logo

oasis-root

Compiled tree of Oasis Linux based on own branch at <https://hacktivis.me/git/oasis/> git clone https://anongit.hacktivis.me/git/oasis-root.git

md_p.h (15912B)


  1. /* SPDX-License-Identifier: GPL-2.0+ WITH Linux-syscall-note */
  2. /*
  3. md_p.h : physical layout of Linux RAID devices
  4. Copyright (C) 1996-98 Ingo Molnar, Gadi Oxman
  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. #ifndef _MD_P_H
  11. #define _MD_P_H
  12. #include <linux/types.h>
  13. #include <asm/byteorder.h>
  14. /*
  15. * RAID superblock.
  16. *
  17. * The RAID superblock maintains some statistics on each RAID configuration.
  18. * Each real device in the RAID set contains it near the end of the device.
  19. * Some of the ideas are copied from the ext2fs implementation.
  20. *
  21. * We currently use 4096 bytes as follows:
  22. *
  23. * word offset function
  24. *
  25. * 0 - 31 Constant generic RAID device information.
  26. * 32 - 63 Generic state information.
  27. * 64 - 127 Personality specific information.
  28. * 128 - 511 12 32-words descriptors of the disks in the raid set.
  29. * 512 - 911 Reserved.
  30. * 912 - 1023 Disk specific descriptor.
  31. */
  32. /*
  33. * If x is the real device size in bytes, we return an apparent size of:
  34. *
  35. * y = (x & ~(MD_RESERVED_BYTES - 1)) - MD_RESERVED_BYTES
  36. *
  37. * and place the 4kB superblock at offset y.
  38. */
  39. #define MD_RESERVED_BYTES (64 * 1024)
  40. #define MD_RESERVED_SECTORS (MD_RESERVED_BYTES / 512)
  41. #define MD_NEW_SIZE_SECTORS(x) ((x & ~(MD_RESERVED_SECTORS - 1)) - MD_RESERVED_SECTORS)
  42. #define MD_SB_BYTES 4096
  43. #define MD_SB_WORDS (MD_SB_BYTES / 4)
  44. #define MD_SB_SECTORS (MD_SB_BYTES / 512)
  45. /*
  46. * The following are counted in 32-bit words
  47. */
  48. #define MD_SB_GENERIC_OFFSET 0
  49. #define MD_SB_PERSONALITY_OFFSET 64
  50. #define MD_SB_DISKS_OFFSET 128
  51. #define MD_SB_DESCRIPTOR_OFFSET 992
  52. #define MD_SB_GENERIC_CONSTANT_WORDS 32
  53. #define MD_SB_GENERIC_STATE_WORDS 32
  54. #define MD_SB_GENERIC_WORDS (MD_SB_GENERIC_CONSTANT_WORDS + MD_SB_GENERIC_STATE_WORDS)
  55. #define MD_SB_PERSONALITY_WORDS 64
  56. #define MD_SB_DESCRIPTOR_WORDS 32
  57. #define MD_SB_DISKS 27
  58. #define MD_SB_DISKS_WORDS (MD_SB_DISKS*MD_SB_DESCRIPTOR_WORDS)
  59. #define MD_SB_RESERVED_WORDS (1024 - MD_SB_GENERIC_WORDS - MD_SB_PERSONALITY_WORDS - MD_SB_DISKS_WORDS - MD_SB_DESCRIPTOR_WORDS)
  60. #define MD_SB_EQUAL_WORDS (MD_SB_GENERIC_WORDS + MD_SB_PERSONALITY_WORDS + MD_SB_DISKS_WORDS)
  61. /*
  62. * Device "operational" state bits
  63. */
  64. #define MD_DISK_FAULTY 0 /* disk is faulty / operational */
  65. #define MD_DISK_ACTIVE 1 /* disk is running or spare disk */
  66. #define MD_DISK_SYNC 2 /* disk is in sync with the raid set */
  67. #define MD_DISK_REMOVED 3 /* disk is in sync with the raid set */
  68. #define MD_DISK_CLUSTER_ADD 4 /* Initiate a disk add across the cluster
  69. * For clustered enviroments only.
  70. */
  71. #define MD_DISK_CANDIDATE 5 /* disk is added as spare (local) until confirmed
  72. * For clustered enviroments only.
