spa.h revision bc9014e6a81272073b9854d9f65dd59e18d18c35
1fa9e406ahrens/*
2fa9e406ahrens * CDDL HEADER START
3fa9e406ahrens *
4fa9e406ahrens * The contents of this file are subject to the terms of the
5ea8dc4beschrock * Common Development and Distribution License (the "License").
6ea8dc4beschrock * You may not use this file except in compliance with the License.
7fa9e406ahrens *
8fa9e406ahrens * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9fa9e406ahrens * or http://www.opensolaris.org/os/licensing.
10fa9e406ahrens * See the License for the specific language governing permissions
11fa9e406ahrens * and limitations under the License.
12fa9e406ahrens *
13fa9e406ahrens * When distributing Covered Code, include this CDDL HEADER in each
14fa9e406ahrens * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15fa9e406ahrens * If applicable, add the following below this CDDL HEADER, with the
16fa9e406ahrens * fields enclosed by brackets "[]" replaced with your own identifying
17fa9e406ahrens * information: Portions Copyright [yyyy] [name of copyright owner]
18fa9e406ahrens *
19fa9e406ahrens * CDDL HEADER END
20fa9e406ahrens */
21fa9e406ahrens/*
223f9d6adLin Ling * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
237802d7bMatthew Ahrens * Copyright (c) 2011, 2014 by Delphix. All rights reserved.
24e9103aaGarrett D'Amore * Copyright 2011 Nexenta Systems, Inc.  All rights reserved.
25bc9014eJustin Gibbs * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
26fa9e406ahrens */
27fa9e406ahrens
28fa9e406ahrens#ifndef _SYS_SPA_H
29fa9e406ahrens#define	_SYS_SPA_H
30fa9e406ahrens
31fa9e406ahrens#include <sys/avl.h>
32fa9e406ahrens#include <sys/zfs_context.h>
33fa9e406ahrens#include <sys/nvpair.h>
34fa9e406ahrens#include <sys/sysmacros.h>
35fa9e406ahrens#include <sys/types.h>
36fa9e406ahrens#include <sys/fs/zfs.h>
37fa9e406ahrens
38fa9e406ahrens#ifdef	__cplusplus
39fa9e406ahrensextern "C" {
40fa9e406ahrens#endif
41fa9e406ahrens
42fa9e406ahrens/*
43fa9e406ahrens * Forward references that lots of things need.
44fa9e406ahrens */
45fa9e406ahrenstypedef struct spa spa_t;
46fa9e406ahrenstypedef struct vdev vdev_t;
47fa9e406ahrenstypedef struct metaslab metaslab_t;
48b24ab67Jeff Bonwicktypedef struct metaslab_group metaslab_group_t;
49b24ab67Jeff Bonwicktypedef struct metaslab_class metaslab_class_t;
50b24ab67Jeff Bonwicktypedef struct zio zio_t;
51fa9e406ahrenstypedef struct zilog zilog_t;
52fa94a07brendantypedef struct spa_aux_vdev spa_aux_vdev_t;
53b24ab67Jeff Bonwicktypedef struct ddt ddt_t;
54b24ab67Jeff Bonwicktypedef struct ddt_entry ddt_entry_t;
55fa9e406ahrensstruct dsl_pool;
564445fffMatthew Ahrensstruct dsl_dataset;
57fa9e406ahrens
58fa9e406ahrens/*
59fa9e406ahrens * General-purpose 32-bit and 64-bit bitfield encodings.
60fa9e406ahrens */
61fa9e406ahrens#define	BF32_DECODE(x, low, len)	P2PHASE((x) >> (low), 1U << (len))
62fa9e406ahrens#define	BF64_DECODE(x, low, len)	P2PHASE((x) >> (low), 1ULL << (len))
63fa9e406ahrens#define	BF32_ENCODE(x, low, len)	(P2PHASE((x), 1U << (len)) << (low))
64fa9e406ahrens#define	BF64_ENCODE(x, low, len)	(P2PHASE((x), 1ULL << (len)) << (low))
65fa9e406ahrens
66fa9e406ahrens#define	BF32_GET(x, low, len)		BF32_DECODE(x, low, len)
67fa9e406ahrens#define	BF64_GET(x, low, len)		BF64_DECODE(x, low, len)
68fa9e406ahrens
6943466aaMax Grossman#define	BF32_SET(x, low, len, val) do { \
7043466aaMax Grossman	ASSERT3U(val, <, 1U << (len)); \
7143466aaMax Grossman	ASSERT3U(low + len, <=, 32); \
7243466aaMax Grossman	(x) ^= BF32_ENCODE((x >> low) ^ (val), low, len); \
7343466aaMax Grossman_NOTE(CONSTCOND) } while (0)
7443466aaMax Grossman
7543466aaMax Grossman#define	BF64_SET(x, low, len, val) do { \
7643466aaMax Grossman	ASSERT3U(val, <, 1ULL << (len)); \
7743466aaMax Grossman	ASSERT3U(low + len, <=, 64); \
7843466aaMax Grossman	((x) ^= BF64_ENCODE((x >> low) ^ (val), low, len)); \
7943466aaMax Grossman_NOTE(CONSTCOND) } while (0)
80fa9e406ahrens
81fa9e406ahrens#define	BF32_GET_SB(x, low, len, shift, bias)	\
82fa9e406ahrens	((BF32_GET(x, low, len) + (bias)) << (shift))
83fa9e406ahrens#define	BF64_GET_SB(x, low, len, shift, bias)	\
84fa9e406ahrens	((BF64_GET(x, low, len) + (bias)) << (shift))
85fa9e406ahrens
8643466aaMax Grossman#define	BF32_SET_SB(x, low, len, shift, bias, val) do { \
8743466aaMax Grossman	ASSERT(IS_P2ALIGNED(val, 1U << shift)); \
8843466aaMax Grossman	ASSERT3S((val) >> (shift), >=, bias); \
8943466aaMax Grossman	BF32_SET(x, low, len, ((val) >> (shift)) - (bias)); \
9043466aaMax Grossman_NOTE(CONSTCOND) } while (0)
9143466aaMax Grossman#define	BF64_SET_SB(x, low, len, shift, bias, val) do { \
9243466aaMax Grossman	ASSERT(IS_P2ALIGNED(val, 1ULL << shift)); \
9343466aaMax Grossman	ASSERT3S((val) >> (shift), >=, bias); \
9443466aaMax Grossman	BF64_SET(x, low, len, ((val) >> (shift)) - (bias)); \
9543466aaMax Grossman_NOTE(CONSTCOND) } while (0)
96fa9e406ahrens
97fa9e406ahrens/*
98b515258Matthew Ahrens * We currently support block sizes from 512 bytes to 16MB.
99b515258Matthew Ahrens * The benefits of larger blocks, and thus larger IO, need to be weighed
100b515258Matthew Ahrens * against the cost of COWing a giant block to modify one byte, and the
101b515258Matthew Ahrens * large latency of reading or writing a large block.
102b515258Matthew Ahrens *
103b515258Matthew Ahrens * Note that although blocks up to 16MB are supported, the recordsize
104b515258Matthew Ahrens * property can not be set larger than zfs_max_recordsize (default 1MB).
105b515258Matthew Ahrens * See the comment near zfs_max_recordsize in dsl_dataset.c for details.
106b515258Matthew Ahrens *
107b515258Matthew Ahrens * Note that although the LSIZE field of the blkptr_t can store sizes up
108b515258Matthew Ahrens * to 32MB, the dnode's dn_datablkszsec can only store sizes up to
109b515258Matthew Ahrens * 32MB - 512 bytes.  Therefore, we limit SPA_MAXBLOCKSIZE to 16MB.
