC_AUTOEXPAND	0x20
433
434/* device manipulation */
435extern int spa_vdev_add(spa_t *spa, nvlist_t *nvroot);
436extern int spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot,
437    int replacing);
438extern int spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid,
439    int replace_done);
440extern int spa_vdev_remove(spa_t *spa, uint64_t guid, boolean_t unspare);
441extern int spa_vdev_setpath(spa_t *spa, uint64_t guid, const char *newpath);
442extern int spa_vdev_setfru(spa_t *spa, uint64_t guid, const char *newfru);
443
444/* spare state (which is global across all pools) */
445extern void spa_spare_add(vdev_t *vd);
446extern void spa_spare_remove(vdev_t *vd);
447extern boolean_t spa_spare_exists(uint64_t guid, uint64_t *pool, int *refcnt);
448extern void spa_spare_activate(vdev_t *vd);
449
450/* L2ARC state (which is global across all pools) */
451extern void spa_l2cache_add(vdev_t *vd);
452extern void spa_l2cache_remove(vdev_t *vd);
453extern boolean_t spa_l2cache_exists(uint64_t guid, uint64_t *pool);
454extern void spa_l2cache_activate(vdev_t *vd);
455extern void spa_l2cache_drop(spa_t *spa);
456
457/* scrubbing */
458extern int spa_scrub(spa_t *spa, pool_scrub_type_t type);
459
460/* spa syncing */
461extern void spa_sync(spa_t *spa, uint64_t txg); /* only for DMU use */
462extern void spa_sync_allpools(void);
463
464#define	SYNC_PASS_DEFERRED_FREE	1	/* defer frees after this pass */
465#define	SYNC_PASS_DONT_COMPRESS	4	/* don't compress after this pass */
466#define	SYNC_PASS_REWRITE	1	/* rewrite new bps after this pass */
467
468/* spa namespace global mutex */
469extern kmutex_t spa_namespace_lock;
470
471/*
472 * SPA configuration functions in spa_config.c
473 */
474
475#define	SPA_CONFIG_UPDATE_POOL	0
476#define	SPA_CONFIG_UPDATE_VDEVS	1
477
478extern void spa_config_sync(spa_t *, boolean_t, boolean_t);
479extern void spa_config_load(void);
480extern nvlist_t *spa_all_configs(uint64_t *);
481extern void spa_config_set(spa_t *spa, nvlist_t *config);
482extern nvlist_t *spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg,
483    int getstats);
484extern void spa_config_update(spa_t *spa, int what);
485
486/*
487 * Miscellaneous SPA routines in spa_misc.c
488 */
489
490/* Namespace manipulation */
491extern spa_t *spa_lookup(const char *name);
492extern spa_t *spa_add(const char *name, nvlist_t *config, const char *altroot);
493extern void spa_remove(spa_t *spa);
494extern spa_t *spa_next(spa_t *prev);
495
496/* Refcount functions */
497extern void spa_open_ref(spa_t *spa, void *tag);
498extern void spa_close(spa_t *spa, void *tag);
499extern boolean_t spa_refcount_zero(spa_t *spa);
500
501#define	SCL_NONE	0x00
502#define	SCL_CONFIG	0x01
503#define	SCL_STATE	0x02
504#define	SCL_L2ARC	0x04		/* hack until L2ARC 2.0 */
505#define	SCL_ALLOC	0x08
506#define	SCL_ZIO		0x10
507#define	SCL_FREE	0x20
508#define	SCL_VDEV	0x40
509#define	SCL_LOCKS	7
510#define	SCL_ALL		((1 << SCL_LOCKS) - 1)
511#define	SCL_STATE_ALL	(SCL_STATE | SCL_L2ARC | SCL_ZIO)
512
513/* Pool configuration locks */
514extern int spa_config_tryenter(spa_t *spa, int locks, void *tag, krw_t rw);
515extern void spa_config_enter(spa_t *spa, int locks, void *tag, krw_t rw);
516extern void spa_config_exit(spa_t *spa, int locks, void *tag);
