spa.h revision 1b8adde7ba7d5e04395c141c5400dc2cffd7d809
1/*
2 *  GRUB  --  GRand Unified Bootloader
3 *  Copyright (C) 1999,2000,2001,2002,2003,2004  Free Software Foundation, Inc.
4 *
5 *  This program is free software; you can redistribute it and/or modify
6 *  it under the terms of the GNU General Public License as published by
7 *  the Free Software Foundation; either version 2 of the License, or
8 *  (at your option) any later version.
9 *
10 *  This program is distributed in the hope that it will be useful,
11 *  but WITHOUT ANY WARRANTY; without even the implied warranty of
12 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
13 *  GNU General Public License for more details.
14 *
15 *  You should have received a copy of the GNU General Public License
16 *  along with this program; if not, write to the Free Software
17 *  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
18 */
19/*
20 * Copyright 2007 Sun Microsystems, Inc.  All rights reserved.
21 * Use is subject to license terms.
22 */
23
24#ifndef _SYS_SPA_H
25#define	_SYS_SPA_H
26
27#pragma ident	"%Z%%M%	%I%	%E% SMI"
28
29/*
30 * General-purpose 32-bit and 64-bit bitfield encodings.
31 */
32#define	BF32_DECODE(x, low, len)	P2PHASE((x) >> (low), 1U << (len))
33#define	BF64_DECODE(x, low, len)	P2PHASE((x) >> (low), 1ULL << (len))
34#define	BF32_ENCODE(x, low, len)	(P2PHASE((x), 1U << (len)) << (low))
35#define	BF64_ENCODE(x, low, len)	(P2PHASE((x), 1ULL << (len)) << (low))
36
37#define	BF32_GET(x, low, len)		BF32_DECODE(x, low, len)
38#define	BF64_GET(x, low, len)		BF64_DECODE(x, low, len)
39
40#define	BF32_SET(x, low, len, val)	\
41	((x) ^= BF32_ENCODE((x >> low) ^ (val), low, len))
42#define	BF64_SET(x, low, len, val)	\
43	((x) ^= BF64_ENCODE((x >> low) ^ (val), low, len))
44
45#define	BF32_GET_SB(x, low, len, shift, bias)	\
46	((BF32_GET(x, low, len) + (bias)) << (shift))
47#define	BF64_GET_SB(x, low, len, shift, bias)	\
48	((BF64_GET(x, low, len) + (bias)) << (shift))
49
50#define	BF32_SET_SB(x, low, len, shift, bias, val)	\
51	BF32_SET(x, low, len, ((val) >> (shift)) - (bias))
52#define	BF64_SET_SB(x, low, len, shift, bias, val)	\
53	BF64_SET(x, low, len, ((val) >> (shift)) - (bias))
54
55/*
56 * We currently support nine block sizes, from 512 bytes to 128K.
57 * We could go higher, but the benefits are near-zero and the cost
58 * of COWing a giant block to modify one byte would become excessive.
59 */
60#define	SPA_MINBLOCKSHIFT	9
61#define	SPA_MAXBLOCKSHIFT	17
62#define	SPA_MINBLOCKSIZE	(1ULL << SPA_MINBLOCKSHIFT)
63#define	SPA_MAXBLOCKSIZE	(1ULL << SPA_MAXBLOCKSHIFT)
64
65#define	SPA_BLOCKSIZES		(SPA_MAXBLOCKSHIFT - SPA_MINBLOCKSHIFT + 1)
66
67/*
68 * The DVA size encodings for LSIZE and PSIZE support blocks up to 32MB.
69 * The ASIZE encoding should be at least 64 times larger (6 more bits)
70 * to support up to 4-way RAID-Z mirror mode with worst-case gang block
71 * overhead, three DVAs per bp, plus one more bit in case we do anything
72 * else that expands the ASIZE.