110fa9e406ahrens */
111fa9e406ahrens#define	SPA_MINBLOCKSHIFT	9
112b515258Matthew Ahrens#define	SPA_OLD_MAXBLOCKSHIFT	17
113b515258Matthew Ahrens#define	SPA_MAXBLOCKSHIFT	24
114fa9e406ahrens#define	SPA_MINBLOCKSIZE	(1ULL << SPA_MINBLOCKSHIFT)
115b515258Matthew Ahrens#define	SPA_OLD_MAXBLOCKSIZE	(1ULL << SPA_OLD_MAXBLOCKSHIFT)
116fa9e406ahrens#define	SPA_MAXBLOCKSIZE	(1ULL << SPA_MAXBLOCKSHIFT)
117fa9e406ahrens
118fa9e406ahrens/*
119f7991baTim Haley * Size of block to hold the configuration data (a packed nvlist)
120f7991baTim Haley */
121ad135b5Christopher Siden#define	SPA_CONFIG_BLOCKSIZE	(1ULL << 14)
122f7991baTim Haley
123f7991baTim Haley/*
124fa9e406ahrens * The DVA size encodings for LSIZE and PSIZE support blocks up to 32MB.
125fa9e406ahrens * The ASIZE encoding should be at least 64 times larger (6 more bits)
126fa9e406ahrens * to support up to 4-way RAID-Z mirror mode with worst-case gang block
127fa9e406ahrens * overhead, three DVAs per bp, plus one more bit in case we do anything
128fa9e406ahrens * else that expands the ASIZE.
129fa9e406ahrens */
130fa9e406ahrens#define	SPA_LSIZEBITS		16	/* LSIZE up to 32M (2^16 * 512)	*/
131fa9e406ahrens#define	SPA_PSIZEBITS		16	/* PSIZE up to 32M (2^16 * 512)	*/
132fa9e406ahrens#define	SPA_ASIZEBITS		24	/* ASIZE up to 64 times larger	*/
133fa9e406ahrens
134fa9e406ahrens/*
135fa9e406ahrens * All SPA data is represented by 128-bit data virtual addresses (DVAs).
136fa9e406ahrens * The members of the dva_t should be considered opaque outside the SPA.
137fa9e406ahrens */
138fa9e406ahrenstypedef struct dva {
139fa9e406ahrens	uint64_t	dva_word[2];
140fa9e406ahrens} dva_t;
141fa9e406ahrens
142fa9e406ahrens/*
143fa9e406ahrens * Each block has a 256-bit checksum -- strong enough for cryptographic hashes.
144fa9e406ahrens */
145fa9e406ahrenstypedef struct zio_cksum {
146fa9e406ahrens	uint64_t	zc_word[4];
147fa9e406ahrens} zio_cksum_t;
148fa9e406ahrens
149fa9e406ahrens/*
150fa9e406ahrens * Each block is described by its DVAs, time of birth, checksum, etc.
151fa9e406ahrens * The word-by-word, bit-by-bit layout of the blkptr is as follows:
152fa9e406ahrens *
153fa9e406ahrens *	64	56	48	40	32	24	16	8	0
154fa9e406ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
155fa9e406ahrens * 0	|		vdev1		| GRID  |	  ASIZE		|
156fa9e406ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
157fa9e406ahrens * 1	|G|			 offset1				|
158fa9e406ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
159fa9e406ahrens * 2	|		vdev2		| GRID  |	  ASIZE		|
160fa9e406ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
161fa9e406ahrens * 3	|G|			 offset2				|
162fa9e406ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
163fa9e406ahrens * 4	|		vdev3		| GRID  |	  ASIZE		|
164fa9e406ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
165fa9e406ahrens * 5	|G|			 offset3				|
166fa9e406ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
1675d7b4d4Matthew Ahrens * 6	|BDX|lvl| type	| cksum |E| comp|    PSIZE	|     LSIZE	|
168fa9e406ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
169fa9e406ahrens * 7	|			padding					|
170fa9e406ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
171fa9e406ahrens * 8	|			padding					|
172fa9e406ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
173b24ab67Jeff Bonwick * 9	|			physical birth txg			|
174fa9e406ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
175b24ab67Jeff Bonwick * a	|			logical birth txg			|
176fa9e406ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
177fa9e406ahrens * b	|			fill count				|
178fa9e406ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
179fa9e406ahrens * c	|			checksum[0]				|
180fa9e406ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
181fa9e406ahrens * d	|			checksum[1]				|
182fa9e406ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
183fa9e406ahrens * e	|			checksum[2]				|
184fa9e406ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
185fa9e406ahrens * f	|			checksum[3]				|
186fa9e406ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
187fa9e406ahrens *
188fa9e406ahrens * Legend:
189fa9e406ahrens *
190fa9e406ahrens * vdev		virtual device ID
191fa9e406ahrens * offset	offset into virtual device
192fa9e406ahrens * LSIZE	logical size
193fa9e406ahrens * PSIZE	physical size (after compression)
194fa9e406ahrens * ASIZE	allocated size (including RAID-Z parity and gang block headers)
195fa9e406ahrens * GRID		RAID-Z layout information (reserved for future use)
196fa9e406ahrens * cksum	checksum function
197fa9e406ahrens * comp		compression function
198fa9e406ahrens * G		gang block indicator
199b24ab67Jeff Bonwick * B		byteorder (endianness)
200b24ab67Jeff Bonwick * D		dedup
2015d7b4d4Matthew Ahrens * X		encryption (on version 30, which is not supported)
2025d7b4d4Matthew Ahrens * E		blkptr_t contains embedded data (see below)
203fa9e406ahrens * lvl		level of indirection
204b24ab67Jeff Bonwick * type		DMU object type
205b24ab67Jeff Bonwick * phys birth	txg of block allocation; zero if same as logical birth txg
206b24ab67Jeff Bonwick * log. birth	transaction group in which the block was logically born
207fa9e406ahrens * fill count	number of non-zero blocks under this bp
208fa9e406ahrens * checksum[4]	256-bit checksum of the data this bp describes
209fa9e406ahrens */
2105d7b4d4Matthew Ahrens
2115d7b4d4Matthew Ahrens/*
2125d7b4d4Matthew Ahrens * "Embedded" blkptr_t's don't actually point to a block, instead they
2135d7b4d4Matthew Ahrens * have a data payload embedded in the blkptr_t itself.  See the comment
2145d7b4d4Matthew Ahrens * in blkptr.c for more details.
2155d7b4d4Matthew Ahrens *
2165d7b4d4Matthew Ahrens * The blkptr_t is laid out as follows:
2175d7b4d4Matthew Ahrens *
2185d7b4d4Matthew Ahrens *	64	56	48	40	32	24	16	8	0
2195d7b4d4Matthew Ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
2205d7b4d4Matthew Ahrens * 0	|      payload                                                  |
2215d7b4d4Matthew Ahrens * 1	|      payload                                                  |
2225d7b4d4Matthew Ahrens * 2	|      payload                                                  |
2235d7b4d4Matthew Ahrens * 3	|      payload                                                  |
2245d7b4d4Matthew Ahrens * 4	|      payload                                                  |
2255d7b4d4Matthew Ahrens * 5	|      payload                                                  |
2265d7b4d4Matthew Ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
2275d7b4d4Matthew Ahrens * 6	|BDX|lvl| type	| etype |E| comp| PSIZE|              LSIZE	|
2285d7b4d4Matthew Ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
2295d7b4d4Matthew Ahrens * 7	|      payload                                                  |
2305d7b4d4Matthew Ahrens * 8	|      payload                                                  |
2315d7b4d4Matthew Ahrens * 9	|      payload                                                  |
2325d7b4d4Matthew Ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
2335d7b4d4Matthew Ahrens * a	|			logical birth txg			|
2345d7b4d4Matthew Ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
2355d7b4d4Matthew Ahrens * b	|      payload                                                  |
2365d7b4d4Matthew Ahrens * c	|      payload                                                  |
2375d7b4d4Matthew Ahrens * d	|      payload                                                  |
2385d7b4d4Matthew Ahrens * e	|      payload                                                  |
2395d7b4d4Matthew Ahrens * f	|      payload                                                  |
2405d7b4d4Matthew Ahrens *	+-------+-------+-------+-------+-------+-------+-------+-------+
2415d7b4d4Matthew Ahrens *
2425d7b4d4Matthew Ahrens * Legend:
2435d7b4d4Matthew Ahrens *
2445d7b4d4Matthew Ahrens * payload		contains the embedded data
2455d7b4d4Matthew Ahrens * B (byteorder)	byteorder (endianness)
2465d7b4d4Matthew Ahrens * D (dedup)		padding (set to zero)
2475d7b4d4Matthew Ahrens * X			encryption (set to zero; see above)
2485d7b4d4Matthew Ahrens * E (embedded)		set to one
2495d7b4d4Matthew Ahrens * lvl			indirection level
2505d7b4d4Matthew Ahrens * type			DMU object type
2515d7b4d4Matthew Ahrens * etype		how to interpret embedded data (BP_EMBEDDED_TYPE_*)
2525d7b4d4Matthew Ahrens * comp			compression function of payload
2535d7b4d4Matthew Ahrens * PSIZE		size of payload after compression, in bytes
2545d7b4d4Matthew Ahrens * LSIZE		logical size of payload, in bytes
2555d7b4d4Matthew Ahrens *			note that 25 bits is enough to store the largest
2565d7b4d4Matthew Ahrens *			"normal" BP's LSIZE (2^16 * 2^9) in bytes
2575d7b4d4Matthew Ahrens * log. birth		transaction group in which the block was logically born
2585d7b4d4Matthew Ahrens *
2595d7b4d4Matthew Ahrens * Note that LSIZE and PSIZE are stored in bytes, whereas for non-embedded
2605d7b4d4Matthew Ahrens * bp's they are stored in units of SPA_MINBLOCKSHIFT.