  73. */
  74. #define MD_DISK_FAILFAST 10 /* Send REQ_FAILFAST if there are multiple
  75. * devices available - and don't try to
  76. * correct read errors.
  77. */
  78. #define MD_DISK_WRITEMOSTLY 9 /* disk is "write-mostly" is RAID1 config.
  79. * read requests will only be sent here in
  80. * dire need
  81. */
  82. #define MD_DISK_JOURNAL 18 /* disk is used as the write journal in RAID-5/6 */
  83. #define MD_DISK_ROLE_SPARE 0xffff
  84. #define MD_DISK_ROLE_FAULTY 0xfffe
  85. #define MD_DISK_ROLE_JOURNAL 0xfffd
  86. #define MD_DISK_ROLE_MAX 0xff00 /* max value of regular disk role */
  87. typedef struct mdp_device_descriptor_s {
  88. __u32 number; /* 0 Device number in the entire set */
  89. __u32 major; /* 1 Device major number */
  90. __u32 minor; /* 2 Device minor number */
  91. __u32 raid_disk; /* 3 The role of the device in the raid set */
  92. __u32 state; /* 4 Operational state */
  93. __u32 reserved[MD_SB_DESCRIPTOR_WORDS - 5];
  94. } mdp_disk_t;
  95. #define MD_SB_MAGIC 0xa92b4efc
  96. /*
  97. * Superblock state bits
  98. */
  99. #define MD_SB_CLEAN 0
  100. #define MD_SB_ERRORS 1
  101. #define MD_SB_CLUSTERED 5 /* MD is clustered */
  102. #define MD_SB_BITMAP_PRESENT 8 /* bitmap may be present nearby */
  103. /*
  104. * Notes:
  105. * - if an array is being reshaped (restriped) in order to change
  106. * the number of active devices in the array, 'raid_disks' will be
  107. * the larger of the old and new numbers. 'delta_disks' will
  108. * be the "new - old". So if +ve, raid_disks is the new value, and
  109. * "raid_disks-delta_disks" is the old. If -ve, raid_disks is the
  110. * old value and "raid_disks+delta_disks" is the new (smaller) value.
  111. */
  112. typedef struct mdp_superblock_s {
  113. /*
  114. * Constant generic information
  115. */
  116. __u32 md_magic; /* 0 MD identifier */
  117. __u32 major_version; /* 1 major version to which the set conforms */
  118. __u32 minor_version; /* 2 minor version ... */
  119. __u32 patch_version; /* 3 patchlevel version ... */
  120. __u32 gvalid_words; /* 4 Number of used words in this section */
  121. __u32 set_uuid0; /* 5 Raid set identifier */
  122. __u32 ctime; /* 6 Creation time */
  123. __u32 level; /* 7 Raid personality */
  124. __u32 size; /* 8 Apparent size of each individual disk */
  125. __u32 nr_disks; /* 9 total disks in the raid set */
  126. __u32 raid_disks; /* 10 disks in a fully functional raid set */
  127. __u32 md_minor; /* 11 preferred MD minor device number */
  128. __u32 not_persistent; /* 12 does it have a persistent superblock */
  129. __u32 set_uuid1; /* 13 Raid set identifier #2 */
  130. __u32 set_uuid2; /* 14 Raid set identifier #3 */
  131. __u32 set_uuid3; /* 15 Raid set identifier #4 */
  132. __u32 gstate_creserved[MD_SB_GENERIC_CONSTANT_WORDS - 16];
  133. /*
  134. * Generic state information
  135. */
  136. __u32 utime; /* 0 Superblock update time */
  137. __u32 state; /* 1 State bits (clean, ...) */
  138. __u32 active_disks; /* 2 Number of currently active disks */
  139. __u32 working_disks; /* 3 Number of working disks */
  140. __u32 failed_disks; /* 4 Number of failed disks */
  141. __u32 spare_disks; /* 5 Number of spare disks */
  142. __u32 sb_csum; /* 6 checksum of the whole superblock */
  143. #if defined(__BYTE_ORDER) ? __BYTE_ORDER == __BIG_ENDIAN : defined(__BIG_ENDIAN)
  144. __u32 events_hi; /* 7 high-order of superblock update count */
  145. __u32 events_lo; /* 8 low-order of superblock update count */
  146. __u32 cp_events_hi; /* 9 high-order of checkpoint update count */
  147. __u32 cp_events_lo; /* 10 low-order of checkpoint update count */