73 */
74#define	SPA_LSIZEBITS		16	/* LSIZE up to 32M (2^16 * 512)	*/
75#define	SPA_PSIZEBITS		16	/* PSIZE up to 32M (2^16 * 512)	*/
76#define	SPA_ASIZEBITS		24	/* ASIZE up to 64 times larger	*/
77
78/*
79 * All SPA data is represented by 128-bit data virtual addresses (DVAs).
80 * The members of the dva_t should be considered opaque outside the SPA.
81 */
82typedef struct dva {
83	uint64_t	dva_word[2];
84} dva_t;
85
86/*
87 * Each block has a 256-bit checksum -- strong enough for cryptographic hashes.
88 */
89typedef struct zio_cksum {
90	uint64_t	zc_word[4];
91} zio_cksum_t;
92
93/*
94 * Each block is described by its DVAs, time of birth, checksum, etc.
95 * The word-by-word, bit-by-bit layout of the blkptr is as follows:
96 *
97 *	64	56	48	40	32	24	16	8	0
98 *	+-------+-------+-------+-------+-------+-------+-------+-------+
99 * 0	|		vdev1		| GRID  |	  ASIZE		|
100 *	+-------+-------+-------+-------+-------+-------+-------+-------+
101 * 1	|G|			 offset1				|
102 *	+-------+-------+-------+-------+-------+-------+-------+-------+
103 * 2	|		vdev2		| GRID  |	  ASIZE		|
104 *	+-------+-------+-------+-------+-------+-------+-------+-------+
105 * 3	|G|			 offset2				|
106 *	+-------+-------+-------+-------+-------+-------+-------+-------+
107 * 4	|		vdev3		| GRID  |	  ASIZE		|
108 *	+-------+-------+-------+-------+-------+-------+-------+-------+
109 * 5	|G|			 offset3				|
110 *	+-------+-------+-------+-------+-------+-------+-------+-------+
111 * 6	|E| lvl | type	| cksum | comp	|     PSIZE	|     LSIZE	|
112 *	+-------+-------+-------+-------+-------+-------+-------+-------+
113 * 7	|			padding					|
114 *	+-------+-------+-------+-------+-------+-------+-------+-------+
115 * 8	|			padding					|
116 *	+-------+-------+-------+-------+-------+-------+-------+-------+
117 * 9	|			padding					|
118 *	+-------+-------+-------+-------+-------+-------+-------+-------+
119 * a	|			birth txg				|
120 *	+-------+-------+-------+-------+-------+-------+-------+-------+
121 * b	|			fill count				|
122 *	+-------+-------+-------+-------+-------+-------+-------+-------+
123 * c	|			checksum[0]				|
124 *	+-------+-------+-------+-------+-------+-------+-------+-------+
125 * d	|			checksum[1]				|
126 *	+-------+-------+-------+-------+-------+-------+-------+-------+
127 * e	|			checksum[2]				|
128 *	+-------+-------+-------+-------+-------+-------+-------+-------+
129 * f	|			checksum[3]				|
130 *	+-------+-------+-------+-------+-------+-------+-------+-------+
131 *
132 * Legend:
133 *
134 * vdev		virtual device ID
135 * offset	offset into virtual device
136 * LSIZE	logical size
137 * PSIZE	physical size (after compression)
138 * ASIZE	allocated size (including RAID-Z parity and gang block headers)
139 * GRID		RAID-Z layout information (reserved for future use)
140 * cksum	checksum function
141 * comp		compression function
142 * G		gang block indicator
143 * E		endianness
144 * type		DMU object type
145 * lvl		level of indirection
146 * birth txg	transaction group in which the block was born
147 * fill count	number of non-zero blocks under this bp
148 * checksum[4]	256-bit checksum of the data this bp describes
149 */
150typedef struct blkptr {
151	dva_t		blk_dva[3];	/* 128-bit Data Virtual Address	*/
152	uint64_t	blk_prop;	/* size, compression, type, etc	*/
153	uint64_t	blk_pad[3];	/* Extra space for the future	*/
154	uint64_t	blk_birth;	/* transaction group at birth	*/
155	uint64_t	blk_fill;	/* fill count			*/
156	zio_cksum_t	blk_cksum;	/* 256-bit checksum		*/
157} blkptr_t;
158
159#define	SPA_BLKPTRSHIFT	7		/* blkptr_t is 128 bytes	*/
160#define	SPA_DVAS_PER_BP	3		/* Number of DVAs in a bp	*/
161
162/*
163 * Macros to get and set fields in a bp or DVA.