2615d7b4d4Matthew Ahrens * Generally, the generic BP_GET_*() macros can be used on embedded BP's.
2625d7b4d4Matthew Ahrens * The B, D, X, lvl, type, and comp fields are stored the same as with normal
2635d7b4d4Matthew Ahrens * BP's so the BP_SET_* macros can be used with them.  etype, PSIZE, LSIZE must
2645d7b4d4Matthew Ahrens * be set with the BPE_SET_* macros.  BP_SET_EMBEDDED() should be called before
2655d7b4d4Matthew Ahrens * other macros, as they assert that they are only used on BP's of the correct
2665d7b4d4Matthew Ahrens * "embedded-ness".
2675d7b4d4Matthew Ahrens */
2685d7b4d4Matthew Ahrens
2695d7b4d4Matthew Ahrens#define	BPE_GET_ETYPE(bp)	\
2705d7b4d4Matthew Ahrens	(ASSERT(BP_IS_EMBEDDED(bp)), \
2715d7b4d4Matthew Ahrens	BF64_GET((bp)->blk_prop, 40, 8))
2725d7b4d4Matthew Ahrens#define	BPE_SET_ETYPE(bp, t)	do { \
2735d7b4d4Matthew Ahrens	ASSERT(BP_IS_EMBEDDED(bp)); \
2745d7b4d4Matthew Ahrens	BF64_SET((bp)->blk_prop, 40, 8, t); \
2755d7b4d4Matthew Ahrens_NOTE(CONSTCOND) } while (0)
2765d7b4d4Matthew Ahrens
2775d7b4d4Matthew Ahrens#define	BPE_GET_LSIZE(bp)	\
2785d7b4d4Matthew Ahrens	(ASSERT(BP_IS_EMBEDDED(bp)), \
2795d7b4d4Matthew Ahrens	BF64_GET_SB((bp)->blk_prop, 0, 25, 0, 1))
2805d7b4d4Matthew Ahrens#define	BPE_SET_LSIZE(bp, x)	do { \
2815d7b4d4Matthew Ahrens	ASSERT(BP_IS_EMBEDDED(bp)); \
2825d7b4d4Matthew Ahrens	BF64_SET_SB((bp)->blk_prop, 0, 25, 0, 1, x); \
2835d7b4d4Matthew Ahrens_NOTE(CONSTCOND) } while (0)
2845d7b4d4Matthew Ahrens
2855d7b4d4Matthew Ahrens#define	BPE_GET_PSIZE(bp)	\
2865d7b4d4Matthew Ahrens	(ASSERT(BP_IS_EMBEDDED(bp)), \
2875d7b4d4Matthew Ahrens	BF64_GET_SB((bp)->blk_prop, 25, 7, 0, 1))
2885d7b4d4Matthew Ahrens#define	BPE_SET_PSIZE(bp, x)	do { \
2895d7b4d4Matthew Ahrens	ASSERT(BP_IS_EMBEDDED(bp)); \
2905d7b4d4Matthew Ahrens	BF64_SET_SB((bp)->blk_prop, 25, 7, 0, 1, x); \
2915d7b4d4Matthew Ahrens_NOTE(CONSTCOND) } while (0)
2925d7b4d4Matthew Ahrens
2935d7b4d4Matthew Ahrenstypedef enum bp_embedded_type {
2945d7b4d4Matthew Ahrens	BP_EMBEDDED_TYPE_DATA,
2955d7b4d4Matthew Ahrens	BP_EMBEDDED_TYPE_RESERVED, /* Reserved for an unintegrated feature. */
2965d7b4d4Matthew Ahrens	NUM_BP_EMBEDDED_TYPES = BP_EMBEDDED_TYPE_RESERVED
2975d7b4d4Matthew Ahrens} bp_embedded_type_t;
2985d7b4d4Matthew Ahrens
2995d7b4d4Matthew Ahrens#define	BPE_NUM_WORDS 14
3005d7b4d4Matthew Ahrens#define	BPE_PAYLOAD_SIZE (BPE_NUM_WORDS * sizeof (uint64_t))
3015d7b4d4Matthew Ahrens#define	BPE_IS_PAYLOADWORD(bp, wp) \
3025d7b4d4Matthew Ahrens	((wp) != &(bp)->blk_prop && (wp) != &(bp)->blk_birth)
3035d7b4d4Matthew Ahrens
304fa9e406ahrens#define	SPA_BLKPTRSHIFT	7		/* blkptr_t is 128 bytes	*/
305fa9e406ahrens#define	SPA_DVAS_PER_BP	3		/* Number of DVAs in a bp	*/
306fa9e406ahrens
30743466aaMax Grossman/*
30843466aaMax Grossman * A block is a hole when it has either 1) never been written to, or
30943466aaMax Grossman * 2) is zero-filled. In both cases, ZFS can return all zeroes for all reads
31043466aaMax Grossman * without physically allocating disk space. Holes are represented in the
31143466aaMax Grossman * blkptr_t structure by zeroed blk_dva. Correct checking for holes is
31243466aaMax Grossman * done through the BP_IS_HOLE macro. For holes, the logical size, level,
31343466aaMax Grossman * DMU object type, and birth times are all also stored for holes that
31443466aaMax Grossman * were written to at some point (i.e. were punched after having been filled).
31543466aaMax Grossman */
316b24ab67Jeff Bonwicktypedef struct blkptr {
317b24ab67Jeff Bonwick	dva_t		blk_dva[SPA_DVAS_PER_BP]; /* Data Virtual Addresses */
318b24ab67Jeff Bonwick	uint64_t	blk_prop;	/* size, compression, type, etc	    */
319b24ab67Jeff Bonwick	uint64_t	blk_pad[2];	/* Extra space for the future	    */
320b24ab67Jeff Bonwick	uint64_t	blk_phys_birth;	/* txg when block was allocated	    */
321b24ab67Jeff Bonwick	uint64_t	blk_birth;	/* transaction group at birth	    */
322b24ab67Jeff Bonwick	uint64_t	blk_fill;	/* fill count			    */
323b24ab67Jeff Bonwick	zio_cksum_t	blk_cksum;	/* 256-bit checksum		    */
324b24ab67Jeff Bonwick} blkptr_t;
325b24ab67Jeff Bonwick
326fa9e406ahrens/*
327fa9e406ahrens * Macros to get and set fields in a bp or DVA.