  148. #elif defined(__BYTE_ORDER) ? __BYTE_ORDER == __LITTLE_ENDIAN : defined(__LITTLE_ENDIAN)
  149. __u32 events_lo; /* 7 low-order of superblock update count */
  150. __u32 events_hi; /* 8 high-order of superblock update count */
  151. __u32 cp_events_lo; /* 9 low-order of checkpoint update count */
  152. __u32 cp_events_hi; /* 10 high-order of checkpoint update count */
  153. #else
  154. #error unspecified endianness
  155. #endif
  156. __u32 recovery_cp; /* 11 recovery checkpoint sector count */
  157. /* There are only valid for minor_version > 90 */
  158. __u64 reshape_position; /* 12,13 next address in array-space for reshape */
  159. __u32 new_level; /* 14 new level we are reshaping to */
  160. __u32 delta_disks; /* 15 change in number of raid_disks */
  161. __u32 new_layout; /* 16 new layout */
  162. __u32 new_chunk; /* 17 new chunk size (bytes) */
  163. __u32 gstate_sreserved[MD_SB_GENERIC_STATE_WORDS - 18];
  164. /*
  165. * Personality information
  166. */
  167. __u32 layout; /* 0 the array's physical layout */
  168. __u32 chunk_size; /* 1 chunk size in bytes */
  169. __u32 root_pv; /* 2 LV root PV */
  170. __u32 root_block; /* 3 LV root block */
  171. __u32 pstate_reserved[MD_SB_PERSONALITY_WORDS - 4];
  172. /*
  173. * Disks information
  174. */
  175. mdp_disk_t disks[MD_SB_DISKS];
  176. /*
  177. * Reserved
  178. */
  179. __u32 reserved[MD_SB_RESERVED_WORDS];
  180. /*
  181. * Active descriptor
  182. */
  183. mdp_disk_t this_disk;
  184. } mdp_super_t;
  185. static __inline__ __u64 md_event(mdp_super_t *sb) {
  186. __u64 ev = sb->events_hi;
  187. return (ev<<32)| sb->events_lo;
  188. }
  189. #define MD_SUPERBLOCK_1_TIME_SEC_MASK ((1ULL<<40) - 1)
  190. /*
  191. * The version-1 superblock :
  192. * All numeric fields are little-endian.
  193. *
  194. * total size: 256 bytes plus 2 per device.
  195. * 1K allows 384 devices.
  196. */
  197. struct mdp_superblock_1 {
  198. /* constant array information - 128 bytes */
  199. __le32 magic; /* MD_SB_MAGIC: 0xa92b4efc - little endian */
  200. __le32 major_version; /* 1 */
  201. __le32 feature_map; /* bit 0 set if 'bitmap_offset' is meaningful */
  202. __le32 pad0; /* always set to 0 when writing */
  203. __u8 set_uuid[16]; /* user-space generated. */
  204. char set_name[32]; /* set and interpreted by user-space */
  205. __le64 ctime; /* lo 40 bits are seconds, top 24 are microseconds or 0*/
  206. __le32 level; /* 0,1,4,5 */
  207. __le32 layout; /* only for raid5 and raid10 currently */
  208. __le64 size; /* used size of component devices, in 512byte sectors */
  209. __le32 chunksize; /* in 512byte sectors */
  210. __le32 raid_disks;
  211. union {
  212. __le32 bitmap_offset; /* sectors after start of superblock that bitmap starts
  213. * NOTE: signed, so bitmap can be before superblock
  214. * only meaningful of feature_map[0] is set.
  215. */
  216. /* only meaningful when feature_map[MD_FEATURE_PPL] is set */
  217. struct {
  218. __le16 offset; /* sectors from start of superblock that ppl starts (signed) */
  219. __le16 size; /* ppl size in sectors */
  220. } ppl;
  221. };
  222. /* These are only valid with feature bit '4' */
  223. __le32 new_level; /* new level we are reshaping to */
  224. __le64 reshape_position; /* next address in array-space for reshape */
  225. __le32 delta_disks; /* change in number of raid_disks */
  226. __le32 new_layout; /* new layout */
  227. __le32 new_chunk; /* new chunk size (512byte sectors) */
  228. __le32 new_offset; /* signed number to add to data_offset in new
  229. * layout. 0 == no-change. This can be
  230. * different on each device in the array.
