ext4.h 89 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087208820892090209120922093209420952096209720982099210021012102210321042105210621072108210921102111211221132114211521162117211821192120212121222123212421252126212721282129213021312132213321342135213621372138213921402141214221432144214521462147214821492150215121522153215421552156215721582159216021612162216321642165216621672168216921702171217221732174217521762177217821792180218121822183218421852186218721882189219021912192219321942195219621972198219922002201220222032204220522062207220822092210221122122213221422152216221722182219222022212222222322242225222622272228222922302231223222332234223522362237223822392240224122422243224422452246224722482249225022512252225322542255225622572258225922602261226222632264226522662267226822692270227122722273227422752276227722782279228022812282228322842285228622872288228922902291229222932294229522962297229822992300230123022303230423052306230723082309231023112312231323142315231623172318231923202321232223232324232523262327232823292330233123322333233423352336233723382339234023412342234323442345234623472348234923502351235223532354235523562357235823592360236123622363236423652366236723682369237023712372237323742375237623772378237923802381238223832384238523862387238823892390239123922393239423952396239723982399240024012402240324042405240624072408240924102411241224132414241524162417241824192420242124222423242424252426242724282429243024312432243324342435243624372438243924402441244224432444244524462447244824492450245124522453245424552456245724582459246024612462246324642465246624672468246924702471247224732474247524762477247824792480248124822483248424852486248724882489249024912492249324942495249624972498249925002501250225032504250525062507250825092510251125122513251425152516251725182519252025212522
  1. /*
  2. * ext4.h
  3. *
  4. * Copyright (C) 1992, 1993, 1994, 1995
  5. * Remy Card (card@masi.ibp.fr)
  6. * Laboratoire MASI - Institut Blaise Pascal
  7. * Universite Pierre et Marie Curie (Paris VI)
  8. *
  9. * from
  10. *
  11. * linux/include/linux/minix_fs.h
  12. *
  13. * Copyright (C) 1991, 1992 Linus Torvalds
  14. */
  15. #ifndef _EXT4_H
  16. #define _EXT4_H
  17. #include <linux/types.h>
  18. #include <linux/blkdev.h>
  19. #include <linux/magic.h>
  20. #include <linux/jbd2.h>
  21. #include <linux/quota.h>
  22. #include <linux/rwsem.h>
  23. #include <linux/rbtree.h>
  24. #include <linux/seqlock.h>
  25. #include <linux/mutex.h>
  26. #include <linux/timer.h>
  27. #include <linux/version.h>
  28. #include <linux/wait.h>
  29. #include <linux/blockgroup_lock.h>
  30. #include <linux/percpu_counter.h>
  31. #include <crypto/hash.h>
  32. #ifdef __KERNEL__
  33. #include <linux/compat.h>
  34. #endif
  35. /*
  36. * The fourth extended filesystem constants/structures
  37. */
  38. /*
  39. * Define EXT4FS_DEBUG to produce debug messages
  40. */
  41. #undef EXT4FS_DEBUG
  42. /*
  43. * Debug code
  44. */
  45. #ifdef EXT4FS_DEBUG
  46. #define ext4_debug(f, a...) \
  47. do { \
  48. printk(KERN_DEBUG "EXT4-fs DEBUG (%s, %d): %s:", \
  49. __FILE__, __LINE__, __func__); \
  50. printk(KERN_DEBUG f, ## a); \
  51. } while (0)
  52. #else
  53. #define ext4_debug(fmt, ...) no_printk(fmt, ##__VA_ARGS__)
  54. #endif
  55. #define EXT4_ERROR_INODE(inode, fmt, a...) \
  56. ext4_error_inode((inode), __func__, __LINE__, 0, (fmt), ## a)
  57. #define EXT4_ERROR_INODE_BLOCK(inode, block, fmt, a...) \
  58. ext4_error_inode((inode), __func__, __LINE__, (block), (fmt), ## a)
  59. #define EXT4_ERROR_FILE(file, block, fmt, a...) \
  60. ext4_error_file((file), __func__, __LINE__, (block), (fmt), ## a)
  61. /* data type for block offset of block group */
  62. typedef int ext4_grpblk_t;
  63. /* data type for filesystem-wide blocks number */
  64. typedef unsigned long long ext4_fsblk_t;
  65. /* data type for file logical block number */
  66. typedef __u32 ext4_lblk_t;
  67. /* data type for block group number */
  68. typedef unsigned int ext4_group_t;
  69. /*
  70. * Flags used in mballoc's allocation_context flags field.
  71. *
  72. * Also used to show what's going on for debugging purposes when the
  73. * flag field is exported via the traceport interface
  74. */
  75. /* prefer goal again. length */
  76. #define EXT4_MB_HINT_MERGE 0x0001
  77. /* blocks already reserved */
  78. #define EXT4_MB_HINT_RESERVED 0x0002
  79. /* metadata is being allocated */
  80. #define EXT4_MB_HINT_METADATA 0x0004
  81. /* first blocks in the file */
  82. #define EXT4_MB_HINT_FIRST 0x0008
  83. /* search for the best chunk */
  84. #define EXT4_MB_HINT_BEST 0x0010
  85. /* data is being allocated */
  86. #define EXT4_MB_HINT_DATA 0x0020
  87. /* don't preallocate (for tails) */
  88. #define EXT4_MB_HINT_NOPREALLOC 0x0040
  89. /* allocate for locality group */
  90. #define EXT4_MB_HINT_GROUP_ALLOC 0x0080
  91. /* allocate goal blocks or none */
  92. #define EXT4_MB_HINT_GOAL_ONLY 0x0100
  93. /* goal is meaningful */
  94. #define EXT4_MB_HINT_TRY_GOAL 0x0200
  95. /* blocks already pre-reserved by delayed allocation */
  96. #define EXT4_MB_DELALLOC_RESERVED 0x0400
  97. /* We are doing stream allocation */
  98. #define EXT4_MB_STREAM_ALLOC 0x0800
  99. /* Use reserved root blocks if needed */
  100. #define EXT4_MB_USE_ROOT_BLOCKS 0x1000
  101. struct ext4_allocation_request {
  102. /* target inode for block we're allocating */
  103. struct inode *inode;
  104. /* how many blocks we want to allocate */
  105. unsigned int len;
  106. /* logical block in target inode */
  107. ext4_lblk_t logical;
  108. /* the closest logical allocated block to the left */
  109. ext4_lblk_t lleft;
  110. /* the closest logical allocated block to the right */
  111. ext4_lblk_t lright;
  112. /* phys. target (a hint) */
  113. ext4_fsblk_t goal;
  114. /* phys. block for the closest logical allocated block to the left */
  115. ext4_fsblk_t pleft;
  116. /* phys. block for the closest logical allocated block to the right */
  117. ext4_fsblk_t pright;
  118. /* flags. see above EXT4_MB_HINT_* */
  119. unsigned int flags;
  120. };
  121. /*
  122. * Logical to physical block mapping, used by ext4_map_blocks()
  123. *
  124. * This structure is used to pass requests into ext4_map_blocks() as
  125. * well as to store the information returned by ext4_map_blocks(). It
  126. * takes less room on the stack than a struct buffer_head.