164 */
165#define	DVA_GET_ASIZE(dva)	\
166	BF64_GET_SB((dva)->dva_word[0], 0, 24, SPA_MINBLOCKSHIFT, 0)
167#define	DVA_SET_ASIZE(dva, x)	\
168	BF64_SET_SB((dva)->dva_word[0], 0, 24, SPA_MINBLOCKSHIFT, 0, x)
169
170#define	DVA_GET_GRID(dva)	BF64_GET((dva)->dva_word[0], 24, 8)
171#define	DVA_SET_GRID(dva, x)	BF64_SET((dva)->dva_word[0], 24, 8, x)
172
173#define	DVA_GET_VDEV(dva)	BF64_GET((dva)->dva_word[0], 32, 32)
174#define	DVA_SET_VDEV(dva, x)	BF64_SET((dva)->dva_word[0], 32, 32, x)
175
176#define	DVA_GET_OFFSET(dva)	\
177	BF64_GET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0)
178#define	DVA_SET_OFFSET(dva, x)	\
179	BF64_SET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0, x)
180
181#define	DVA_GET_GANG(dva)	BF64_GET((dva)->dva_word[1], 63, 1)
182#define	DVA_SET_GANG(dva, x)	BF64_SET((dva)->dva_word[1], 63, 1, x)
183
184#define	BP_GET_LSIZE(bp)	\
185	(BP_IS_HOLE(bp) ? 0 : \
186	BF64_GET_SB((bp)->blk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1))
187#define	BP_SET_LSIZE(bp, x)	\
188	BF64_SET_SB((bp)->blk_prop, 0, 16, SPA_MINBLOCKSHIFT, 1, x)
189
190#define	BP_GET_PSIZE(bp)	\
191	BF64_GET_SB((bp)->blk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1)
192#define	BP_SET_PSIZE(bp, x)	\
193	BF64_SET_SB((bp)->blk_prop, 16, 16, SPA_MINBLOCKSHIFT, 1, x)
194
195#define	BP_GET_COMPRESS(bp)	BF64_GET((bp)->blk_prop, 32, 8)
196#define	BP_SET_COMPRESS(bp, x)	BF64_SET((bp)->blk_prop, 32, 8, x)
197
198#define	BP_GET_CHECKSUM(bp)	BF64_GET((bp)->blk_prop, 40, 8)
199#define	BP_SET_CHECKSUM(bp, x)	BF64_SET((bp)->blk_prop, 40, 8, x)
200
201#define	BP_GET_TYPE(bp)		BF64_GET((bp)->blk_prop, 48, 8)
202#define	BP_SET_TYPE(bp, x)	BF64_SET((bp)->blk_prop, 48, 8, x)
203
204#define	BP_GET_LEVEL(bp)	BF64_GET((bp)->blk_prop, 56, 5)
205#define	BP_SET_LEVEL(bp, x)	BF64_SET((bp)->blk_prop, 56, 5, x)
206
207#define	BP_GET_BYTEORDER(bp)	(0 - BF64_GET((bp)->blk_prop, 63, 1))
208#define	BP_SET_BYTEORDER(bp, x)	BF64_SET((bp)->blk_prop, 63, 1, x)
209
210#define	BP_GET_ASIZE(bp)	\
211	(DVA_GET_ASIZE(&(bp)->blk_dva[0]) + DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
212		DVA_GET_ASIZE(&(bp)->blk_dva[2]))
213
214#define	BP_GET_UCSIZE(bp) \
215	((BP_GET_LEVEL(bp) > 0 || dmu_ot[BP_GET_TYPE(bp)].ot_metadata) ? \