  231. */
  232. /* constant this-device information - 64 bytes */
  233. __le64 data_offset; /* sector start of data, often 0 */
  234. __le64 data_size; /* sectors in this device that can be used for data */
  235. __le64 super_offset; /* sector start of this superblock */
  236. union {
  237. __le64 recovery_offset;/* sectors before this offset (from data_offset) have been recovered */
  238. __le64 journal_tail;/* journal tail of journal device (from data_offset) */
  239. };
  240. __le32 dev_number; /* permanent identifier of this device - not role in raid */
  241. __le32 cnt_corrected_read; /* number of read errors that were corrected by re-writing */
  242. __u8 device_uuid[16]; /* user-space setable, ignored by kernel */
  243. __u8 devflags; /* per-device flags. Only two defined...*/
  244. #define WriteMostly1 1 /* mask for writemostly flag in above */
  245. #define FailFast1 2 /* Should avoid retries and fixups and just fail */
  246. /* Bad block log. If there are any bad blocks the feature flag is set.
  247. * If offset and size are non-zero, that space is reserved and available
  248. */
  249. __u8 bblog_shift; /* shift from sectors to block size */
  250. __le16 bblog_size; /* number of sectors reserved for list */
  251. __le32 bblog_offset; /* sector offset from superblock to bblog,
  252. * signed - not unsigned */
  253. /* array state information - 64 bytes */
  254. __le64 utime; /* 40 bits second, 24 bits microseconds */
  255. __le64 events; /* incremented when superblock updated */
  256. __le64 resync_offset; /* data before this offset (from data_offset) known to be in sync */
  257. __le32 sb_csum; /* checksum up to devs[max_dev] */
  258. __le32 max_dev; /* size of devs[] array to consider */
  259. __u8 pad3[64-32]; /* set to 0 when writing */
  260. /* device state information. Indexed by dev_number.
  261. * 2 bytes per device
  262. * Note there are no per-device state flags. State information is rolled
  263. * into the 'roles' value. If a device is spare or faulty, then it doesn't
  264. * have a meaningful role.
  265. */
  266. __le16 dev_roles[]; /* role in array, or 0xffff for a spare, or 0xfffe for faulty */
  267. };
  268. /* feature_map bits */
  269. #define MD_FEATURE_BITMAP_OFFSET 1
  270. #define MD_FEATURE_RECOVERY_OFFSET 2 /* recovery_offset is present and
  271. * must be honoured
  272. */
  273. #define MD_FEATURE_RESHAPE_ACTIVE 4
  274. #define MD_FEATURE_BAD_BLOCKS 8 /* badblock list is not empty */
  275. #define MD_FEATURE_REPLACEMENT 16 /* This device is replacing an
  276. * active device with same 'role'.
  277. * 'recovery_offset' is also set.
  278. */
  279. #define MD_FEATURE_RESHAPE_BACKWARDS 32 /* Reshape doesn't change number
  280. * of devices, but is going
  281. * backwards anyway.
  282. */
  283. #define MD_FEATURE_NEW_OFFSET 64 /* new_offset must be honoured */
  284. #define MD_FEATURE_RECOVERY_BITMAP 128 /* recovery that is happening
  285. * is guided by bitmap.