  127. */
  128. #define EXT4_MAP_NEW (1 << BH_New)
  129. #define EXT4_MAP_MAPPED (1 << BH_Mapped)
  130. #define EXT4_MAP_UNWRITTEN (1 << BH_Unwritten)
  131. #define EXT4_MAP_BOUNDARY (1 << BH_Boundary)
  132. #define EXT4_MAP_UNINIT (1 << BH_Uninit)
  133. /* Sometimes (in the bigalloc case, from ext4_da_get_block_prep) the caller of
  134. * ext4_map_blocks wants to know whether or not the underlying cluster has
  135. * already been accounted for. EXT4_MAP_FROM_CLUSTER conveys to the caller that
  136. * the requested mapping was from previously mapped (or delayed allocated)
  137. * cluster. We use BH_AllocFromCluster only for this flag. BH_AllocFromCluster
  138. * should never appear on buffer_head's state flags.
  139. */
  140. #define EXT4_MAP_FROM_CLUSTER (1 << BH_AllocFromCluster)
  141. #define EXT4_MAP_FLAGS (EXT4_MAP_NEW | EXT4_MAP_MAPPED |\
  142. EXT4_MAP_UNWRITTEN | EXT4_MAP_BOUNDARY |\
  143. EXT4_MAP_UNINIT | EXT4_MAP_FROM_CLUSTER)
  144. struct ext4_map_blocks {
  145. ext4_fsblk_t m_pblk;
  146. ext4_lblk_t m_lblk;
  147. unsigned int m_len;
  148. unsigned int m_flags;
  149. };
  150. /*
  151. * For delayed allocation tracking
  152. */
  153. struct mpage_da_data {
  154. struct inode *inode;
  155. sector_t b_blocknr; /* start block number of extent */
  156. size_t b_size; /* size of extent */
  157. unsigned long b_state; /* state of the extent */
  158. unsigned long first_page, next_page; /* extent of pages */
  159. struct writeback_control *wbc;
  160. int io_done;
  161. int pages_written;
  162. int retval;
  163. };
  164. /*
  165. * Flags for ext4_io_end->flags
  166. */
  167. #define EXT4_IO_END_UNWRITTEN 0x0001
  168. #define EXT4_IO_END_ERROR 0x0002
  169. #define EXT4_IO_END_QUEUED 0x0004
  170. #define EXT4_IO_END_DIRECT 0x0008
  171. #define EXT4_IO_END_IN_FSYNC 0x0010
  172. struct ext4_io_page {
  173. struct page *p_page;
  174. atomic_t p_count;
  175. };
  176. #define MAX_IO_PAGES 128
  177. /*
  178. * For converting uninitialized extents on a work queue.
  179. *
  180. * 'page' is only used from the writepage() path; 'pages' is only used for
  181. * buffered writes; they are used to keep page references until conversion
  182. * takes place. For AIO/DIO, neither field is filled in.
  183. */
  184. typedef struct ext4_io_end {
  185. struct list_head list; /* per-file finished IO list */
  186. struct inode *inode; /* file being written to */
  187. unsigned int flag; /* unwritten or not */
  188. struct page *page; /* for writepage() path */
  189. loff_t offset; /* offset in the file */
  190. ssize_t size; /* size of the extent */
  191. struct work_struct work; /* data work queue */
  192. struct kiocb *iocb; /* iocb struct for AIO */
  193. int result; /* error value for AIO */
  194. int num_io_pages; /* for writepages() */
  195. struct ext4_io_page *pages[MAX_IO_PAGES]; /* for writepages() */
  196. } ext4_io_end_t;
  197. struct ext4_io_submit {
  198. int io_op;
  199. struct bio *io_bio;
  200. ext4_io_end_t *io_end;
  201. struct ext4_io_page *io_page;
  202. sector_t io_next_block;
  203. };
  204. /*
  205. * Special inodes numbers
  206. */
  207. #define EXT4_BAD_INO 1 /* Bad blocks inode */
  208. #define EXT4_ROOT_INO 2 /* Root inode */
  209. #define EXT4_USR_QUOTA_INO 3 /* User quota inode */
  210. #define EXT4_GRP_QUOTA_INO 4 /* Group quota inode */
  211. #define EXT4_BOOT_LOADER_INO 5 /* Boot loader inode */
  212. #define EXT4_UNDEL_DIR_INO 6 /* Undelete directory inode */
  213. #define EXT4_RESIZE_INO 7 /* Reserved group descriptors inode */
  214. #define EXT4_JOURNAL_INO 8 /* Journal inode */
  215. /* First non-reserved inode for old ext4 filesystems */
  216. #define EXT4_GOOD_OLD_FIRST_INO 11
  217. /*
  218. * Maximal count of links to a file
  219. */
  220. #define EXT4_LINK_MAX 65000
  221. /*
  222. * Macro-instructions used to manage several block sizes
  223. */
  224. #define EXT4_MIN_BLOCK_SIZE 1024
  225. #define EXT4_MAX_BLOCK_SIZE 65536
  226. #define EXT4_MIN_BLOCK_LOG_SIZE 10
  227. #define EXT4_MAX_BLOCK_LOG_SIZE 16
  228. #ifdef __KERNEL__
  229. # define EXT4_BLOCK_SIZE(s) ((s)->s_blocksize)
  230. #else
  231. # define EXT4_BLOCK_SIZE(s) (EXT4_MIN_BLOCK_SIZE << (s)->s_log_block_size)
  232. #endif
  233. #define EXT4_ADDR_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / sizeof(__u32))
  234. #define EXT4_CLUSTER_SIZE(s) (EXT4_BLOCK_SIZE(s) << \
  235. EXT4_SB(s)->s_cluster_bits)
  236. #ifdef __KERNEL__
  237. # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_blocksize_bits)
  238. # define EXT4_CLUSTER_BITS(s) (EXT4_SB(s)->s_cluster_bits)
  239. #else
  240. # define EXT4_BLOCK_SIZE_BITS(s) ((s)->s_log_block_size + 10)
  241. #endif
  242. #ifdef __KERNEL__
  243. #define EXT4_ADDR_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_addr_per_block_bits)
  244. #define EXT4_INODE_SIZE(s) (EXT4_SB(s)->s_inode_size)