517extern int spa_config_held(spa_t *spa, int locks, krw_t rw);
518
519/* Pool vdev add/remove lock */
520extern uint64_t spa_vdev_enter(spa_t *spa);
521extern uint64_t spa_vdev_config_enter(spa_t *spa);
522extern void spa_vdev_config_exit(spa_t *spa, vdev_t *vd, uint64_t txg,
523    int error, char *tag);
524extern int spa_vdev_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error);
525
526/* Pool vdev state change lock */
527extern void spa_vdev_state_enter(spa_t *spa, int oplock);
528extern int spa_vdev_state_exit(spa_t *spa, vdev_t *vd, int error);
529
530/* Log state */
531typedef enum spa_log_state {
532	SPA_LOG_UNKNOWN = 0,	/* unknown log state */
533	SPA_LOG_MISSING,	/* missing log(s) */
534	SPA_LOG_CLEAR,		/* clear the log(s) */
535	SPA_LOG_GOOD,		/* log(s) are good */
536} spa_log_state_t;
537
538extern spa_log_state_t spa_get_log_state(spa_t *spa);
539extern void spa_set_log_state(spa_t *spa, spa_log_state_t state);
540
541/* Log claim callback */
542extern void spa_claim_notify(zio_t *zio);
543
544/* Accessor functions */
545extern boolean_t spa_shutting_down(spa_t *spa);
546extern struct dsl_pool *spa_get_dsl(spa_t *spa);
547extern blkptr_t *spa_get_rootblkptr(spa_t *spa);
548extern void spa_set_rootblkptr(spa_t *spa, const blkptr_t *bp);
549extern void spa_altroot(spa_t *, char *, size_t);
550extern int spa_sync_pass(spa_t *spa);
551extern char *spa_name(spa_t *spa);
552extern uint64_t spa_guid(spa_t *spa);
553extern uint64_t spa_last_synced_txg(spa_t *spa);
554extern uint64_t spa_first_txg(spa_t *spa);
555extern uint64_t spa_syncing_txg(spa_t *spa);
556extern uint64_t spa_version(spa_t *spa);
557extern pool_state_t spa_state(spa_t *spa);
558extern uint64_t spa_freeze_txg(spa_t *spa);
559extern uint64_t spa_get_asize(spa_t *spa, uint64_t lsize);
560extern uint64_t spa_get_dspace(spa_t *spa);
561extern void spa_update_dspace(spa_t *spa);
562extern uint64_t spa_version(spa_t *spa);
563extern boolean_t spa_deflate(spa_t *spa);
564extern metaslab_class_t *spa_normal_class(spa_t *spa);
565extern metaslab_class_t *spa_log_class(spa_t *spa);
566extern int spa_max_replication(spa_t *spa);
567extern int spa_busy(void);
568extern uint8_t spa_get_failmode(spa_t *spa);
569extern boolean_t spa_suspended(spa_t *spa);
570extern uint64_t spa_bootfs(spa_t *spa);
571extern uint64_t spa_delegation(spa_t *spa);
572extern objset_t *spa_meta_objset(spa_t *spa);
573extern enum zio_checksum spa_dedup_checksum(spa_t *spa);
574
575/* Miscellaneous support routines */
576extern int spa_rename(const char *oldname, const char *newname);
577extern boolean_t spa_guid_exists(uint64_t pool_guid, uint64_t device_guid);
578extern char *spa_strdup(const char *);
579extern void spa_strfree(char *);
580extern uint64_t spa_get_random(uint64_t range);
581extern void sprintf_blkptr(char *buf, const blkptr_t *bp);
582extern void spa_freeze(spa_t *spa);
583extern void spa_upgrade(spa_t *spa, uint64_t version);
584extern void spa_evict_all(void);
585extern vdev_t *spa_lookup_by_guid(spa_t *spa, uint64_t guid,
586    boolean_t l2cache);
587extern boolean_t spa_has_spare(spa_t *, uint64_t guid);
588extern uint64_t dva_get_dsize_sync(spa_t *spa, const dva_t *dva);
589extern uint64_t bp_get_dsize_sync(spa_t *spa, const blkptr_t *bp);
