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/*
21 * Copyright 2010 Sun Microsystems, Inc.  All rights reserved.
22 * Use is subject to license terms.
23 */
24
25/*
26 * Copyright (c) 2013, 2015 by Delphix. All rights reserved.
27 * Copyright (c) 2014 Integros [integros.com]
28 */
29
30#ifndef _SYS_SPA_H
31#define	_SYS_SPA_H
32
33/*
34 * General-purpose 32-bit and 64-bit bitfield encodings.
35 */
36#define	BF32_DECODE(x, low, len)	P2PHASE((x) >> (low), 1U << (len))
37#define	BF64_DECODE(x, low, len)	P2PHASE((x) >> (low), 1ULL << (len))
38#define	BF32_ENCODE(x, low, len)	(P2PHASE((x), 1U << (len)) << (low))
39#define	BF64_ENCODE(x, low, len)	(P2PHASE((x), 1ULL << (len)) << (low))
40
41#define	BF32_GET(x, low, len)		BF32_DECODE(x, low, len)
42#define	BF64_GET(x, low, len)		BF64_DECODE(x, low, len)
43
44#define	BF32_SET(x, low, len, val)	\
45	((x) ^= BF32_ENCODE((x >> low) ^ (val), low, len))
46#define	BF64_SET(x, low, len, val)	\
47	((x) ^= BF64_ENCODE((x >> low) ^ (val), low, len))
48
49#define	BF32_GET_SB(x, low, len, shift, bias)	\
50	((BF32_GET(x, low, len) + (bias)) << (shift))
51#define	BF64_GET_SB(x, low, len, shift, bias)	\
52	((BF64_GET(x, low, len) + (bias)) << (shift))
53
54#define	BF32_SET_SB(x, low, len, shift, bias, val)	\
55	BF32_SET(x, low, len, ((val) >> (shift)) - (bias))
56#define	BF64_SET_SB(x, low, len, shift, bias, val)	\
57	BF64_SET(x, low, len, ((val) >> (shift)) - (bias))
58
59/*
60 * Note: GRUB can't actually read blocks larger than 128KB, due to lack
61 * of memory.  Therefore its SPA_MAXBLOCKSIZE is still 128KB.
62 */
63#define	SPA_MINBLOCKSHIFT	9
64#define	SPA_MAXBLOCKSHIFT	17
65#define	SPA_MINBLOCKSIZE	(1ULL << SPA_MINBLOCKSHIFT)
66#define	SPA_MAXBLOCKSIZE	(1ULL << SPA_MAXBLOCKSHIFT)
67
68/*
69 * Size of block to hold the configuration data (a packed nvlist)
70 */
71#define	SPA_CONFIG_BLOCKSIZE	(1ULL << 14)
72
73/*
74 * The DVA size encodings for LSIZE and PSIZE support blocks up to 32MB.
75 * The ASIZE encoding should be at least 64 times larger (6 more bits)
76 * to support up to 4-way RAID-Z mirror mode with worst-case gang block
77 * overhead, three DVAs per bp, plus one more bit in case we do anything
78 * else that expands the ASIZE.
79 */
80#define	SPA_LSIZEBITS		16	/* LSIZE up to 32M (2^16 * 512)	*/
81#define	SPA_PSIZEBITS		16	/* PSIZE up to 32M (2^16 * 512)	*/
82#define	SPA_ASIZEBITS		24	/* ASIZE up to 64 times larger	*/
83
84#define	SPA_COMPRESSBITS	7
85
86/*
87 * All SPA data is represented by 128-bit data virtual addresses (DVAs).
88 * The members of the dva_t should be considered opaque outside the SPA.
89 */
90typedef struct dva {
91	uint64_t	dva_word[2];
92} dva_t;
93
94/*
95 * Each block has a 256-bit checksum -- strong enough for cryptographic hashes.
96 */
97typedef struct zio_cksum {
98	uint64_t	zc_word[4];
99} zio_cksum_t;
100
101/*
102 * Each block is described by its DVAs, time of birth, checksum, etc.