  245. #define EXT4_FIRST_INO(s) (EXT4_SB(s)->s_first_ino)
  246. #else
  247. #define EXT4_INODE_SIZE(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
  248. EXT4_GOOD_OLD_INODE_SIZE : \
  249. (s)->s_inode_size)
  250. #define EXT4_FIRST_INO(s) (((s)->s_rev_level == EXT4_GOOD_OLD_REV) ? \
  251. EXT4_GOOD_OLD_FIRST_INO : \
  252. (s)->s_first_ino)
  253. #endif
  254. #define EXT4_BLOCK_ALIGN(size, blkbits) ALIGN((size), (1 << (blkbits)))
  255. /* Translate a block number to a cluster number */
  256. #define EXT4_B2C(sbi, blk) ((blk) >> (sbi)->s_cluster_bits)
  257. /* Translate a cluster number to a block number */
  258. #define EXT4_C2B(sbi, cluster) ((cluster) << (sbi)->s_cluster_bits)
  259. /* Translate # of blks to # of clusters */
  260. #define EXT4_NUM_B2C(sbi, blks) (((blks) + (sbi)->s_cluster_ratio - 1) >> \
  261. (sbi)->s_cluster_bits)
  262. /*
  263. * Structure of a blocks group descriptor
  264. */
  265. struct ext4_group_desc
  266. {
  267. __le32 bg_block_bitmap_lo; /* Blocks bitmap block */
  268. __le32 bg_inode_bitmap_lo; /* Inodes bitmap block */
  269. __le32 bg_inode_table_lo; /* Inodes table block */
  270. __le16 bg_free_blocks_count_lo;/* Free blocks count */
  271. __le16 bg_free_inodes_count_lo;/* Free inodes count */
  272. __le16 bg_used_dirs_count_lo; /* Directories count */
  273. __le16 bg_flags; /* EXT4_BG_flags (INODE_UNINIT, etc) */
  274. __le32 bg_exclude_bitmap_lo; /* Exclude bitmap for snapshots */
  275. __le16 bg_block_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+bbitmap) LE */
  276. __le16 bg_inode_bitmap_csum_lo;/* crc32c(s_uuid+grp_num+ibitmap) LE */
  277. __le16 bg_itable_unused_lo; /* Unused inodes count */
  278. __le16 bg_checksum; /* crc16(sb_uuid+group+desc) */
  279. __le32 bg_block_bitmap_hi; /* Blocks bitmap block MSB */
  280. __le32 bg_inode_bitmap_hi; /* Inodes bitmap block MSB */
  281. __le32 bg_inode_table_hi; /* Inodes table block MSB */
  282. __le16 bg_free_blocks_count_hi;/* Free blocks count MSB */
  283. __le16 bg_free_inodes_count_hi;/* Free inodes count MSB */
  284. __le16 bg_used_dirs_count_hi; /* Directories count MSB */
  285. __le16 bg_itable_unused_hi; /* Unused inodes count MSB */
  286. __le32 bg_exclude_bitmap_hi; /* Exclude bitmap block MSB */
  287. __le16 bg_block_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+bbitmap) BE */
  288. __le16 bg_inode_bitmap_csum_hi;/* crc32c(s_uuid+grp_num+ibitmap) BE */
  289. __u32 bg_reserved;
  290. };
  291. #define EXT4_BG_INODE_BITMAP_CSUM_HI_END \
  292. (offsetof(struct ext4_group_desc, bg_inode_bitmap_csum_hi) + \
  293. sizeof(__le16))
  294. #define EXT4_BG_BLOCK_BITMAP_CSUM_HI_END \
  295. (offsetof(struct ext4_group_desc, bg_block_bitmap_csum_hi) + \
  296. sizeof(__le16))
  297. /*
  298. * Structure of a flex block group info
  299. */
  300. struct flex_groups {
  301. atomic64_t free_clusters;
  302. atomic_t free_inodes;
  303. atomic_t used_dirs;
  304. };
  305. #define EXT4_BG_INODE_UNINIT 0x0001 /* Inode table/bitmap not in use */
  306. #define EXT4_BG_BLOCK_UNINIT 0x0002 /* Block bitmap not in use */
  307. #define EXT4_BG_INODE_ZEROED 0x0004 /* On-disk itable initialized to zero */
  308. /*
  309. * Macro-instructions used to manage group descriptors
  310. */
  311. #define EXT4_MIN_DESC_SIZE 32
  312. #define EXT4_MIN_DESC_SIZE_64BIT 64
  313. #define EXT4_MAX_DESC_SIZE EXT4_MIN_BLOCK_SIZE
  314. #define EXT4_DESC_SIZE(s) (EXT4_SB(s)->s_desc_size)
  315. #ifdef __KERNEL__
  316. # define EXT4_BLOCKS_PER_GROUP(s) (EXT4_SB(s)->s_blocks_per_group)
  317. # define EXT4_CLUSTERS_PER_GROUP(s) (EXT4_SB(s)->s_clusters_per_group)
  318. # define EXT4_DESC_PER_BLOCK(s) (EXT4_SB(s)->s_desc_per_block)
  319. # define EXT4_INODES_PER_GROUP(s) (EXT4_SB(s)->s_inodes_per_group)
  320. # define EXT4_DESC_PER_BLOCK_BITS(s) (EXT4_SB(s)->s_desc_per_block_bits)
  321. #else
  322. # define EXT4_BLOCKS_PER_GROUP(s) ((s)->s_blocks_per_group)
  323. # define EXT4_DESC_PER_BLOCK(s) (EXT4_BLOCK_SIZE(s) / EXT4_DESC_SIZE(s))
  324. # define EXT4_INODES_PER_GROUP(s) ((s)->s_inodes_per_group)
  325. #endif
  326. /*
  327. * Constants relative to the data blocks
  328. */
  329. #define EXT4_NDIR_BLOCKS 12
  330. #define EXT4_IND_BLOCK EXT4_NDIR_BLOCKS
  331. #define EXT4_DIND_BLOCK (EXT4_IND_BLOCK + 1)
  332. #define EXT4_TIND_BLOCK (EXT4_DIND_BLOCK + 1)
  333. #define EXT4_N_BLOCKS (EXT4_TIND_BLOCK + 1)
  334. /*
  335. * Inode flags
  336. */
  337. #define EXT4_SECRM_FL 0x00000001 /* Secure deletion */
  338. #define EXT4_UNRM_FL 0x00000002 /* Undelete */
  339. #define EXT4_COMPR_FL 0x00000004 /* Compress file */
  340. #define EXT4_SYNC_FL 0x00000008 /* Synchronous updates */
  341. #define EXT4_IMMUTABLE_FL 0x00000010 /* Immutable file */
  342. #define EXT4_APPEND_FL 0x00000020 /* writes to file may only append */