328fa9e406ahrens */
329fa9e406ahrens#define	DVA_GET_ASIZE(dva)	\
33043466aaMax Grossman	BF64_GET_SB((dva)->dva_word[0], 0, SPA_ASIZEBITS, SPA_MINBLOCKSHIFT, 0)
331fa9e406ahrens#define	DVA_SET_ASIZE(dva, x)	\
33243466aaMax Grossman	BF64_SET_SB((dva)->dva_word[0], 0, SPA_ASIZEBITS, \
33343466aaMax Grossman	SPA_MINBLOCKSHIFT, 0, x)
334fa9e406ahrens
335fa9e406ahrens#define	DVA_GET_GRID(dva)	BF64_GET((dva)->dva_word[0], 24, 8)
336fa9e406ahrens#define	DVA_SET_GRID(dva, x)	BF64_SET((dva)->dva_word[0], 24, 8, x)
337fa9e406ahrens
338fa9e406ahrens#define	DVA_GET_VDEV(dva)	BF64_GET((dva)->dva_word[0], 32, 32)
339fa9e406ahrens#define	DVA_SET_VDEV(dva, x)	BF64_SET((dva)->dva_word[0], 32, 32, x)
340fa9e406ahrens
341fa9e406ahrens#define	DVA_GET_OFFSET(dva)	\
342fa9e406ahrens	BF64_GET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0)
343fa9e406ahrens#define	DVA_SET_OFFSET(dva, x)	\
344fa9e406ahrens	BF64_SET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0, x)
345fa9e406ahrens
346fa9e406ahrens#define	DVA_GET_GANG(dva)	BF64_GET((dva)->dva_word[1], 63, 1)
347fa9e406ahrens#define	DVA_SET_GANG(dva, x)	BF64_SET((dva)->dva_word[1], 63, 1, x)
348fa9e406ahrens
349fa9e406ahrens#define	BP_GET_LSIZE(bp)	\
3505d7b4d4Matthew Ahrens	(BP_IS_EMBEDDED(bp) ?	\
3515d7b4d4Matthew Ahrens	(BPE_GET_ETYPE(bp) == BP_EMBEDDED_TYPE_DATA ? BPE_GET_LSIZE(bp) : 0): \
3525d7b4d4Matthew Ahrens	BF64_GET_SB((bp)->blk_prop, 0, SPA_LSIZEBITS, SPA_MINBLOCKSHIFT, 1))
3535d7b4d4Matthew Ahrens#define	BP_SET_LSIZE(bp, x)	do { \
3545d7b4d4Matthew Ahrens	ASSERT(!BP_IS_EMBEDDED(bp)); \
3555d7b4d4Matthew Ahrens	BF64_SET_SB((bp)->blk_prop, \
3565d7b4d4Matthew Ahrens	    0, SPA_LSIZEBITS, SPA_MINBLOCKSHIFT, 1, x); \
3575d7b4d4Matthew Ahrens_NOTE(CONSTCOND) } while (0)
358fa9e406ahrens
359fa9e406ahrens#define	BP_GET_PSIZE(bp)	\
3605d7b4d4Matthew Ahrens	(BP_IS_EMBEDDED(bp) ? 0 : \
3615d7b4d4Matthew Ahrens	BF64_GET_SB((bp)->blk_prop, 16, SPA_PSIZEBITS, SPA_MINBLOCKSHIFT, 1))
3625d7b4d4Matthew Ahrens#define	BP_SET_PSIZE(bp, x)	do { \
3635d7b4d4Matthew Ahrens	ASSERT(!BP_IS_EMBEDDED(bp)); \
3645d7b4d4Matthew Ahrens	BF64_SET_SB((bp)->blk_prop, \
3655d7b4d4Matthew Ahrens	    16, SPA_PSIZEBITS, SPA_MINBLOCKSHIFT, 1, x); \
3665d7b4d4Matthew Ahrens_NOTE(CONSTCOND) } while (0)
3675d7b4d4Matthew Ahrens
3685d7b4d4Matthew Ahrens#define	BP_GET_COMPRESS(bp)		BF64_GET((bp)->blk_prop, 32, 7)
3695d7b4d4Matthew Ahrens#define	BP_SET_COMPRESS(bp, x)		BF64_SET((bp)->blk_prop, 32, 7, x)
370fa9e406ahrens
3715d7b4d4Matthew Ahrens#define	BP_IS_EMBEDDED(bp)		BF64_GET((bp)->blk_prop, 39, 1)
3725d7b4d4Matthew Ahrens#define	BP_SET_EMBEDDED(bp, x)		BF64_SET((bp)->blk_prop, 39, 1, x)
373b24ab67Jeff Bonwick
3745d7b4d4Matthew Ahrens#define	BP_GET_CHECKSUM(bp)		\
3755d7b4d4Matthew Ahrens	(BP_IS_EMBEDDED(bp) ? ZIO_CHECKSUM_OFF : \
3765d7b4d4Matthew Ahrens	BF64_GET((bp)->blk_prop, 40, 8))
3775d7b4d4Matthew Ahrens#define	BP_SET_CHECKSUM(bp, x)		do { \
3785d7b4d4Matthew Ahrens	ASSERT(!BP_IS_EMBEDDED(bp)); \
3795d7b4d4Matthew Ahrens	BF64_SET((bp)->blk_prop, 40, 8, x); \
3805d7b4d4Matthew Ahrens_NOTE(CONSTCOND) } while (0)
381b24ab67Jeff Bonwick
382b24ab67Jeff Bonwick#define	BP_GET_TYPE(bp)			BF64_GET((bp)->blk_prop, 48, 8)
383b24ab67Jeff Bonwick#define	BP_SET_TYPE(bp, x)		BF64_SET((bp)->blk_prop, 48, 8, x)
384fa9e406ahrens
385b24ab67Jeff Bonwick#define	BP_GET_LEVEL(bp)		BF64_GET((bp)->blk_prop, 56, 5)
386b24ab67Jeff Bonwick#define	BP_SET_LEVEL(bp, x)		BF64_SET((bp)->blk_prop, 56, 5, x)
387fa9e406ahrens
388b24ab67Jeff Bonwick#define	BP_GET_DEDUP(bp)		BF64_GET((bp)->blk_prop, 62, 1)
389b24ab67Jeff Bonwick#define	BP_SET_DEDUP(bp, x)		BF64_SET((bp)->blk_prop, 62, 1, x)
390fa9e406ahrens
39143466aaMax Grossman#define	BP_GET_BYTEORDER(bp)		BF64_GET((bp)->blk_prop, 63, 1)
392b24ab67Jeff Bonwick#define	BP_SET_BYTEORDER(bp, x)		BF64_SET((bp)->blk_prop, 63, 1, x)
393b24ab67Jeff Bonwick
394b24ab67Jeff Bonwick#define	BP_PHYSICAL_BIRTH(bp)		\
3955d7b4d4Matthew Ahrens	(BP_IS_EMBEDDED(bp) ? 0 : \
3965d7b4d4Matthew Ahrens	(bp)->blk_phys_birth ? (bp)->blk_phys_birth : (bp)->blk_birth)
397b24ab67Jeff Bonwick
398b24ab67Jeff Bonwick#define	BP_SET_BIRTH(bp, logical, physical)	\
399b24ab67Jeff Bonwick{						\
4005d7b4d4Matthew Ahrens	ASSERT(!BP_IS_EMBEDDED(bp));		\
401b24ab67Jeff Bonwick	(bp)->blk_birth = (logical);		\
402b24ab67Jeff Bonwick	(bp)->blk_phys_birth = ((logical) == (physical) ? 0 : (physical)); \
403b24ab67Jeff Bonwick}
404fa9e406ahrens
4055d7b4d4Matthew Ahrens#define	BP_GET_FILL(bp) (BP_IS_EMBEDDED(bp) ? 1 : (bp)->blk_fill)
4065d7b4d4Matthew Ahrens
407fa9e406ahrens#define	BP_GET_ASIZE(bp)	\
