1 /*
2 md_p.h : physical layout of Linux RAID devices
3 Copyright (C) 1996-98 Ingo Molnar, Gadi Oxman
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, or (at your option)
8 any later version.
9
10 You should have received a copy of the GNU General Public License
11 (for example /usr/src/linux/COPYING); if not, write to the Free
12 Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
13 */
14
15 #ifndef _MD_P_H
16 #define _MD_P_H
17
18 #include <linux/types.h>
19
20 /*
21 * RAID superblock.
22 *
23 * The RAID superblock maintains some statistics on each RAID configuration.
24 * Each real device in the RAID set contains it near the end of the device.
25 * Some of the ideas are copied from the ext2fs implementation.
26 *
27 * We currently use 4096 bytes as follows:
28 *
29 * word offset function
30 *
31 * 0 - 31 Constant generic RAID device information.
32 * 32 - 63 Generic state information.
33 * 64 - 127 Personality specific information.
34 * 128 - 511 12 32-words descriptors of the disks in the raid set.
35 * 512 - 911 Reserved.
36 * 912 - 1023 Disk specific descriptor.
37 */
38
39 /*
40 * If x is the real device size in bytes, we return an apparent size of:
41 *
42 * y = (x & ~(MD_RESERVED_BYTES - 1)) - MD_RESERVED_BYTES
43 *
44 * and place the 4kB superblock at offset y.
45 */
46 #define MD_RESERVED_BYTES (64 * 1024)
47 #define MD_RESERVED_SECTORS (MD_RESERVED_BYTES / 512)
48
49 #define MD_NEW_SIZE_SECTORS(x) ((x & ~(MD_RESERVED_SECTORS - 1)) - MD_RESERVED_SECTORS)
50
51 #define MD_SB_BYTES 4096
52 #define MD_SB_WORDS (MD_SB_BYTES / 4)
53 #define MD_SB_SECTORS (MD_SB_BYTES / 512)
54
55 /*
56 * The following are counted in 32-bit words
57 */
58 #define MD_SB_GENERIC_OFFSET 0
59 #define MD_SB_PERSONALITY_OFFSET 64
60 #define MD_SB_DISKS_OFFSET 128
61 #define MD_SB_DESCRIPTOR_OFFSET 992
62
63 #define MD_SB_GENERIC_CONSTANT_WORDS 32
64 #define MD_SB_GENERIC_STATE_WORDS 32
65 #define MD_SB_GENERIC_WORDS (MD_SB_GENERIC_CONSTANT_WORDS + MD_SB_GENERIC_STATE_WORDS)
66 #define MD_SB_PERSONALITY_WORDS 64
67 #define MD_SB_DESCRIPTOR_WORDS 32
68 #define MD_SB_DISKS 27
69 #define MD_SB_DISKS_WORDS (MD_SB_DISKS*MD_SB_DESCRIPTOR_WORDS)
70 #define MD_SB_RESERVED_WORDS (1024 - MD_SB_GENERIC_WORDS - MD_SB_PERSONALITY_WORDS - MD_SB_DISKS_WORDS - MD_SB_DESCRIPTOR_WORDS)
71 #define MD_SB_EQUAL_WORDS (MD_SB_GENERIC_WORDS + MD_SB_PERSONALITY_WORDS + MD_SB_DISKS_WORDS)
72
73 /*
74 * Device "operational" state bits
75 */
76 #define MD_DISK_FAULTY 0 /* disk is faulty / operational */
77 #define MD_DISK_ACTIVE 1 /* disk is running or spare disk */
78 #define MD_DISK_SYNC 2 /* disk is in sync with the raid set */
79 #define MD_DISK_REMOVED 3 /* disk is in sync with the raid set */
80
81 #define MD_DISK_WRITEMOSTLY 9 /* disk is "write-mostly" is RAID1 config.
82 * read requests will only be sent here in
83 * dire need
84 */
85
86 typedef struct mdp_device_descriptor_s {
87 __u32 number; /* 0 Device number in the entire set */
88 __u32 major; /* 1 Device major number */
89 __u32 minor; /* 2 Device minor number */
90 __u32 raid_disk; /* 3 The role of the device in the raid set */
91 __u32 state; /* 4 Operational state */
92 __u32 reserved[MD_SB_DESCRIPTOR_WORDS - 5];
93 } mdp_disk_t;
94
95 #define MD_SB_MAGIC 0xa92b4efc
96
97 /*
98 * Superblock state bits
99 */
100 #define MD_SB_CLEAN 0
101 #define MD_SB_ERRORS 1
102
103 #define MD_SB_BITMAP_PRESENT 8 /* bitmap may be present nearby */
104
105 /*
106 * Notes:
107 * - if an array is being reshaped (restriped) in order to change the
108 * the number of active devices in the array, 'raid_disks' will be
109 * the larger of the old and new numbers. 'delta_disks' will
110 * be the "new - old". So if +ve, raid_disks is the new value, and
111 * "raid_disks-delta_disks" is the old. If -ve, raid_disks is the
112 * old value and "raid_disks+delta_disks" is the new (smaller) value.