  343. #define EXT4_NODUMP_FL 0x00000040 /* do not dump file */
  344. #define EXT4_NOATIME_FL 0x00000080 /* do not update atime */
  345. /* Reserved for compression usage... */
  346. #define EXT4_DIRTY_FL 0x00000100
  347. #define EXT4_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */
  348. #define EXT4_NOCOMPR_FL 0x00000400 /* Don't compress */
  349. #define EXT4_ECOMPR_FL 0x00000800 /* Compression error */
  350. /* End compression flags --- maybe not all used */
  351. #define EXT4_INDEX_FL 0x00001000 /* hash-indexed directory */
  352. #define EXT4_IMAGIC_FL 0x00002000 /* AFS directory */
  353. #define EXT4_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */
  354. #define EXT4_NOTAIL_FL 0x00008000 /* file tail should not be merged */
  355. #define EXT4_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */
  356. #define EXT4_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/
  357. #define EXT4_HUGE_FILE_FL 0x00040000 /* Set to each huge file */
  358. #define EXT4_EXTENTS_FL 0x00080000 /* Inode uses extents */
  359. #define EXT4_EA_INODE_FL 0x00200000 /* Inode used for large EA */
  360. #define EXT4_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */
  361. #define EXT4_RESERVED_FL 0x80000000 /* reserved for ext4 lib */
  362. #define EXT4_FL_USER_VISIBLE 0x004BDFFF /* User visible flags */
  363. #define EXT4_FL_USER_MODIFIABLE 0x004B80FF /* User modifiable flags */
  364. /* Flags that should be inherited by new inodes from their parent. */
  365. #define EXT4_FL_INHERITED (EXT4_SECRM_FL | EXT4_UNRM_FL | EXT4_COMPR_FL |\
  366. EXT4_SYNC_FL | EXT4_NODUMP_FL | EXT4_NOATIME_FL |\
  367. EXT4_NOCOMPR_FL | EXT4_JOURNAL_DATA_FL |\
  368. EXT4_NOTAIL_FL | EXT4_DIRSYNC_FL)
  369. /* Flags that are appropriate for regular files (all but dir-specific ones). */
  370. #define EXT4_REG_FLMASK (~(EXT4_DIRSYNC_FL | EXT4_TOPDIR_FL))
  371. /* Flags that are appropriate for non-directories/regular files. */
  372. #define EXT4_OTHER_FLMASK (EXT4_NODUMP_FL | EXT4_NOATIME_FL)
  373. /* Mask out flags that are inappropriate for the given type of inode. */
  374. static inline __u32 ext4_mask_flags(umode_t mode, __u32 flags)
  375. {
  376. if (S_ISDIR(mode))
  377. return flags;
  378. else if (S_ISREG(mode))
  379. return flags & EXT4_REG_FLMASK;
  380. else
  381. return flags & EXT4_OTHER_FLMASK;
  382. }
  383. /*
  384. * Inode flags used for atomic set/get
  385. */
  386. enum {
  387. EXT4_INODE_SECRM = 0, /* Secure deletion */
  388. EXT4_INODE_UNRM = 1, /* Undelete */
  389. EXT4_INODE_COMPR = 2, /* Compress file */
  390. EXT4_INODE_SYNC = 3, /* Synchronous updates */
  391. EXT4_INODE_IMMUTABLE = 4, /* Immutable file */
  392. EXT4_INODE_APPEND = 5, /* writes to file may only append */
  393. EXT4_INODE_NODUMP = 6, /* do not dump file */
  394. EXT4_INODE_NOATIME = 7, /* do not update atime */
  395. /* Reserved for compression usage... */
  396. EXT4_INODE_DIRTY = 8,
  397. EXT4_INODE_COMPRBLK = 9, /* One or more compressed clusters */
  398. EXT4_INODE_NOCOMPR = 10, /* Don't compress */
  399. EXT4_INODE_ECOMPR = 11, /* Compression error */
  400. /* End compression flags --- maybe not all used */
  401. EXT4_INODE_INDEX = 12, /* hash-indexed directory */
  402. EXT4_INODE_IMAGIC = 13, /* AFS directory */
  403. EXT4_INODE_JOURNAL_DATA = 14, /* file data should be journaled */
  404. EXT4_INODE_NOTAIL = 15, /* file tail should not be merged */
  405. EXT4_INODE_DIRSYNC = 16, /* dirsync behaviour (directories only) */
  406. EXT4_INODE_TOPDIR = 17, /* Top of directory hierarchies*/
  407. EXT4_INODE_HUGE_FILE = 18, /* Set to each huge file */
  408. EXT4_INODE_EXTENTS = 19, /* Inode uses extents */
  409. EXT4_INODE_EA_INODE = 21, /* Inode used for large EA */
  410. EXT4_INODE_EOFBLOCKS = 22, /* Blocks allocated beyond EOF */
  411. EXT4_INODE_RESERVED = 31, /* reserved for ext4 lib */
  412. };
  413. #define TEST_FLAG_VALUE(FLAG) (EXT4_##FLAG##_FL == (1 << EXT4_INODE_##FLAG))
  414. #define CHECK_FLAG_VALUE(FLAG) if (!TEST_FLAG_VALUE(FLAG)) { \
  415. printk(KERN_EMERG "EXT4 flag fail: " #FLAG ": %d %d\n", \
  416. EXT4_##FLAG##_FL, EXT4_INODE_##FLAG); BUG_ON(1); }
  417. /*
  418. * Since it's pretty easy to mix up bit numbers and hex values, and we
  419. * can't do a compile-time test for ENUM values, we use a run-time
  420. * test to make sure that EXT4_XXX_FL is consistent with respect to
  421. * EXT4_INODE_XXX. If all is well the printk and BUG_ON will all drop
  422. * out so it won't cost any extra space in the compiled kernel image.
  423. * But it's important that these values are the same, since we are
  424. * using EXT4_INODE_XXX to test for the flag values, but EXT4_XX_FL
  425. * must be consistent with the values of FS_XXX_FL defined in
  426. * include/linux/fs.h and the on-disk values found in ext2, ext3, and
  427. * ext4 filesystems, and of course the values defined in e2fsprogs.