4085d7b4d4Matthew Ahrens	(BP_IS_EMBEDDED(bp) ? 0 : \
4095d7b4d4Matthew Ahrens	DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \
4105d7b4d4Matthew Ahrens	DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
4115d7b4d4Matthew Ahrens	DVA_GET_ASIZE(&(bp)->blk_dva[2]))
41299653d4eschrock
41399653d4eschrock#define	BP_GET_UCSIZE(bp) \
414ad135b5Christopher Siden	((BP_GET_LEVEL(bp) > 0 || DMU_OT_IS_METADATA(BP_GET_TYPE(bp))) ? \
4153f9d6adLin Ling	BP_GET_PSIZE(bp) : BP_GET_LSIZE(bp))
416fa9e406ahrens
41744cd46cbillm#define	BP_GET_NDVAS(bp)	\
4185d7b4d4Matthew Ahrens	(BP_IS_EMBEDDED(bp) ? 0 : \
4195d7b4d4Matthew Ahrens	!!DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \
42044cd46cbillm	!!DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
42144cd46cbillm	!!DVA_GET_ASIZE(&(bp)->blk_dva[2]))
42244cd46cbillm
42344cd46cbillm#define	BP_COUNT_GANG(bp)	\
4245d7b4d4Matthew Ahrens	(BP_IS_EMBEDDED(bp) ? 0 : \
42544cd46cbillm	(DVA_GET_GANG(&(bp)->blk_dva[0]) + \
42644cd46cbillm	DVA_GET_GANG(&(bp)->blk_dva[1]) + \
4275d7b4d4Matthew Ahrens	DVA_GET_GANG(&(bp)->blk_dva[2])))
42844cd46cbillm
429fa9e406ahrens#define	DVA_EQUAL(dva1, dva2)	\
430fa9e406ahrens	((dva1)->dva_word[1] == (dva2)->dva_word[1] && \
431fa9e406ahrens	(dva1)->dva_word[0] == (dva2)->dva_word[0])
432fa9e406ahrens
433b24ab67Jeff Bonwick#define	BP_EQUAL(bp1, bp2)	\
434b24ab67Jeff Bonwick	(BP_PHYSICAL_BIRTH(bp1) == BP_PHYSICAL_BIRTH(bp2) &&	\
4355d7b4d4Matthew Ahrens	(bp1)->blk_birth == (bp2)->blk_birth &&			\
436b24ab67Jeff Bonwick	DVA_EQUAL(&(bp1)->blk_dva[0], &(bp2)->blk_dva[0]) &&	\
437b24ab67Jeff Bonwick	DVA_EQUAL(&(bp1)->blk_dva[1], &(bp2)->blk_dva[1]) &&	\
438b24ab67Jeff Bonwick	DVA_EQUAL(&(bp1)->blk_dva[2], &(bp2)->blk_dva[2]))
439b24ab67Jeff Bonwick
4406b4acc8ahrens#define	ZIO_CHECKSUM_EQUAL(zc1, zc2) \
4416b4acc8ahrens	(0 == (((zc1).zc_word[0] - (zc2).zc_word[0]) | \
4426b4acc8ahrens	((zc1).zc_word[1] - (zc2).zc_word[1]) | \
4436b4acc8ahrens	((zc1).zc_word[2] - (zc2).zc_word[2]) | \
4446b4acc8ahrens	((zc1).zc_word[3] - (zc2).zc_word[3])))
4456b4acc8ahrens
446fa9e406ahrens#define	DVA_IS_VALID(dva)	(DVA_GET_ASIZE(dva) != 0)
447fa9e406ahrens
448fa9e406ahrens#define	ZIO_SET_CHECKSUM(zcp, w0, w1, w2, w3)	\
449fa9e406ahrens{						\
450fa9e406ahrens	(zcp)->zc_word[0] = w0;			\
451fa9e406ahrens	(zcp)->zc_word[1] = w1;			\
452fa9e406ahrens	(zcp)->zc_word[2] = w2;			\
453fa9e406ahrens	(zcp)->zc_word[3] = w3;			\
454fa9e406ahrens}
455fa9e406ahrens
4565d7b4d4Matthew Ahrens#define	BP_IDENTITY(bp)		(ASSERT(!BP_IS_EMBEDDED(bp)), &(bp)->blk_dva[0])
4575d7b4d4Matthew Ahrens#define	BP_IS_GANG(bp)		\
4585d7b4d4Matthew Ahrens	(BP_IS_EMBEDDED(bp) ? B_FALSE : DVA_GET_GANG(BP_IDENTITY(bp)))
45943466aaMax Grossman#define	DVA_IS_EMPTY(dva)	((dva)->dva_word[0] == 0ULL &&	\
46043466aaMax Grossman				(dva)->dva_word[1] == 0ULL)
4615d7b4d4Matthew Ahrens#define	BP_IS_HOLE(bp) \
4625d7b4d4Matthew Ahrens	(!BP_IS_EMBEDDED(bp) && DVA_IS_EMPTY(BP_IDENTITY(bp)))
463fa9e406ahrens
4646e1f5caNeil Perrin/* BP_IS_RAIDZ(bp) assumes no block compression */
4656e1f5caNeil Perrin#define	BP_IS_RAIDZ(bp)		(DVA_GET_ASIZE(&(bp)->blk_dva[0]) > \
4666e1f5caNeil Perrin				BP_GET_PSIZE(bp))
4676e1f5caNeil Perrin
468e14bb32Jeff Bonwick#define	BP_ZERO(bp)				\
469fa9e406ahrens{						\
470fa9e406ahrens	(bp)->blk_dva[0].dva_word[0] = 0;	\
471fa9e406ahrens	(bp)->blk_dva[0].dva_word[1] = 0;	\
472fa9e406ahrens	(bp)->blk_dva[1].dva_word[0] = 0;	\
473fa9e406ahrens	(bp)->blk_dva[1].dva_word[1] = 0;	\
474fa9e406ahrens	(bp)->blk_dva[2].dva_word[0] = 0;	\
475fa9e406ahrens	(bp)->blk_dva[2].dva_word[1] = 0;	\
476fa9e406ahrens	(bp)->blk_prop = 0;			\
477fa9e406ahrens	(bp)->blk_pad[0] = 0;			\
478fa9e406ahrens	(bp)->blk_pad[1] = 0;			\
479b24ab67Jeff Bonwick	(bp)->blk_phys_birth = 0;		\
480e14bb32Jeff Bonwick	(bp)->blk_birth = 0;			\
481fa9e406ahrens	(bp)->blk_fill = 0;			\
482fa9e406ahrens	ZIO_SET_CHECKSUM(&(bp)->blk_cksum, 0, 0, 0, 0);	\
483fa9e406ahrens}
484fa9e406ahrens
485fa9e406ahrens#ifdef _BIG_ENDIAN
486fa9e406ahrens#define	ZFS_HOST_BYTEORDER	(0ULL)
487fa9e406ahrens#else
48843466aaMax Grossman#define	ZFS_HOST_BYTEORDER	(1ULL)
489fa9e406ahrens#endif
490fa9e406ahrens
491fa9e406ahrens#define	BP_SHOULD_BYTESWAP(bp)	(BP_GET_BYTEORDER(bp) != ZFS_HOST_BYTEORDER)
492fa9e406ahrens
49344cd46cbillm#define	BP_SPRINTF_LEN	320
494fbabab8maybee
495b24ab67Jeff Bonwick/*
496b24ab67Jeff Bonwick * This macro allows code sharing between zfs, libzpool, and mdb.
497b24ab67Jeff Bonwick * 'func' is either snprintf() or mdb_snprintf().
498b24ab67Jeff Bonwick * 'ws' (whitespace) can be ' ' for single-line format, '\n' for multi-line.