113 */
114
115
116 typedef struct mdp_superblock_s {
117 /*
118 * Constant generic information
119 */
120 __u32 md_magic; /* 0 MD identifier */
121 __u32 major_version; /* 1 major version to which the set conforms */
122 __u32 minor_version; /* 2 minor version ... */
123 __u32 patch_version; /* 3 patchlevel version ... */
124 __u32 gvalid_words; /* 4 Number of used words in this section */
125 __u32 set_uuid0; /* 5 Raid set identifier */
126 __u32 ctime; /* 6 Creation time */
127 __u32 level; /* 7 Raid personality */
128 __u32 size; /* 8 Apparent size of each individual disk */
129 __u32 nr_disks; /* 9 total disks in the raid set */
130 __u32 raid_disks; /* 10 disks in a fully functional raid set */
131 __u32 md_minor; /* 11 preferred MD minor device number */
132 __u32 not_persistent; /* 12 does it have a persistent superblock */
133 __u32 set_uuid1; /* 13 Raid set identifier #2 */
134 __u32 set_uuid2; /* 14 Raid set identifier #3 */
135 __u32 set_uuid3; /* 15 Raid set identifier #4 */
136 __u32 gstate_creserved[MD_SB_GENERIC_CONSTANT_WORDS - 16];
137
138 /*
139 * Generic state information
140 */
141 __u32 utime; /* 0 Superblock update time */
142 __u32 state; /* 1 State bits (clean, ...) */
143 __u32 active_disks; /* 2 Number of currently active disks */
144 __u32 working_disks; /* 3 Number of working disks */
145 __u32 failed_disks; /* 4 Number of failed disks */
146 __u32 spare_disks; /* 5 Number of spare disks */
147 __u32 sb_csum; /* 6 checksum of the whole superblock */
148 #ifdef __BIG_ENDIAN
149 __u32 events_hi; /* 7 high-order of superblock update count */
150 __u32 events_lo; /* 8 low-order of superblock update count */
151 __u32 cp_events_hi; /* 9 high-order of checkpoint update count */
152 __u32 cp_events_lo; /* 10 low-order of checkpoint update count */
153 #else
154 __u32 events_lo; /* 7 low-order of superblock update count */
155 __u32 events_hi; /* 8 high-order of superblock update count */
156 __u32 cp_events_lo; /* 9 low-order of checkpoint update count */
157 __u32 cp_events_hi; /* 10 high-order of checkpoint update count */
158 #endif
159 __u32 recovery_cp; /* 11 recovery checkpoint sector count */
160 /* There are only valid for minor_version > 90 */
161 __u64 reshape_position; /* 12,13 next address in array-space for reshape */
162 __u32 new_level; /* 14 new level we are reshaping to */
163 __u32 delta_disks; /* 15 change in number of raid_disks */
164 __u32 new_layout; /* 16 new layout */
165 __u32 new_chunk; /* 17 new chunk size (bytes) */
166 __u32 gstate_sreserved[MD_SB_GENERIC_STATE_WORDS - 18];
167
168 /*
169 * Personality information
170 */
171 __u32 layout; /* 0 the array's physical layout */
172 __u32 chunk_size; /* 1 chunk size in bytes */
173 __u32 root_pv; /* 2 LV root PV */
174 __u32 root_block; /* 3 LV root block */
175 __u32 pstate_reserved[MD_SB_PERSONALITY_WORDS - 4];
176
177 /*
178 * Disks information
179 */
180 mdp_disk_t disks[MD_SB_DISKS];
181
182 /*
183 * Reserved
184 */
185 __u32 reserved[MD_SB_RESERVED_WORDS];
186
187 /*
188 * Active descriptor
189 */
190 mdp_disk_t this_disk;
191
192 } mdp_super_t;
193
md_event(mdp_super_t * sb)194 static inline __u64 md_event(mdp_super_t *sb) {
195 __u64 ev = sb->events_hi;
196 return (ev<<32)| sb->events_lo;
197 }
198
199 #define MD_SUPERBLOCK_1_TIME_SEC_MASK ((1ULL<<40) - 1)
200
201 /*
202 * The version-1 superblock :
203 * All numeric fields are little-endian.