  428. *
  429. * It's not paranoia if the Murphy's Law really *is* out to get you. :-)
  430. */
  431. static inline void ext4_check_flag_values(void)
  432. {
  433. CHECK_FLAG_VALUE(SECRM);
  434. CHECK_FLAG_VALUE(UNRM);
  435. CHECK_FLAG_VALUE(COMPR);
  436. CHECK_FLAG_VALUE(SYNC);
  437. CHECK_FLAG_VALUE(IMMUTABLE);
  438. CHECK_FLAG_VALUE(APPEND);
  439. CHECK_FLAG_VALUE(NODUMP);
  440. CHECK_FLAG_VALUE(NOATIME);
  441. CHECK_FLAG_VALUE(DIRTY);
  442. CHECK_FLAG_VALUE(COMPRBLK);
  443. CHECK_FLAG_VALUE(NOCOMPR);
  444. CHECK_FLAG_VALUE(ECOMPR);
  445. CHECK_FLAG_VALUE(INDEX);
  446. CHECK_FLAG_VALUE(IMAGIC);
  447. CHECK_FLAG_VALUE(JOURNAL_DATA);
  448. CHECK_FLAG_VALUE(NOTAIL);
  449. CHECK_FLAG_VALUE(DIRSYNC);
  450. CHECK_FLAG_VALUE(TOPDIR);
  451. CHECK_FLAG_VALUE(HUGE_FILE);
  452. CHECK_FLAG_VALUE(EXTENTS);
  453. CHECK_FLAG_VALUE(EA_INODE);
  454. CHECK_FLAG_VALUE(EOFBLOCKS);
  455. CHECK_FLAG_VALUE(RESERVED);
  456. }
  457. /* Used to pass group descriptor data when online resize is done */
  458. struct ext4_new_group_input {
  459. __u32 group; /* Group number for this data */
  460. __u64 block_bitmap; /* Absolute block number of block bitmap */
  461. __u64 inode_bitmap; /* Absolute block number of inode bitmap */
  462. __u64 inode_table; /* Absolute block number of inode table start */
  463. __u32 blocks_count; /* Total number of blocks in this group */
  464. __u16 reserved_blocks; /* Number of reserved blocks in this group */
  465. __u16 unused;
  466. };
  467. #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
  468. struct compat_ext4_new_group_input {
  469. u32 group;
  470. compat_u64 block_bitmap;
  471. compat_u64 inode_bitmap;
  472. compat_u64 inode_table;
  473. u32 blocks_count;
  474. u16 reserved_blocks;
  475. u16 unused;
  476. };
  477. #endif
  478. /* The struct ext4_new_group_input in kernel space, with free_blocks_count */
  479. struct ext4_new_group_data {
  480. __u32 group;
  481. __u64 block_bitmap;
  482. __u64 inode_bitmap;
  483. __u64 inode_table;
  484. __u32 blocks_count;
  485. __u16 reserved_blocks;
  486. __u16 unused;
  487. __u32 free_blocks_count;
  488. };
  489. /* Indexes used to index group tables in ext4_new_group_data */
  490. enum {
  491. BLOCK_BITMAP = 0, /* block bitmap */
  492. INODE_BITMAP, /* inode bitmap */
  493. INODE_TABLE, /* inode tables */
  494. GROUP_TABLE_COUNT,
  495. };
  496. /*
  497. * Flags used by ext4_map_blocks()
  498. */
  499. /* Allocate any needed blocks and/or convert an unitialized
  500. extent to be an initialized ext4 */
  501. #define EXT4_GET_BLOCKS_CREATE 0x0001
  502. /* Request the creation of an unitialized extent */
  503. #define EXT4_GET_BLOCKS_UNINIT_EXT 0x0002
  504. #define EXT4_GET_BLOCKS_CREATE_UNINIT_EXT (EXT4_GET_BLOCKS_UNINIT_EXT|\
  505. EXT4_GET_BLOCKS_CREATE)
  506. /* Caller is from the delayed allocation writeout path,
  507. so set the magic i_delalloc_reserve_flag after taking the
  508. inode allocation semaphore for */
  509. #define EXT4_GET_BLOCKS_DELALLOC_RESERVE 0x0004
  510. /* caller is from the direct IO path, request to creation of an
  511. unitialized extents if not allocated, split the uninitialized
  512. extent if blocks has been preallocated already*/
  513. #define EXT4_GET_BLOCKS_PRE_IO 0x0008
  514. #define EXT4_GET_BLOCKS_CONVERT 0x0010
  515. #define EXT4_GET_BLOCKS_IO_CREATE_EXT (EXT4_GET_BLOCKS_PRE_IO|\
  516. EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
  517. /* Convert extent to initialized after IO complete */
  518. #define EXT4_GET_BLOCKS_IO_CONVERT_EXT (EXT4_GET_BLOCKS_CONVERT|\
  519. EXT4_GET_BLOCKS_CREATE_UNINIT_EXT)
  520. /* Punch out blocks of an extent */
  521. #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020
  522. /* Don't normalize allocation size (used for fallocate) */
  523. #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040
  524. /* Request will not result in inode size update (user for fallocate) */
  525. #define EXT4_GET_BLOCKS_KEEP_SIZE 0x0080
  526. /* Write zeros to newly created written extents */
  527. #define EXT4_GET_BLOCKS_ZERO 0x0200
  528. /*
  529. * Flags used by ext4_free_blocks
  530. */
  531. #define EXT4_FREE_BLOCKS_METADATA 0x0001
  532. #define EXT4_FREE_BLOCKS_FORGET 0x0002
  533. #define EXT4_FREE_BLOCKS_VALIDATED 0x0004
  534. #define EXT4_FREE_BLOCKS_NO_QUOT_UPDATE 0x0008
  535. #define EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER 0x0010
  536. #define EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER 0x0020
  537. /*
  538. * Flags used by ext4_discard_partial_page_buffers
  539. */
  540. #define EXT4_DISCARD_PARTIAL_PG_ZERO_UNMAPPED 0x0001
  541. /*
  542. * ioctl commands
  543. */
  544. #define EXT4_IOC_GETFLAGS FS_IOC_GETFLAGS
  545. #define EXT4_IOC_SETFLAGS FS_IOC_SETFLAGS
  546. #define EXT4_IOC_GETVERSION _IOR('f', 3, long)
  547. #define EXT4_IOC_SETVERSION _IOW('f', 4, long)
  548. #define EXT4_IOC_GETVERSION_OLD FS_IOC_GETVERSION
  549. #define EXT4_IOC_SETVERSION_OLD FS_IOC_SETVERSION
  550. #define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
  551. #define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
  552. #define EXT4_IOC_GROUP_EXTEND _IOW('f', 7, unsigned long)
  553. #define EXT4_IOC_GROUP_ADD _IOW('f', 8, struct ext4_new_group_input)
  554. #define EXT4_IOC_MIGRATE _IO('f', 9)
  555. /* note ioctl 10 reserved for an early version of the FIEMAP ioctl */
  556. /* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
  557. #define EXT4_IOC_ALLOC_DA_BLKS _IO('f', 12)
  558. #define EXT4_IOC_MOVE_EXT _IOWR('f', 15, struct move_extent)
  559. #define EXT4_IOC_RESIZE_FS _IOW('f', 16, __u64)
  560. #if defined(__KERNEL__) && defined(CONFIG_COMPAT)
  561. /*