499b24ab67Jeff Bonwick */
50043466aaMax Grossman#define	SNPRINTF_BLKPTR(func, ws, buf, size, bp, type, checksum, compress) \
501b24ab67Jeff Bonwick{									\
502b24ab67Jeff Bonwick	static const char *copyname[] =					\
503b24ab67Jeff Bonwick	    { "zero", "single", "double", "triple" };			\
504b24ab67Jeff Bonwick	int len = 0;							\
505b24ab67Jeff Bonwick	int copies = 0;							\
506b24ab67Jeff Bonwick									\
507b24ab67Jeff Bonwick	if (bp == NULL) {						\
50843466aaMax Grossman		len += func(buf + len, size - len, "<NULL>");		\
509b24ab67Jeff Bonwick	} else if (BP_IS_HOLE(bp)) {					\
51043466aaMax Grossman		len += func(buf + len, size - len, "<hole>");		\
51143466aaMax Grossman		if (bp->blk_birth > 0) {				\
51243466aaMax Grossman			len += func(buf + len, size - len,		\
51343466aaMax Grossman			    " birth=%lluL",				\
51443466aaMax Grossman			    (u_longlong_t)bp->blk_birth);		\
51543466aaMax Grossman		}							\
5165d7b4d4Matthew Ahrens	} else if (BP_IS_EMBEDDED(bp)) {				\
5175d7b4d4Matthew Ahrens		len = func(buf + len, size - len,			\
5185d7b4d4Matthew Ahrens		    "EMBEDDED [L%llu %s] et=%u %s "			\
5195d7b4d4Matthew Ahrens		    "size=%llxL/%llxP birth=%lluL",			\
5205d7b4d4Matthew Ahrens		    (u_longlong_t)BP_GET_LEVEL(bp),			\
5215d7b4d4Matthew Ahrens		    type,						\
5225d7b4d4Matthew Ahrens		    (int)BPE_GET_ETYPE(bp),				\
5235d7b4d4Matthew Ahrens		    compress,						\
5245d7b4d4Matthew Ahrens		    (u_longlong_t)BPE_GET_LSIZE(bp),			\
5255d7b4d4Matthew Ahrens		    (u_longlong_t)BPE_GET_PSIZE(bp),			\
5265d7b4d4Matthew Ahrens		    (u_longlong_t)bp->blk_birth);			\
527b24ab67Jeff Bonwick	} else {							\
528b24ab67Jeff Bonwick		for (int d = 0; d < BP_GET_NDVAS(bp); d++) {		\
529b24ab67Jeff Bonwick			const dva_t *dva = &bp->blk_dva[d];		\
530b24ab67Jeff Bonwick			if (DVA_IS_VALID(dva))				\
531b24ab67Jeff Bonwick				copies++;				\
532b24ab67Jeff Bonwick			len += func(buf + len, size - len,		\
533b24ab67Jeff Bonwick			    "DVA[%d]=<%llu:%llx:%llx>%c", d,		\
534b24ab67Jeff Bonwick			    (u_longlong_t)DVA_GET_VDEV(dva),		\
535b24ab67Jeff Bonwick			    (u_longlong_t)DVA_GET_OFFSET(dva),		\
536b24ab67Jeff Bonwick			    (u_longlong_t)DVA_GET_ASIZE(dva),		\
537b24ab67Jeff Bonwick			    ws);					\
538b24ab67Jeff Bonwick		}							\
539b24ab67Jeff Bonwick		if (BP_IS_GANG(bp) &&					\
540b24ab67Jeff Bonwick		    DVA_GET_ASIZE(&bp->blk_dva[2]) <=			\
541b24ab67Jeff Bonwick		    DVA_GET_ASIZE(&bp->blk_dva[1]) / 2)			\
542b24ab67Jeff Bonwick			copies--;					\
543b24ab67Jeff Bonwick		len += func(buf + len, size - len,			\
544b24ab67Jeff Bonwick		    "[L%llu %s] %s %s %s %s %s %s%c"			\
545b24ab67Jeff Bonwick		    "size=%llxL/%llxP birth=%lluL/%lluP fill=%llu%c"	\
546b24ab67Jeff Bonwick		    "cksum=%llx:%llx:%llx:%llx",			\
547b24ab67Jeff Bonwick		    (u_longlong_t)BP_GET_LEVEL(bp),			\
548b24ab67Jeff Bonwick		    type,						\
549b24ab67Jeff Bonwick		    checksum,						\
550b24ab67Jeff Bonwick		    compress,						\
551b24ab67Jeff Bonwick		    BP_GET_BYTEORDER(bp) == 0 ? "BE" : "LE",		\
552b24ab67Jeff Bonwick		    BP_IS_GANG(bp) ? "gang" : "contiguous",		\
553b24ab67Jeff Bonwick		    BP_GET_DEDUP(bp) ? "dedup" : "unique",		\
554b24ab67Jeff Bonwick		    copyname[copies],					\
555b24ab67Jeff Bonwick		    ws,							\
556b24ab67Jeff Bonwick		    (u_longlong_t)BP_GET_LSIZE(bp),			\
557b24ab67Jeff Bonwick		    (u_longlong_t)BP_GET_PSIZE(bp),			\
558b24ab67Jeff Bonwick		    (u_longlong_t)bp->blk_birth,			\
559b24ab67Jeff Bonwick		    (u_longlong_t)BP_PHYSICAL_BIRTH(bp),		\
5605d7b4d4Matthew Ahrens		    (u_longlong_t)BP_GET_FILL(bp),			\
561b24ab67Jeff Bonwick		    ws,							\
562b24ab67Jeff Bonwick		    (u_longlong_t)bp->blk_cksum.zc_word[0],		\
563b24ab67Jeff Bonwick		    (u_longlong_t)bp->blk_cksum.zc_word[1],		\
564b24ab67Jeff Bonwick		    (u_longlong_t)bp->blk_cksum.zc_word[2],		\
565b24ab67Jeff Bonwick		    (u_longlong_t)bp->blk_cksum.zc_word[3]);		\
566b24ab67Jeff Bonwick	}								\
567b24ab67Jeff Bonwick	ASSERT(len < size);						\
568b24ab67Jeff Bonwick}
569b24ab67Jeff Bonwick
570fa9e406ahrens#include <sys/dmu.h>
571fa9e406ahrens
572ad23a2djohansen#define	BP_GET_BUFC_TYPE(bp)						\
573ad135b5Christopher Siden	(((BP_GET_LEVEL(bp) > 0) || (DMU_OT_IS_METADATA(BP_GET_TYPE(bp)))) ? \
574ad135b5Christopher Siden	ARC_BUFC_METADATA : ARC_BUFC_DATA)
575fa9e406ahrens
5761195e68Mark J Musantetypedef enum spa_import_type {
5771195e68Mark J Musante	SPA_IMPORT_EXISTING,
5781195e68Mark J Musante	SPA_IMPORT_ASSEMBLE
5791195e68Mark J Musante} spa_import_type_t;
5801195e68Mark J Musante
581fa9e406ahrens/* state manipulation functions */
582fa9e406ahrensextern int spa_open(const char *pool, spa_t **, void *tag);
583468c413Tim Haleyextern int spa_open_rewind(const char *pool, spa_t **, void *tag,
584468c413Tim Haley    nvlist_t *policy, nvlist_t **config);
585ad135b5Christopher Sidenextern int spa_get_stats(const char *pool, nvlist_t **config, char *altroot,
586ad135b5Christopher Siden    size_t buflen);
587990b485llingextern int spa_create(const char *pool, nvlist_t *config, nvlist_t *props,
5884445fffMatthew Ahrens    nvlist_t *zplprops);
589051aabetaylorextern int spa_import_rootpool(char *devpath, char *devid);
5904b964adGeorge Wilsonextern int spa_import(const char *pool, nvlist_t *config, nvlist_t *props,
5914b964adGeorge Wilson    uint64_t flags);
592fa9e406ahrensextern nvlist_t *spa_tryimport(nvlist_t *tryconfig);
593fa9e406ahrensextern int spa_destroy(char *pool);
594394ab0cGeorge Wilsonextern int spa_export(char *pool, nvlist_t **oldconfig, boolean_t force,
595394ab0cGeorge Wilson    boolean_t hardforce);
596ea8dc4beschrockextern int spa_reset(char *pool);
597ea8dc4beschrockextern void spa_async_request(spa_t *spa, int flag);
598088f389ahrensextern void spa_async_unrequest(spa_t *spa, int flag);
599ea8dc4beschrockextern void spa_async_suspend(spa_t *spa);
600ea8dc4beschrockextern void spa_async_resume(spa_t *spa);
601ea8dc4beschrockextern spa_t *spa_inject_addref(char *pool);
602ea8dc4beschrockextern void spa_inject_delref(spa_t *spa);
6033f9d6adLin Lingextern void spa_scan_stat_init(spa_t *spa);
6043f9d6adLin Lingextern int spa_scan_get_stats(spa_t *spa, pool_scan_stat_t *ps);
605ea8dc4beschrock
606e14bb32Jeff Bonwick#define	SPA_ASYNC_CONFIG_UPDATE	0x01
607e14bb32Jeff Bonwick#define	SPA_ASYNC_REMOVE	0x02
608e14bb32Jeff Bonwick#define	SPA_ASYNC_PROBE		0x04
609e14bb32Jeff Bonwick#define	SPA_ASYNC_RESILVER_DONE	0x08
610e14bb32Jeff Bonwick#define	SPA_ASYNC_RESILVER	0x10
611573ca77George Wilson#define	SPA_ASYNC_AUTOEXPAND	0x20
6123f9d6adLin Ling#define	SPA_ASYNC_REMOVE_DONE	0x40
6133f9d6adLin Ling#define	SPA_ASYNC_REMOVE_STOP	0x80
6143f9d6adLin Ling
6153f9d6adLin Ling/*
6163f9d6adLin Ling * Controls the behavior of spa_vdev_remove().