103 * The word-by-word, bit-by-bit layout of the blkptr is as follows:
104 *
105 *	64	56	48	40	32	24	16	8	0
106 *	+-------+-------+-------+-------+-------+-------+-------+-------+
107 * 0	|		vdev1		| GRID  |	  ASIZE		|
108 *	+-------+-------+-------+-------+-------+-------+-------+-------+
109 * 1	|G|			 offset1				|
110 *	+-------+-------+-------+-------+-------+-------+-------+-------+
111 * 2	|		vdev2		| GRID  |	  ASIZE		|
112 *	+-------+-------+-------+-------+-------+-------+-------+-------+
113 * 3	|G|			 offset2				|
114 *	+-------+-------+-------+-------+-------+-------+-------+-------+
115 * 4	|		vdev3		| GRID  |	  ASIZE		|
116 *	+-------+-------+-------+-------+-------+-------+-------+-------+
117 * 5	|G|			 offset3				|
118 *	+-------+-------+-------+-------+-------+-------+-------+-------+
119 * 6	|BDX|lvl| type	| cksum |E| comp|    PSIZE	|     LSIZE	|
120 *	+-------+-------+-------+-------+-------+-------+-------+-------+
121 * 7	|			padding					|
122 *	+-------+-------+-------+-------+-------+-------+-------+-------+
123 * 8	|			padding					|
124 *	+-------+-------+-------+-------+-------+-------+-------+-------+
125 * 9	|			physical birth txg			|
126 *	+-------+-------+-------+-------+-------+-------+-------+-------+
127 * a	|			logical birth txg			|
128 *	+-------+-------+-------+-------+-------+-------+-------+-------+
129 * b	|			fill count				|
130 *	+-------+-------+-------+-------+-------+-------+-------+-------+
131 * c	|			checksum[0]				|
132 *	+-------+-------+-------+-------+-------+-------+-------+-------+
133 * d	|			checksum[1]				|
134 *	+-------+-------+-------+-------+-------+-------+-------+-------+
135 * e	|			checksum[2]				|
136 *	+-------+-------+-------+-------+-------+-------+-------+-------+
137 * f	|			checksum[3]				|
138 *	+-------+-------+-------+-------+-------+-------+-------+-------+
139 *
140 * Legend:
141 *
142 * vdev		virtual device ID
143 * offset	offset into virtual device
144 * LSIZE	logical size
145 * PSIZE	physical size (after compression)
146 * ASIZE	allocated size (including RAID-Z parity and gang block headers)
147 * GRID		RAID-Z layout information (reserved for future use)
148 * cksum	checksum function
149 * comp		compression function
150 * G		gang block indicator
151 * B		byteorder (endianness)
152 * D		dedup
153 * X		encryption (on version 30, which is not supported)
154 * E		blkptr_t contains embedded data
155 * lvl		level of indirection
156 * type		DMU object type
157 * phys birth	txg of block allocation; zero if same as logical birth txg
158 * log. birth	transaction group in which the block was logically born
159 * fill count	number of non-zero blocks under this bp
160 * checksum[4]	256-bit checksum of the data this bp describes
161 */
162#define	SPA_BLKPTRSHIFT	7		/* blkptr_t is 128 bytes	*/
163#define	SPA_DVAS_PER_BP	3		/* Number of DVAs in a bp	*/
164
165typedef struct blkptr {
166	dva_t		blk_dva[SPA_DVAS_PER_BP]; /* Data Virtual Addresses */
167	uint64_t	blk_prop;	/* size, compression, type, etc	    */
168	uint64_t	blk_pad[2];	/* Extra space for the future	    */
169	uint64_t	blk_phys_birth;	/* txg when block was allocated	    */
170	uint64_t	blk_birth;	/* transaction group at birth	    */
171	uint64_t	blk_fill;	/* fill count			    */
172	zio_cksum_t	blk_cksum;	/* 256-bit checksum		    */
173} blkptr_t;
174
175/*
176 * Macros to get and set fields in a bp or DVA.