  562. * ioctl commands in 32 bit emulation
  563. */
  564. #define EXT4_IOC32_GETFLAGS FS_IOC32_GETFLAGS
  565. #define EXT4_IOC32_SETFLAGS FS_IOC32_SETFLAGS
  566. #define EXT4_IOC32_GETVERSION _IOR('f', 3, int)
  567. #define EXT4_IOC32_SETVERSION _IOW('f', 4, int)
  568. #define EXT4_IOC32_GETRSVSZ _IOR('f', 5, int)
  569. #define EXT4_IOC32_SETRSVSZ _IOW('f', 6, int)
  570. #define EXT4_IOC32_GROUP_EXTEND _IOW('f', 7, unsigned int)
  571. #define EXT4_IOC32_GROUP_ADD _IOW('f', 8, struct compat_ext4_new_group_input)
  572. #define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
  573. #define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
  574. #endif
  575. /* Max physical block we can address w/o extents */
  576. #define EXT4_MAX_BLOCK_FILE_PHYS 0xFFFFFFFF
  577. /*
  578. * Structure of an inode on the disk
  579. */
  580. struct ext4_inode {
  581. __le16 i_mode; /* File mode */
  582. __le16 i_uid; /* Low 16 bits of Owner Uid */
  583. __le32 i_size_lo; /* Size in bytes */
  584. __le32 i_atime; /* Access time */
  585. __le32 i_ctime; /* Inode Change time */
  586. __le32 i_mtime; /* Modification time */
  587. __le32 i_dtime; /* Deletion Time */
  588. __le16 i_gid; /* Low 16 bits of Group Id */
  589. __le16 i_links_count; /* Links count */
  590. __le32 i_blocks_lo; /* Blocks count */
  591. __le32 i_flags; /* File flags */
  592. union {
  593. struct {
  594. __le32 l_i_version;
  595. } linux1;
  596. struct {
  597. __u32 h_i_translator;
  598. } hurd1;
  599. struct {
  600. __u32 m_i_reserved1;
  601. } masix1;
  602. } osd1; /* OS dependent 1 */
  603. __le32 i_block[EXT4_N_BLOCKS];/* Pointers to blocks */
  604. __le32 i_generation; /* File version (for NFS) */
  605. __le32 i_file_acl_lo; /* File ACL */
  606. __le32 i_size_high;
  607. __le32 i_obso_faddr; /* Obsoleted fragment address */
  608. union {
  609. struct {
  610. __le16 l_i_blocks_high; /* were l_i_reserved1 */
  611. __le16 l_i_file_acl_high;
  612. __le16 l_i_uid_high; /* these 2 fields */
  613. __le16 l_i_gid_high; /* were reserved2[0] */
  614. __le16 l_i_checksum_lo;/* crc32c(uuid+inum+inode) LE */
  615. __le16 l_i_reserved;
  616. } linux2;
  617. struct {
  618. __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
  619. __u16 h_i_mode_high;
  620. __u16 h_i_uid_high;
  621. __u16 h_i_gid_high;
  622. __u32 h_i_author;
  623. } hurd2;
  624. struct {
  625. __le16 h_i_reserved1; /* Obsoleted fragment number/size which are removed in ext4 */
  626. __le16 m_i_file_acl_high;
  627. __u32 m_i_reserved2[2];
  628. } masix2;
  629. } osd2; /* OS dependent 2 */
  630. __le16 i_extra_isize;
  631. __le16 i_checksum_hi; /* crc32c(uuid+inum+inode) BE */
  632. __le32 i_ctime_extra; /* extra Change time (nsec << 2 | epoch) */
  633. __le32 i_mtime_extra; /* extra Modification time(nsec << 2 | epoch) */
  634. __le32 i_atime_extra; /* extra Access time (nsec << 2 | epoch) */
  635. __le32 i_crtime; /* File Creation time */
  636. __le32 i_crtime_extra; /* extra FileCreationtime (nsec << 2 | epoch) */
  637. __le32 i_version_hi; /* high 32 bits for 64-bit version */
  638. };
  639. struct move_extent {
  640. __u32 reserved; /* should be zero */
  641. __u32 donor_fd; /* donor file descriptor */
  642. __u64 orig_start; /* logical start offset in block for orig */
  643. __u64 donor_start; /* logical start offset in block for donor */
  644. __u64 len; /* block length to be moved */
  645. __u64 moved_len; /* moved block length */
  646. };
  647. #define EXT4_EPOCH_BITS 2
  648. #define EXT4_EPOCH_MASK ((1 << EXT4_EPOCH_BITS) - 1)
  649. #define EXT4_NSEC_MASK (~0UL << EXT4_EPOCH_BITS)
  650. /*
  651. * Extended fields will fit into an inode if the filesystem was formatted
  652. * with large inodes (-I 256 or larger) and there are not currently any EAs
  653. * consuming all of the available space. For new inodes we always reserve
  654. * enough space for the kernel's known extended fields, but for inodes
  655. * created with an old kernel this might not have been the case. None of
  656. * the extended inode fields is critical for correct filesystem operation.
  657. * This macro checks if a certain field fits in the inode. Note that
  658. * inode-size = GOOD_OLD_INODE_SIZE + i_extra_isize
  659. */
  660. #define EXT4_FITS_IN_INODE(ext4_inode, einode, field) \
  661. ((offsetof(typeof(*ext4_inode), field) + \
  662. sizeof((ext4_inode)->field)) \
  663. <= (EXT4_GOOD_OLD_INODE_SIZE + \
  664. (einode)->i_extra_isize)) \
  665. /*
  666. * We use an encoding that preserves the times for extra epoch "00":
  667. *
  668. * extra msb of adjust for signed
  669. * epoch 32-bit 32-bit tv_sec to
  670. * bits time decoded 64-bit tv_sec 64-bit tv_sec valid time range
  671. * 0 0 1 -0x80000000..-0x00000001 0x000000000 1901-12-13..1969-12-31
  672. * 0 0 0 0x000000000..0x07fffffff 0x000000000 1970-01-01..2038-01-19
  673. * 0 1 1 0x080000000..0x0ffffffff 0x100000000 2038-01-19..2106-02-07
  674. * 0 1 0 0x100000000..0x17fffffff 0x100000000 2106-02-07..2174-02-25
  675. * 1 0 1 0x180000000..0x1ffffffff 0x200000000 2174-02-25..2242-03-16
  676. * 1 0 0 0x200000000..0x27fffffff 0x200000000 2242-03-16..2310-04-04
  677. * 1 1 1 0x280000000..0x2ffffffff 0x300000000 2310-04-04..2378-04-22
  678. * 1 1 0 0x300000000..0x37fffffff 0x300000000 2378-04-22..2446-05-10
  679. *
  680. * Note that previous versions of the kernel on 64-bit systems would
  681. * incorrectly use extra epoch bits 1,1 for dates between 1901 and
  682. * 1970. e2fsck will correct this, assuming that it is run on the
  683. * affected filesystem before 2242.
  684. */
  685. static inline __le32 ext4_encode_extra_time(struct timespec *time)
  686. {
  687. u32 extra = sizeof(time->tv_sec) > 4 ?