6173f9d6adLin Ling */
6183f9d6adLin Ling#define	SPA_REMOVE_UNSPARE	0x01
6193f9d6adLin Ling#define	SPA_REMOVE_DONE		0x02
620fa9e406ahrens
621fa9e406ahrens/* device manipulation */
622fa9e406ahrensextern int spa_vdev_add(spa_t *spa, nvlist_t *nvroot);
623ea8dc4beschrockextern int spa_vdev_attach(spa_t *spa, uint64_t guid, nvlist_t *nvroot,
624fa9e406ahrens    int replacing);
6258ad4d6dJeff Bonwickextern int spa_vdev_detach(spa_t *spa, uint64_t guid, uint64_t pguid,
6268ad4d6dJeff Bonwick    int replace_done);
62799653d4eschrockextern int spa_vdev_remove(spa_t *spa, uint64_t guid, boolean_t unspare);
6283f9d6adLin Lingextern boolean_t spa_vdev_remove_active(spa_t *spa);
629c67d967eschrockextern int spa_vdev_setpath(spa_t *spa, uint64_t guid, const char *newpath);
6306809eb4Eric Schrockextern int spa_vdev_setfru(spa_t *spa, uint64_t guid, const char *newfru);
6311195e68Mark J Musanteextern int spa_vdev_split_mirror(spa_t *spa, char *newname, nvlist_t *config,
6321195e68Mark J Musante    nvlist_t *props, boolean_t exp);
633fa9e406ahrens
63499653d4eschrock/* spare state (which is global across all pools) */
63539c2341eschrockextern void spa_spare_add(vdev_t *vd);
63639c2341eschrockextern void spa_spare_remove(vdev_t *vd);
63789a89ebllingextern boolean_t spa_spare_exists(uint64_t guid, uint64_t *pool, int *refcnt);
63839c2341eschrockextern void spa_spare_activate(vdev_t *vd);
63999653d4eschrock
640fa94a07brendan/* L2ARC state (which is global across all pools) */
641fa94a07brendanextern void spa_l2cache_add(vdev_t *vd);
642fa94a07brendanextern void spa_l2cache_remove(vdev_t *vd);
643fa94a07brendanextern boolean_t spa_l2cache_exists(uint64_t guid, uint64_t *pool);
644fa94a07brendanextern void spa_l2cache_activate(vdev_t *vd);
645fa94a07brendanextern void spa_l2cache_drop(spa_t *spa);
646fa94a07brendan
6473f9d6adLin Ling/* scanning */
6483f9d6adLin Lingextern int spa_scan(spa_t *spa, pool_scan_func_t func);
6493f9d6adLin Lingextern int spa_scan_stop(spa_t *spa);
650fa9e406ahrens
651fa9e406ahrens/* spa syncing */
652fa9e406ahrensextern void spa_sync(spa_t *spa, uint64_t txg); /* only for DMU use */
653fa9e406ahrensextern void spa_sync_allpools(void);
654fa9e406ahrens
6553a737e0brendan/* spa namespace global mutex */
6563a737e0brendanextern kmutex_t spa_namespace_lock;
6573a737e0brendan
658fa9e406ahrens/*
659fa9e406ahrens * SPA configuration functions in spa_config.c
660fa9e406ahrens */
6610373e76bonwick
6620373e76bonwick#define	SPA_CONFIG_UPDATE_POOL	0
6630373e76bonwick#define	SPA_CONFIG_UPDATE_VDEVS	1
6640373e76bonwick
665c5904d1eschrockextern void spa_config_sync(spa_t *, boolean_t, boolean_t);
666fa9e406ahrensextern void spa_config_load(void);
667fa9e406ahrensextern nvlist_t *spa_all_configs(uint64_t *);
668fa9e406ahrensextern void spa_config_set(spa_t *spa, nvlist_t *config);
669fa9e406ahrensextern nvlist_t *spa_config_generate(spa_t *spa, vdev_t *vd, uint64_t txg,
670fa9e406ahrens    int getstats);
6710373e76bonwickextern void spa_config_update(spa_t *spa, int what);
672fa9e406ahrens
673fa9e406ahrens/*
674fa9e406ahrens * Miscellaneous SPA routines in spa_misc.c
675fa9e406ahrens */
676fa9e406ahrens
677fa9e406ahrens/* Namespace manipulation */
678fa9e406ahrensextern spa_t *spa_lookup(const char *name);
679468c413Tim Haleyextern spa_t *spa_add(const char *name, nvlist_t *config, const char *altroot);
680fa9e406ahrensextern void spa_remove(spa_t *spa);
681fa9e406ahrensextern spa_t *spa_next(spa_t *prev);
682fa9e406ahrens
683fa9e406ahrens/* Refcount functions */
684fa9e406ahrensextern void spa_open_ref(spa_t *spa, void *tag);
685fa9e406ahrensextern void spa_close(spa_t *spa, void *tag);
686bc9014eJustin Gibbsextern void spa_async_close(spa_t *spa, void *tag);
687fa9e406ahrensextern boolean_t spa_refcount_zero(spa_t *spa);
688fa9e406ahrens
6898f18d1fGeorge Wilson#define	SCL_NONE	0x00
690e14bb32Jeff Bonwick#define	SCL_CONFIG	0x01
691e14bb32Jeff Bonwick#define	SCL_STATE	0x02
692e14bb32Jeff Bonwick#define	SCL_L2ARC	0x04		/* hack until L2ARC 2.0 */
693e14bb32Jeff Bonwick#define	SCL_ALLOC	0x08
694e14bb32Jeff Bonwick#define	SCL_ZIO		0x10
695e14bb32Jeff Bonwick#define	SCL_FREE	0x20
696e14bb32Jeff Bonwick#define	SCL_VDEV	0x40
697e14bb32Jeff Bonwick#define	SCL_LOCKS	7
698e14bb32Jeff Bonwick#define	SCL_ALL		((1 << SCL_LOCKS) - 1)
699e14bb32Jeff Bonwick#define	SCL_STATE_ALL	(SCL_STATE | SCL_L2ARC | SCL_ZIO)
700e14bb32Jeff Bonwick
701e14bb32Jeff Bonwick/* Pool configuration locks */
702e14bb32Jeff Bonwickextern int spa_config_tryenter(spa_t *spa, int locks, void *tag, krw_t rw);
703e14bb32Jeff Bonwickextern void spa_config_enter(spa_t *spa, int locks, void *tag, krw_t rw);
704e14bb32Jeff Bonwickextern void spa_config_exit(spa_t *spa, int locks, void *tag);
705e14bb32Jeff Bonwickextern int spa_config_held(spa_t *spa, int locks, krw_t rw);
706fa9e406ahrens
707fa9e406ahrens/* Pool vdev add/remove lock */
708fa9e406ahrensextern uint64_t spa_vdev_enter(spa_t *spa);
70988ecc94George Wilsonextern uint64_t spa_vdev_config_enter(spa_t *spa);
71088ecc94George Wilsonextern void spa_vdev_config_exit(spa_t *spa, vdev_t *vd, uint64_t txg,
71188ecc94George Wilson    int error, char *tag);
712fa9e406ahrensextern int spa_vdev_exit(spa_t *spa, vdev_t *vd, uint64_t txg, int error);
713fa9e406ahrens
714e14bb32Jeff Bonwick/* Pool vdev state change lock */
7158f18d1fGeorge Wilsonextern void spa_vdev_state_enter(spa_t *spa, int oplock);
716e14bb32Jeff Bonwickextern int spa_vdev_state_exit(spa_t *spa, vdev_t *vd, int error);
717e14bb32Jeff Bonwick
718b24ab67Jeff Bonwick/* Log state */
719b24ab67Jeff Bonwicktypedef enum spa_log_state {
720b24ab67Jeff Bonwick	SPA_LOG_UNKNOWN = 0,	/* unknown log state */
721b24ab67Jeff Bonwick	SPA_LOG_MISSING,	/* missing log(s) */
722b24ab67Jeff Bonwick	SPA_LOG_CLEAR,		/* clear the log(s) */
723b24ab67Jeff Bonwick	SPA_LOG_GOOD,		/* log(s) are good */
724b24ab67Jeff Bonwick} spa_log_state_t;
725b24ab67Jeff Bonwick
726b24ab67Jeff Bonwickextern spa_log_state_t spa_get_log_state(spa_t *spa);