  286. */
  287. #define MD_FEATURE_CLUSTERED 256 /* clustered MD */
  288. #define MD_FEATURE_JOURNAL 512 /* support write cache */
  289. #define MD_FEATURE_PPL 1024 /* support PPL */
  290. #define MD_FEATURE_MULTIPLE_PPLS 2048 /* support for multiple PPLs */
  291. #define MD_FEATURE_RAID0_LAYOUT 4096 /* layout is meaningful for RAID0 */
  292. #define MD_FEATURE_ALL (MD_FEATURE_BITMAP_OFFSET \
  293. |MD_FEATURE_RECOVERY_OFFSET \
  294. |MD_FEATURE_RESHAPE_ACTIVE \
  295. |MD_FEATURE_BAD_BLOCKS \
  296. |MD_FEATURE_REPLACEMENT \
  297. |MD_FEATURE_RESHAPE_BACKWARDS \
  298. |MD_FEATURE_NEW_OFFSET \
  299. |MD_FEATURE_RECOVERY_BITMAP \
  300. |MD_FEATURE_CLUSTERED \
  301. |MD_FEATURE_JOURNAL \
  302. |MD_FEATURE_PPL \
  303. |MD_FEATURE_MULTIPLE_PPLS \
  304. |MD_FEATURE_RAID0_LAYOUT \
  305. )
  306. struct r5l_payload_header {
  307. __le16 type;
  308. __le16 flags;
  309. } __attribute__ ((__packed__));
  310. enum r5l_payload_type {
  311. R5LOG_PAYLOAD_DATA = 0,
  312. R5LOG_PAYLOAD_PARITY = 1,
  313. R5LOG_PAYLOAD_FLUSH = 2,
  314. };
  315. struct r5l_payload_data_parity {
  316. struct r5l_payload_header header;
  317. __le32 size; /* sector. data/parity size. each 4k
  318. * has a checksum */
  319. __le64 location; /* sector. For data, it's raid sector. For
  320. * parity, it's stripe sector */
  321. __le32 checksum[];
  322. } __attribute__ ((__packed__));
  323. enum r5l_payload_data_parity_flag {
  324. R5LOG_PAYLOAD_FLAG_DISCARD = 1, /* payload is discard */
  325. /*
  326. * RESHAPED/RESHAPING is only set when there is reshape activity. Note,
  327. * both data/parity of a stripe should have the same flag set
  328. *
  329. * RESHAPED: reshape is running, and this stripe finished reshape
  330. * RESHAPING: reshape is running, and this stripe isn't reshaped
  331. */
  332. R5LOG_PAYLOAD_FLAG_RESHAPED = 2,
  333. R5LOG_PAYLOAD_FLAG_RESHAPING = 3,
  334. };
  335. struct r5l_payload_flush {
  336. struct r5l_payload_header header;
  337. __le32 size; /* flush_stripes size, bytes */
  338. __le64 flush_stripes[];
  339. } __attribute__ ((__packed__));
  340. enum r5l_payload_flush_flag {
  341. R5LOG_PAYLOAD_FLAG_FLUSH_STRIPE = 1, /* data represents whole stripe */
  342. };
  343. struct r5l_meta_block {
  344. __le32 magic;
  345. __le32 checksum;
  346. __u8 version;
  347. __u8 __zero_pading_1;
  348. __le16 __zero_pading_2;
  349. __le32 meta_size; /* whole size of the block */
  350. __le64 seq;
  351. __le64 position; /* sector, start from rdev->data_offset, current position */
  352. struct r5l_payload_header payloads[];
  353. } __attribute__ ((__packed__));
  354. #define R5LOG_VERSION 0x1
  355. #define R5LOG_MAGIC 0x6433c509
  356. struct ppl_header_entry {
  357. __le64 data_sector; /* raid sector of the new data */
  358. __le32 pp_size; /* length of partial parity */
  359. __le32 data_size; /* length of data */
  360. __le32 parity_disk; /* member disk containing parity */
  361. __le32 checksum; /* checksum of partial parity data for this
  362. * entry (~crc32c) */
  363. } __attribute__ ((__packed__));
  364. #define PPL_HEADER_SIZE 4096
  365. #define PPL_HDR_RESERVED 512
  366. #define PPL_HDR_ENTRY_SPACE \
  367. (PPL_HEADER_SIZE - PPL_HDR_RESERVED - 4 * sizeof(__le32) - sizeof(__le64))
  368. #define PPL_HDR_MAX_ENTRIES \
  369. (PPL_HDR_ENTRY_SPACE / sizeof(struct ppl_header_entry))
  370. struct ppl_header {
  371. __u8 reserved[PPL_HDR_RESERVED];/* reserved space, fill with 0xff */
  372. __le32 signature; /* signature (family number of volume) */
  373. __le32 padding; /* zero pad */
  374. __le64 generation; /* generation number of the header */
  375. __le32 entries_count; /* number of entries in entry array */
  376. __le32 checksum; /* checksum of the header (~crc32c) */
  377. struct ppl_header_entry entries[PPL_HDR_MAX_ENTRIES];
  378. } __attribute__ ((__packed__));
  379. #endif