177 */
178#define	DVA_GET_ASIZE(dva)	\
179	BF64_GET_SB((dva)->dva_word[0], 0, SPA_ASIZEBITS, SPA_MINBLOCKSHIFT, 0)
180#define	DVA_SET_ASIZE(dva, x)	\
181	BF64_SET_SB((dva)->dva_word[0], 0, SPA_ASIZEBITS, \
182	SPA_MINBLOCKSHIFT, 0, x)
183
184#define	DVA_GET_GRID(dva)	BF64_GET((dva)->dva_word[0], 24, 8)
185#define	DVA_SET_GRID(dva, x)	BF64_SET((dva)->dva_word[0], 24, 8, x)
186
187#define	DVA_GET_VDEV(dva)	BF64_GET((dva)->dva_word[0], 32, 32)
188#define	DVA_SET_VDEV(dva, x)	BF64_SET((dva)->dva_word[0], 32, 32, x)
189
190#define	DVA_GET_OFFSET(dva)	\
191	BF64_GET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0)
192#define	DVA_SET_OFFSET(dva, x)	\
193	BF64_SET_SB((dva)->dva_word[1], 0, 63, SPA_MINBLOCKSHIFT, 0, x)
194
195#define	DVA_GET_GANG(dva)	BF64_GET((dva)->dva_word[1], 63, 1)
196#define	DVA_SET_GANG(dva, x)	BF64_SET((dva)->dva_word[1], 63, 1, x)
197
198#define	BP_GET_LSIZE(bp)	\
199	BF64_GET_SB((bp)->blk_prop, 0, SPA_LSIZEBITS, SPA_MINBLOCKSHIFT, 1)
200#define	BP_SET_LSIZE(bp, x)	\
201	BF64_SET_SB((bp)->blk_prop, 0, SPA_LSIZEBITS, SPA_MINBLOCKSHIFT, 1, x)
202
203#define	BP_GET_PSIZE(bp)	\
204	BF64_GET_SB((bp)->blk_prop, 16, SPA_PSIZEBITS, SPA_MINBLOCKSHIFT, 1)
205#define	BP_SET_PSIZE(bp, x)	\
206	BF64_SET_SB((bp)->blk_prop, 16, SPA_PSIZEBITS, SPA_MINBLOCKSHIFT, 1, x)
207
208#define	BP_GET_COMPRESS(bp)		\
209	BF64_GET((bp)->blk_prop, 32, SPA_COMPRESSBITS)
210#define	BP_SET_COMPRESS(bp, x)		\
211	BF64_SET((bp)->blk_prop, 32, SPA_COMPRESSBITS, x)
212
213#define	BP_GET_CHECKSUM(bp)		BF64_GET((bp)->blk_prop, 40, 8)
214#define	BP_SET_CHECKSUM(bp, x)		BF64_SET((bp)->blk_prop, 40, 8, x)
215
216#define	BP_GET_TYPE(bp)			BF64_GET((bp)->blk_prop, 48, 8)
217#define	BP_SET_TYPE(bp, x)		BF64_SET((bp)->blk_prop, 48, 8, x)
218
219#define	BP_GET_LEVEL(bp)		BF64_GET((bp)->blk_prop, 56, 5)
220#define	BP_SET_LEVEL(bp, x)		BF64_SET((bp)->blk_prop, 56, 5, x)
221
222#define	BP_IS_EMBEDDED(bp)		BF64_GET((bp)->blk_prop, 39, 1)
223
224#define	BP_GET_DEDUP(bp)		BF64_GET((bp)->blk_prop, 62, 1)
225#define	BP_SET_DEDUP(bp, x)		BF64_SET((bp)->blk_prop, 62, 1, x)
226
227#define	BP_GET_BYTEORDER(bp)		BF64_GET((bp)->blk_prop, 63, 1)
228#define	BP_SET_BYTEORDER(bp, x)		BF64_SET((bp)->blk_prop, 63, 1, x)
229
230#define	BP_PHYSICAL_BIRTH(bp)		\
231	((bp)->blk_phys_birth ? (bp)->blk_phys_birth : (bp)->blk_birth)
232
233#define	BP_SET_BIRTH(bp, logical, physical)	\
234{						\
235	(bp)->blk_birth = (logical);		\
236	(bp)->blk_phys_birth = ((logical) == (physical) ? 0 : (physical)); \
237}
238
239#define	BP_GET_ASIZE(bp)	\
240	(DVA_GET_ASIZE(&(bp)->blk_dva[0]) + DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
241		DVA_GET_ASIZE(&(bp)->blk_dva[2]))
242
243#define	BP_GET_UCSIZE(bp) \
244	((BP_GET_LEVEL(bp) > 0 || dmu_ot[BP_GET_TYPE(bp)].ot_metadata) ? \
245	BP_GET_PSIZE(bp) : BP_GET_LSIZE(bp));