727b24ab67Jeff Bonwickextern void spa_set_log_state(spa_t *spa, spa_log_state_t state);
7281195e68Mark J Musanteextern int spa_offline_log(spa_t *spa);
729b24ab67Jeff Bonwick
730b24ab67Jeff Bonwick/* Log claim callback */
731b24ab67Jeff Bonwickextern void spa_claim_notify(zio_t *zio);
732b24ab67Jeff Bonwick
733fa9e406ahrens/* Accessor functions */
73488b7b0fMatthew Ahrensextern boolean_t spa_shutting_down(spa_t *spa);
735fa9e406ahrensextern struct dsl_pool *spa_get_dsl(spa_t *spa);
736ad135b5Christopher Sidenextern boolean_t spa_is_initializing(spa_t *spa);
737fa9e406ahrensextern blkptr_t *spa_get_rootblkptr(spa_t *spa);
738fa9e406ahrensextern void spa_set_rootblkptr(spa_t *spa, const blkptr_t *bp);
739fa9e406ahrensextern void spa_altroot(spa_t *, char *, size_t);
740fa9e406ahrensextern int spa_sync_pass(spa_t *spa);
741fa9e406ahrensextern char *spa_name(spa_t *spa);
742fa9e406ahrensextern uint64_t spa_guid(spa_t *spa);
743e9103aaGarrett D'Amoreextern uint64_t spa_load_guid(spa_t *spa);
744fa9e406ahrensextern uint64_t spa_last_synced_txg(spa_t *spa);
745fa9e406ahrensextern uint64_t spa_first_txg(spa_t *spa);
746b24ab67Jeff Bonwickextern uint64_t spa_syncing_txg(spa_t *spa);
74799653d4eschrockextern uint64_t spa_version(spa_t *spa);
74888b7b0fMatthew Ahrensextern pool_state_t spa_state(spa_t *spa);
749b16da2eGeorge Wilsonextern spa_load_state_t spa_load_state(spa_t *spa);
750fa9e406ahrensextern uint64_t spa_freeze_txg(spa_t *spa);
751fa9e406ahrensextern uint64_t spa_get_asize(spa_t *spa, uint64_t lsize);
752485bbbfGeorge Wilsonextern uint64_t spa_get_dspace(spa_t *spa);
753c39f2c8Christopher Sidenextern uint64_t spa_get_slop_space(spa_t *spa);
754485bbbfGeorge Wilsonextern void spa_update_dspace(spa_t *spa);
75544cd46cbillmextern uint64_t spa_version(spa_t *spa);
756b24ab67Jeff Bonwickextern boolean_t spa_deflate(spa_t *spa);
757b24ab67Jeff Bonwickextern metaslab_class_t *spa_normal_class(spa_t *spa);
758b24ab67Jeff Bonwickextern metaslab_class_t *spa_log_class(spa_t *spa);
759bc9014eJustin Gibbsextern void spa_evicting_os_register(spa_t *, objset_t *os);
760bc9014eJustin Gibbsextern void spa_evicting_os_deregister(spa_t *, objset_t *os);
761bc9014eJustin Gibbsextern void spa_evicting_os_wait(spa_t *spa);
76244cd46cbillmextern int spa_max_replication(spa_t *spa);
7633f9d6adLin Lingextern int spa_prev_software_version(spa_t *spa);
764fa9e406ahrensextern int spa_busy(void);
7650a4e951gwextern uint8_t spa_get_failmode(spa_t *spa);
766e14bb32Jeff Bonwickextern boolean_t spa_suspended(spa_t *spa);
767b24ab67Jeff Bonwickextern uint64_t spa_bootfs(spa_t *spa);
768b24ab67Jeff Bonwickextern uint64_t spa_delegation(spa_t *spa);
769b24ab67Jeff Bonwickextern objset_t *spa_meta_objset(spa_t *spa);
770283b846George.Wilsonextern uint64_t spa_deadman_synctime(spa_t *spa);
771fa9e406ahrens
772fa9e406ahrens/* Miscellaneous support routines */
77343466aaMax Grossmanextern void spa_activate_mos_feature(spa_t *spa, const char *feature,
77443466aaMax Grossman    dmu_tx_t *tx);
775ad135b5Christopher Sidenextern void spa_deactivate_mos_feature(spa_t *spa, const char *feature);
776fa9e406ahrensextern int spa_rename(const char *oldname, const char *newname);
777f9af39bGeorge Wilsonextern spa_t *spa_by_guid(uint64_t pool_guid, uint64_t device_guid);
778fa9e406ahrensextern boolean_t spa_guid_exists(uint64_t pool_guid, uint64_t device_guid);
779fa9e406ahrensextern char *spa_strdup(const char *);
780fa9e406ahrensextern void spa_strfree(char *);
781fa9e406ahrensextern uint64_t spa_get_random(uint64_t range);
7821195e68Mark J Musanteextern uint64_t spa_generate_guid(spa_t *spa);
78343466aaMax Grossmanextern void snprintf_blkptr(char *buf, size_t buflen, const blkptr_t *bp);
784fa9e406ahrensextern void spa_freeze(spa_t *spa);
785e9103aaGarrett D'Amoreextern int spa_change_guid(spa_t *spa);
786990b485llingextern void spa_upgrade(spa_t *spa, uint64_t version);
787fa9e406ahrensextern void spa_evict_all(void);
788c5904d1eschrockextern vdev_t *spa_lookup_by_guid(spa_t *spa, uint64_t guid,
789c5904d1eschrock    boolean_t l2cache);
79099653d4eschrockextern boolean_t spa_has_spare(spa_t *, uint64_t guid);
791b24ab67Jeff Bonwickextern uint64_t dva_get_dsize_sync(spa_t *spa, const dva_t *dva);
792b24ab67Jeff Bonwickextern uint64_t bp_get_dsize_sync(spa_t *spa, const blkptr_t *bp);
793b24ab67Jeff Bonwickextern uint64_t bp_get_dsize(spa_t *spa, const blkptr_t *bp);
7946ce0521perrinextern boolean_t spa_has_slogs(spa_t *spa);
795bf82a41eschrockextern boolean_t spa_is_root(spa_t *spa);
7968ad4d6dJeff Bonwickextern boolean_t spa_writeable(spa_t *spa);
79773527f4Alex Reeceextern boolean_t spa_has_pending_synctask(spa_t *spa);
798b515258Matthew Ahrensextern int spa_maxblocksize(spa_t *spa);
799f63ab3dMatthew Ahrensextern void zfs_blkptr_verify(spa_t *spa, const blkptr_t *bp);
800468c413Tim Haley
8018ad4d6dJeff Bonwickextern int spa_mode(spa_t *spa);
802ca45db4Chris Kirbyextern uint64_t strtonum(const char *str, char **nptr);
803ea8dc4beschrock
804ecd6cf8marksextern char *spa_his_ievent_table[];
805ecd6cf8marks
80606eeb2aekextern void spa_history_create_obj(spa_t *spa, dmu_tx_t *tx);
80706eeb2aekextern int spa_history_get(spa_t *spa, uint64_t *offset, uint64_t *len_read,
80806eeb2aek    char *his_buf);
8094445fffMatthew Ahrensextern int spa_history_log(spa_t *spa, const char *his_buf);
8104445fffMatthew Ahrensextern int spa_history_log_nvl(spa_t *spa, nvlist_t *nvl);
8114445fffMatthew Ahrensextern void spa_history_log_version(spa_t *spa, const char *operation);
8124445fffMatthew Ahrensextern void spa_history_log_internal(spa_t *spa, const char *operation,
8134445fffMatthew Ahrens    dmu_tx_t *tx, const char *fmt, ...);
8144445fffMatthew Ahrensextern void spa_history_log_internal_ds(struct dsl_dataset *ds, const char *op,
8154445fffMatthew Ahrens    dmu_tx_t *tx, const char *fmt, ...);
8164445fffMatthew Ahrensextern void spa_history_log_internal_dd(dsl_dir_t *dd, const char *operation,
8174445fffMatthew Ahrens    dmu_tx_t *tx, const char *fmt, ...);
81806eeb2aek
819