  688. ((time->tv_sec - (s32)time->tv_sec) >> 32) & EXT4_EPOCH_MASK : 0;
  689. return cpu_to_le32(extra | (time->tv_nsec << EXT4_EPOCH_BITS));
  690. }
  691. static inline void ext4_decode_extra_time(struct timespec *time, __le32 extra)
  692. {
  693. if (unlikely(sizeof(time->tv_sec) > 4 &&
  694. (extra & cpu_to_le32(EXT4_EPOCH_MASK)))) {
  695. #if LINUX_VERSION_CODE < KERNEL_VERSION(4,20,0)
  696. /* Handle legacy encoding of pre-1970 dates with epoch
  697. * bits 1,1. We assume that by kernel version 4.20,
  698. * everyone will have run fsck over the affected
  699. * filesystems to correct the problem. (This
  700. * backwards compatibility may be removed before this
  701. * time, at the discretion of the ext4 developers.)
  702. */
  703. u64 extra_bits = le32_to_cpu(extra) & EXT4_EPOCH_MASK;
  704. if (extra_bits == 3 && ((time->tv_sec) & 0x80000000) != 0)
  705. extra_bits = 0;
  706. time->tv_sec += extra_bits << 32;
  707. #else
  708. time->tv_sec += (u64)(le32_to_cpu(extra) & EXT4_EPOCH_MASK) << 32;
  709. #endif
  710. }
  711. time->tv_nsec = (le32_to_cpu(extra) & EXT4_NSEC_MASK) >> EXT4_EPOCH_BITS;
  712. }
  713. #define EXT4_INODE_SET_XTIME(xtime, inode, raw_inode) \
  714. do { \
  715. (raw_inode)->xtime = cpu_to_le32((inode)->xtime.tv_sec); \
  716. if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
  717. (raw_inode)->xtime ## _extra = \
  718. ext4_encode_extra_time(&(inode)->xtime); \
  719. } while (0)
  720. #define EXT4_EINODE_SET_XTIME(xtime, einode, raw_inode) \
  721. do { \
  722. if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
  723. (raw_inode)->xtime = cpu_to_le32((einode)->xtime.tv_sec); \
  724. if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
  725. (raw_inode)->xtime ## _extra = \
  726. ext4_encode_extra_time(&(einode)->xtime); \
  727. } while (0)
  728. #define EXT4_INODE_GET_XTIME(xtime, inode, raw_inode) \
  729. do { \
  730. (inode)->xtime.tv_sec = (signed)le32_to_cpu((raw_inode)->xtime); \
  731. if (EXT4_FITS_IN_INODE(raw_inode, EXT4_I(inode), xtime ## _extra)) \
  732. ext4_decode_extra_time(&(inode)->xtime, \
  733. raw_inode->xtime ## _extra); \
  734. else \
  735. (inode)->xtime.tv_nsec = 0; \
  736. } while (0)
  737. #define EXT4_EINODE_GET_XTIME(xtime, einode, raw_inode) \
  738. do { \
  739. if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime)) \
  740. (einode)->xtime.tv_sec = \
  741. (signed)le32_to_cpu((raw_inode)->xtime); \
  742. else \
  743. (einode)->xtime.tv_sec = 0; \
  744. if (EXT4_FITS_IN_INODE(raw_inode, einode, xtime ## _extra)) \
  745. ext4_decode_extra_time(&(einode)->xtime, \
  746. raw_inode->xtime ## _extra); \
  747. else \
  748. (einode)->xtime.tv_nsec = 0; \
  749. } while (0)
  750. #define i_disk_version osd1.linux1.l_i_version
  751. #if defined(__KERNEL__) || defined(__linux__)
  752. #define i_reserved1 osd1.linux1.l_i_reserved1
  753. #define i_file_acl_high osd2.linux2.l_i_file_acl_high
  754. #define i_blocks_high osd2.linux2.l_i_blocks_high
  755. #define i_uid_low i_uid
  756. #define i_gid_low i_gid
  757. #define i_uid_high osd2.linux2.l_i_uid_high
  758. #define i_gid_high osd2.linux2.l_i_gid_high
  759. #define i_checksum_lo osd2.linux2.l_i_checksum_lo
  760. #elif defined(__GNU__)
  761. #define i_translator osd1.hurd1.h_i_translator
  762. #define i_uid_high osd2.hurd2.h_i_uid_high
  763. #define i_gid_high osd2.hurd2.h_i_gid_high
  764. #define i_author osd2.hurd2.h_i_author
  765. #elif defined(__masix__)
  766. #define i_reserved1 osd1.masix1.m_i_reserved1
  767. #define i_file_acl_high osd2.masix2.m_i_file_acl_high
  768. #define i_reserved2 osd2.masix2.m_i_reserved2
  769. #endif /* defined(__KERNEL__) || defined(__linux__) */
  770. /*
  771. * storage for cached extent
  772. * If ec_len == 0, then the cache is invalid.
  773. * If ec_start == 0, then the cache represents a gap (null mapping)
  774. */
  775. struct ext4_ext_cache {
  776. ext4_fsblk_t ec_start;
  777. ext4_lblk_t ec_block;
  778. __u32 ec_len; /* must be 32bit to return holes */
  779. };
  780. /*
  781. * fourth extended file system inode data in memory
  782. */
  783. struct ext4_inode_info {
  784. __le32 i_data[15]; /* unconverted */
  785. __u32 i_dtime;
  786. ext4_fsblk_t i_file_acl;
  787. /*
  788. * i_block_group is the number of the block group which contains
  789. * this file's inode. Constant across the lifetime of the inode,
  790. * it is ued for making block allocation decisions - we try to
  791. * place a file's data blocks near its inode block, and new inodes
  792. * near to their parent directory's inode.
  793. */
  794. ext4_group_t i_block_group;
  795. ext4_lblk_t i_dir_start_lookup;
  796. #if (BITS_PER_LONG < 64)
  797. unsigned long i_state_flags; /* Dynamic state flags */
  798. #endif
  799. unsigned long i_flags;
  800. #ifdef CONFIG_EXT4_FS_XATTR
  801. /*
  802. * Extended attributes can be read independently of the main file
  803. * data. Taking i_mutex even when reading would cause contention
  804. * between readers of EAs and writers of regular file data, so
  805. * instead we synchronize on xattr_sem when reading or changing
  806. * EAs.
  807. */
  808. struct rw_semaphore xattr_sem;
  809. #endif
  810. struct list_head i_orphan; /* unlinked but open inodes */
  811. /*
  812. * i_disksize keeps track of what the inode size is ON DISK, not
  813. * in memory. During truncate, i_size is set to the new size by
  814. * the VFS prior to calling ext4_truncate(), but the filesystem won't
  815. * set i_disksize to 0 until the truncate is actually under way.
  816. *
  817. * The intent is that i_disksize always represents the blocks which
  818. * are used by this file. This allows recovery to restart truncate
  819. * on orphans if we crash during truncate. We actually write i_disksize
  820. * into the on-disk inode when writing inodes out, instead of i_size.
  821. *
  822. * The only time when i_disksize and i_size may be different is when
  823. * a truncate is in progress. The only things which change i_disksize
  824. * are ext4_get_block (growth) and ext4_truncate (shrinkth).