246
247#define	BP_GET_NDVAS(bp)	\
248	(!!DVA_GET_ASIZE(&(bp)->blk_dva[0]) + \
249	!!DVA_GET_ASIZE(&(bp)->blk_dva[1]) + \
250	!!DVA_GET_ASIZE(&(bp)->blk_dva[2]))
251
252#define	DVA_EQUAL(dva1, dva2)	\
253	((dva1)->dva_word[1] == (dva2)->dva_word[1] && \
254	(dva1)->dva_word[0] == (dva2)->dva_word[0])
255
256#define	BP_EQUAL(bp1, bp2)	\
257	(BP_PHYSICAL_BIRTH(bp1) == BP_PHYSICAL_BIRTH(bp2) &&	\
258	DVA_EQUAL(&(bp1)->blk_dva[0], &(bp2)->blk_dva[0]) &&	\
259	DVA_EQUAL(&(bp1)->blk_dva[1], &(bp2)->blk_dva[1]) &&	\
260	DVA_EQUAL(&(bp1)->blk_dva[2], &(bp2)->blk_dva[2]))
261
262#define	ZIO_CHECKSUM_EQUAL(zc1, zc2) \
263	(0 == (((zc1).zc_word[0] - (zc2).zc_word[0]) | \
264	((zc1).zc_word[1] - (zc2).zc_word[1]) | \
265	((zc1).zc_word[2] - (zc2).zc_word[2]) | \
266	((zc1).zc_word[3] - (zc2).zc_word[3])))
267
268#define	DVA_IS_VALID(dva)	(DVA_GET_ASIZE(dva) != 0)
269
270#define	ZIO_SET_CHECKSUM(zcp, w0, w1, w2, w3)	\
271{						\
272	(zcp)->zc_word[0] = w0;			\
273	(zcp)->zc_word[1] = w1;			\
274	(zcp)->zc_word[2] = w2;			\
275	(zcp)->zc_word[3] = w3;			\
276}
277
278#define	BP_IDENTITY(bp)		(&(bp)->blk_dva[0])
279#define	BP_IS_GANG(bp)		DVA_GET_GANG(BP_IDENTITY(bp))
280#define	DVA_IS_EMPTY(dva)	((dva)->dva_word[0] == 0ULL &&	\
281				(dva)->dva_word[1] == 0ULL)
282#define	BP_IS_HOLE(bp)		DVA_IS_EMPTY(BP_IDENTITY(bp))
283
284/* BP_IS_RAIDZ(bp) assumes no block compression */
285#define	BP_IS_RAIDZ(bp)		(DVA_GET_ASIZE(&(bp)->blk_dva[0]) > \
286				BP_GET_PSIZE(bp))
287
288#define	BP_ZERO(bp)				\
289{						\
290	(bp)->blk_dva[0].dva_word[0] = 0;	\
291	(bp)->blk_dva[0].dva_word[1] = 0;	\
292	(bp)->blk_dva[1].dva_word[0] = 0;	\
293	(bp)->blk_dva[1].dva_word[1] = 0;	\
294	(bp)->blk_dva[2].dva_word[0] = 0;	\
295	(bp)->blk_dva[2].dva_word[1] = 0;	\
296	(bp)->blk_prop = 0;			\
297	(bp)->blk_pad[0] = 0;			\
298	(bp)->blk_pad[1] = 0;			\
299	(bp)->blk_phys_birth = 0;		\
300	(bp)->blk_birth = 0;			\
301	(bp)->blk_fill = 0;			\
302	ZIO_SET_CHECKSUM(&(bp)->blk_cksum, 0, 0, 0, 0);	\
303}
304
305#define	BPE_GET_ETYPE(bp)	BP_GET_CHECKSUM(bp)
306#define	BPE_GET_LSIZE(bp)	\
307	BF64_GET_SB((bp)->blk_prop, 0, 25, 0, 1)
308#define	BPE_GET_PSIZE(bp)	\
309	BF64_GET_SB((bp)->blk_prop, 25, 7, 0, 1)
310
311typedef enum bp_embedded_type {
312	BP_EMBEDDED_TYPE_DATA,
313	NUM_BP_EMBEDDED_TYPES
314} bp_embedded_type_t;
315
316#define	BPE_NUM_WORDS 14
317#define	BPE_PAYLOAD_SIZE (BPE_NUM_WORDS * sizeof (uint64_t))
318#define	BPE_IS_PAYLOADWORD(bp, wp) \
319	((wp) != &(bp)->blk_prop && (wp) != &(bp)->blk_birth)
320
321#ifdef _BIG_ENDIAN
322#define	ZFS_HOST_BYTEORDER	(0ULL)
323#else
324#define	ZFS_HOST_BYTEORDER	(1ULL)
325#endif
326
327#define	BP_SHOULD_BYTESWAP(bp)	(BP_GET_BYTEORDER(bp) != ZFS_HOST_BYTEORDER)
328
329#define	BP_SPRINTF_LEN	320
330
331#endif	/* _SYS_SPA_H */
332