216	BP_GET_PSIZE(bp) : BP_GET_LSIZE(bp));
217
218#define	BP_GET_NDVAS(bp)	\
219	(!!DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \
220	!!DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
221	!!DVA_GET_ASIZE(&(bp)->blk_dva[2]))
222
223#define	BP_COUNT_GANG(bp)	\
224	(DVA_GET_GANG(&(bp)->blk_dva[0]) + \
225	DVA_GET_GANG(&(bp)->blk_dva[1]) + \
226	DVA_GET_GANG(&(bp)->blk_dva[2]))
227
228#define	DVA_EQUAL(dva1, dva2)	\
229	((dva1)->dva_word[1] == (dva2)->dva_word[1] && \
230	(dva1)->dva_word[0] == (dva2)->dva_word[0])
231
232#define	ZIO_CHECKSUM_EQUAL(zc1, zc2) \
233	(0 == (((zc1).zc_word[0] - (zc2).zc_word[0]) | \
234	((zc1).zc_word[1] - (zc2).zc_word[1]) | \
235	((zc1).zc_word[2] - (zc2).zc_word[2]) | \
236	((zc1).zc_word[3] - (zc2).zc_word[3])))
237
238
239#define	DVA_IS_VALID(dva)	(DVA_GET_ASIZE(dva) != 0)
240
241#define	ZIO_SET_CHECKSUM(zcp, w0, w1, w2, w3)	\
242{						\
243	(zcp)->zc_word[0] = w0;			\
244	(zcp)->zc_word[1] = w1;			\
245	(zcp)->zc_word[2] = w2;			\
246	(zcp)->zc_word[3] = w3;			\
247}
248
249#define	BP_IDENTITY(bp)		(&(bp)->blk_dva[0])
250#define	BP_IS_GANG(bp)		DVA_GET_GANG(BP_IDENTITY(bp))
251#define	BP_IS_HOLE(bp)		((bp)->blk_birth == 0)
252#define	BP_IS_OLDER(bp, txg)	(!BP_IS_HOLE(bp) && (bp)->blk_birth < (txg))
253
254#define	BP_ZERO(bp)				\
255{						\
256	(bp)->blk_dva[0].dva_word[0] = 0;	\
257	(bp)->blk_dva[0].dva_word[1] = 0;	\
258	(bp)->blk_dva[1].dva_word[0] = 0;	\
259	(bp)->blk_dva[1].dva_word[1] = 0;	\
260	(bp)->blk_dva[2].dva_word[0] = 0;	\
261	(bp)->blk_dva[2].dva_word[1] = 0;	\
262	(bp)->blk_prop = 0;			\
263	(bp)->blk_pad[0] = 0;			\
264	(bp)->blk_pad[1] = 0;			\
265	(bp)->blk_pad[2] = 0;			\
266	(bp)->blk_birth = 0;			\
267	(bp)->blk_fill = 0;			\
268	ZIO_SET_CHECKSUM(&(bp)->blk_cksum, 0, 0, 0, 0);	\
269}
270
271/*
272 * Note: the byteorder is either 0 or -1, both of which are palindromes.
273 * This simplifies the endianness handling a bit.
274 */
275#ifdef _BIG_ENDIAN
276#define	ZFS_HOST_BYTEORDER	(0ULL)
277#else
278#define	ZFS_HOST_BYTEORDER	(-1ULL)
279#endif
280
281#define	BP_SHOULD_BYTESWAP(bp)	(BP_GET_BYTEORDER(bp) != ZFS_HOST_BYTEORDER)
282
283#define	BP_SPRINTF_LEN	320
284
285#endif	/* _SYS_SPA_H */
286