590extern uint64_t bp_get_dsize(spa_t *spa, const blkptr_t *bp);
591extern boolean_t spa_has_slogs(spa_t *spa);
592extern boolean_t spa_is_root(spa_t *spa);
593extern boolean_t spa_writeable(spa_t *spa);
594extern void spa_rewind_data_to_nvlist(spa_t *spa, nvlist_t *to);
595
596extern int spa_mode(spa_t *spa);
597extern uint64_t strtonum(const char *str, char **nptr);
598
599/* history logging */
600typedef enum history_log_type {
601	LOG_CMD_POOL_CREATE,
602	LOG_CMD_NORMAL,
603	LOG_INTERNAL
604} history_log_type_t;
605
606typedef struct history_arg {
607	const char *ha_history_str;
608	history_log_type_t ha_log_type;
609	history_internal_events_t ha_event;
610	char ha_zone[MAXPATHLEN];
611} history_arg_t;
612
613extern char *spa_his_ievent_table[];
614
615extern void spa_history_create_obj(spa_t *spa, dmu_tx_t *tx);
616extern int spa_history_get(spa_t *spa, uint64_t *offset, uint64_t *len_read,
617    char *his_buf);
618extern int spa_history_log(spa_t *spa, const char *his_buf,
619    history_log_type_t what);
620extern void spa_history_internal_log(history_internal_events_t event,
621    spa_t *spa, dmu_tx_t *tx, cred_t *cr, const char *fmt, ...);
622extern void spa_history_log_version(spa_t *spa, history_internal_events_t evt);
623
624/* error handling */
625struct zbookmark;
626extern void spa_log_error(spa_t *spa, zio_t *zio);
627extern void zfs_ereport_post(const char *class, spa_t *spa, vdev_t *vd,
628    zio_t *zio, uint64_t stateoroffset, uint64_t length);
629extern void zfs_post_remove(spa_t *spa, vdev_t *vd);
630extern void zfs_post_state_change(spa_t *spa, vdev_t *vd);
631extern void zfs_post_autoreplace(spa_t *spa, vdev_t *vd);
632extern uint64_t spa_get_errlog_size(spa_t *spa);
633extern int spa_get_errlog(spa_t *spa, void *uaddr, size_t *count);
634extern void spa_errlog_rotate(spa_t *spa);
635extern void spa_errlog_drain(spa_t *spa);
636extern void spa_errlog_sync(spa_t *spa, uint64_t txg);
637extern void spa_get_errlists(spa_t *spa, avl_tree_t *last, avl_tree_t *scrub);
638
639/* vdev cache */
640extern void vdev_cache_stat_init(void);
641extern void vdev_cache_stat_fini(void);
642
643/* Initialization and termination */
644extern void spa_init(int flags);
645extern void spa_fini(void);
646extern void spa_boot_init();
647
648/* properties */
649extern int spa_prop_set(spa_t *spa, nvlist_t *nvp);
650extern int spa_prop_get(spa_t *spa, nvlist_t **nvp);
651extern void spa_prop_clear_bootfs(spa_t *spa, uint64_t obj, dmu_tx_t *tx);
652extern void spa_configfile_set(spa_t *, nvlist_t *, boolean_t);
653
654/* asynchronous event notification */
655extern void spa_event_notify(spa_t *spa, vdev_t *vdev, const char *name);
656
657#ifdef ZFS_DEBUG
658#define	dprintf_bp(bp, fmt, ...) do {				\
659	if (zfs_flags & ZFS_DEBUG_DPRINTF) { 			\
660	char *__blkbuf = kmem_alloc(BP_SPRINTF_LEN, KM_SLEEP);	\
661	sprintf_blkptr(__blkbuf, (bp));				\
662	dprintf(fmt " %s\n", __VA_ARGS__, __blkbuf);		\
663	kmem_free(__blkbuf, BP_SPRINTF_LEN);			\
664	} \
665_NOTE(CONSTCOND) } while (0)
666#else
667#define	dprintf_bp(bp, fmt, ...)
668#endif
669
670extern int spa_mode_global;			/* mode, e.g. FREAD | FWRITE */
671
672#ifdef	__cplusplus
673}
674#endif
675
676#endif	/* _SYS_SPA_H */
677