204 *
205 * total size: 256 bytes plus 2 per device.
206 * 1K allows 384 devices.
207 */
208 struct mdp_superblock_1 {
209 /* constant array information - 128 bytes */
210 __le32 magic; /* MD_SB_MAGIC: 0xa92b4efc - little endian */
211 __le32 major_version; /* 1 */
212 __le32 feature_map; /* bit 0 set if 'bitmap_offset' is meaningful */
213 __le32 pad0; /* always set to 0 when writing */
214
215 __u8 set_uuid[16]; /* user-space generated. */
216 char set_name[32]; /* set and interpreted by user-space */
217
218 __le64 ctime; /* lo 40 bits are seconds, top 24 are microseconds or 0*/
219 __le32 level; /* -4 (multipath), -1 (linear), 0,1,4,5 */
220 __le32 layout; /* only for raid5 and raid10 currently */
221 __le64 size; /* used size of component devices, in 512byte sectors */
222
223 __le32 chunksize; /* in 512byte sectors */
224 __le32 raid_disks;
225 __le32 bitmap_offset; /* sectors after start of superblock that bitmap starts
226 * NOTE: signed, so bitmap can be before superblock
227 * only meaningful of feature_map[0] is set.
228 */
229
230 /* These are only valid with feature bit '4' */
231 __le32 new_level; /* new level we are reshaping to */
232 __le64 reshape_position; /* next address in array-space for reshape */
233 __le32 delta_disks; /* change in number of raid_disks */
234 __le32 new_layout; /* new layout */
235 __le32 new_chunk; /* new chunk size (512byte sectors) */
236 __u8 pad1[128-124]; /* set to 0 when written */
237
238 /* constant this-device information - 64 bytes */
239 __le64 data_offset; /* sector start of data, often 0 */
240 __le64 data_size; /* sectors in this device that can be used for data */
241 __le64 super_offset; /* sector start of this superblock */
242 __le64 recovery_offset;/* sectors before this offset (from data_offset) have been recovered */
243 __le32 dev_number; /* permanent identifier of this device - not role in raid */
244 __le32 cnt_corrected_read; /* number of read errors that were corrected by re-writing */
245 __u8 device_uuid[16]; /* user-space setable, ignored by kernel */
246 __u8 devflags; /* per-device flags. Only one defined...*/
247 #define WriteMostly1 1 /* mask for writemostly flag in above */
248 /* Bad block log. If there are any bad blocks the feature flag is set.
249 * If offset and size are non-zero, that space is reserved and available
250 */
251 __u8 bblog_shift; /* shift from sectors to block size */
252 __le16 bblog_size; /* number of sectors reserved for list */
253 __le32 bblog_offset; /* sector offset from superblock to bblog,
254 * signed - not unsigned */
255
256 /* array state information - 64 bytes */
257 __le64 utime; /* 40 bits second, 24 bits microseconds */
258 __le64 events; /* incremented when superblock updated */
259 __le64 resync_offset; /* data before this offset (from data_offset) known to be in sync */
260 __le32 sb_csum; /* checksum up to devs[max_dev] */
261 __le32 max_dev; /* size of devs[] array to consider */
262 __u8 pad3[64-32]; /* set to 0 when writing */
263
264 /* device state information. Indexed by dev_number.
265 * 2 bytes per device
266 * Note there are no per-device state flags. State information is rolled
267 * into the 'roles' value. If a device is spare or faulty, then it doesn't
268 * have a meaningful role.
269 */
270 __le16 dev_roles[0]; /* role in array, or 0xffff for a spare, or 0xfffe for faulty */
271 };
272
273 /* feature_map bits */
274 #define MD_FEATURE_BITMAP_OFFSET 1
275 #define MD_FEATURE_RECOVERY_OFFSET 2 /* recovery_offset is present and
276 * must be honoured
277 */
278 #define MD_FEATURE_RESHAPE_ACTIVE 4
279 #define MD_FEATURE_BAD_BLOCKS 8 /* badblock list is not empty */
280 #define MD_FEATURE_REPLACEMENT 16 /* This device is replacing an
281 * active device with same 'role'.
282 * 'recovery_offset' is also set.
283 */
284 #define MD_FEATURE_ALL (MD_FEATURE_BITMAP_OFFSET \
285 |MD_FEATURE_RECOVERY_OFFSET \
286 |MD_FEATURE_RESHAPE_ACTIVE \
287 |MD_FEATURE_BAD_BLOCKS \
288 |MD_FEATURE_REPLACEMENT)
289
290 #endif
291