  825. */
  826. loff_t i_disksize;
  827. /*
  828. * i_data_sem is for serialising ext4_truncate() against
  829. * ext4_getblock(). In the 2.4 ext2 design, great chunks of inode's
  830. * data tree are chopped off during truncate. We can't do that in
  831. * ext4 because whenever we perform intermediate commits during
  832. * truncate, the inode and all the metadata blocks *must* be in a
  833. * consistent state which allows truncation of the orphans to restart
  834. * during recovery. Hence we must fix the get_block-vs-truncate race
  835. * by other means, so we have i_data_sem.
  836. */
  837. struct rw_semaphore i_data_sem;
  838. struct inode vfs_inode;
  839. struct jbd2_inode *jinode;
  840. struct ext4_ext_cache i_cached_extent;
  841. /*
  842. * File creation time. Its function is same as that of
  843. * struct timespec i_{a,c,m}time in the generic inode.
  844. */
  845. struct timespec i_crtime;
  846. /* mballoc */
  847. struct list_head i_prealloc_list;
  848. spinlock_t i_prealloc_lock;
  849. /* ialloc */
  850. ext4_group_t i_last_alloc_group;
  851. /* allocation reservation info for delalloc */
  852. /* In case of bigalloc, these refer to clusters rather than blocks */
  853. unsigned int i_reserved_data_blocks;
  854. unsigned int i_reserved_meta_blocks;
  855. unsigned int i_allocated_meta_blocks;
  856. ext4_lblk_t i_da_metadata_calc_last_lblock;
  857. int i_da_metadata_calc_len;
  858. /* on-disk additional length */
  859. __u16 i_extra_isize;
  860. #ifdef CONFIG_QUOTA
  861. /* quota space reservation, managed internally by quota code */
  862. qsize_t i_reserved_quota;
  863. #endif
  864. /* completed IOs that might need unwritten extents handling */
  865. struct list_head i_completed_io_list;
  866. spinlock_t i_completed_io_lock;
  867. atomic_t i_ioend_count; /* Number of outstanding io_end structs */
  868. /* current io_end structure for async DIO write*/
  869. ext4_io_end_t *cur_aio_dio;
  870. atomic_t i_aiodio_unwritten; /* Nr. of inflight conversions pending */
  871. spinlock_t i_block_reservation_lock;
  872. /*
  873. * Transactions that contain inode's metadata needed to complete
  874. * fsync and fdatasync, respectively.
  875. */
  876. tid_t i_sync_tid;
  877. tid_t i_datasync_tid;
  878. /* Precomputed uuid+inum+igen checksum for seeding inode checksums */
  879. __u32 i_csum_seed;
  880. };
  881. /*
  882. * File system states
  883. */
  884. #define EXT4_VALID_FS 0x0001 /* Unmounted cleanly */
  885. #define EXT4_ERROR_FS 0x0002 /* Errors detected */
  886. #define EXT4_ORPHAN_FS 0x0004 /* Orphans being recovered */
  887. /*
  888. * Misc. filesystem flags
  889. */
  890. #define EXT2_FLAGS_SIGNED_HASH 0x0001 /* Signed dirhash in use */
  891. #define EXT2_FLAGS_UNSIGNED_HASH 0x0002 /* Unsigned dirhash in use */
  892. #define EXT2_FLAGS_TEST_FILESYS 0x0004 /* to test development code */
  893. /*
  894. * Mount flags
  895. */
  896. #define EXT4_MOUNT_GRPID 0x00004 /* Create files with directory's group */
  897. #define EXT4_MOUNT_DEBUG 0x00008 /* Some debugging messages */
  898. #define EXT4_MOUNT_ERRORS_CONT 0x00010 /* Continue on errors */
  899. #define EXT4_MOUNT_ERRORS_RO 0x00020 /* Remount fs ro on errors */
  900. #define EXT4_MOUNT_ERRORS_PANIC 0x00040 /* Panic on errors */
  901. #define EXT4_MOUNT_ERRORS_MASK 0x00070
  902. #define EXT4_MOUNT_MINIX_DF 0x00080 /* Mimics the Minix statfs */
  903. #define EXT4_MOUNT_NOLOAD 0x00100 /* Don't use existing journal*/
  904. #define EXT4_MOUNT_DATA_FLAGS 0x00C00 /* Mode for data writes: */
  905. #define EXT4_MOUNT_JOURNAL_DATA 0x00400 /* Write data to journal */
  906. #define EXT4_MOUNT_ORDERED_DATA 0x00800 /* Flush data before commit */
  907. #define EXT4_MOUNT_WRITEBACK_DATA 0x00C00 /* No data ordering */
  908. #define EXT4_MOUNT_UPDATE_JOURNAL 0x01000 /* Update the journal format */
  909. #define EXT4_MOUNT_NO_UID32 0x02000 /* Disable 32-bit UIDs */
  910. #define EXT4_MOUNT_XATTR_USER 0x04000 /* Extended user attributes */
  911. #define EXT4_MOUNT_POSIX_ACL 0x08000 /* POSIX Access Control Lists */
  912. #define EXT4_MOUNT_NO_AUTO_DA_ALLOC 0x10000 /* No auto delalloc mapping */
  913. #define EXT4_MOUNT_BARRIER 0x20000 /* Use block barriers */
  914. #define EXT4_MOUNT_NO_SEHASH_XATTR 0x40000 /* Ignore security.sehash extended attribute */
  915. #define EXT4_MOUNT_QUOTA 0x80000 /* Some quota option set */
  916. #define EXT4_MOUNT_USRQUOTA 0x100000 /* "old" user quota */
  917. #define EXT4_MOUNT_GRPQUOTA 0x200000 /* "old" group quota */
  918. #define EXT4_MOUNT_DIOREAD_NOLOCK 0x400000 /* Enable support for dio read nolocking */
  919. #define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000 /* Journal checksums */
  920. #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000 /* Journal Async Commit */
  921. #define EXT4_MOUNT_MBLK_IO_SUBMIT 0x4000000 /* multi-block io submits */
  922. #define EXT4_MOUNT_DELALLOC 0x8000000 /* Delalloc support */
  923. #define EXT4_MOUNT_DATA_ERR_ABORT 0x10000000 /* Abort on file data write */
  924. #define EXT4_MOUNT_BLOCK_VALIDITY 0x20000000 /* Block validity checking */
  925. #define EXT4_MOUNT_DISCARD 0x40000000 /* Issue DISCARD requests */
  926. #define EXT4_MOUNT_INIT_INODE_TABLE 0x80000000 /* Initialize uninitialized itables */
  927. #define EXT4_MOUNT2_EXPLICIT_DELALLOC 0x00000001 /* User explicitly
  928. specified delalloc */
  929. #define clear_opt(sb, opt) EXT4_SB(sb)->s_mount_opt &= \
  930. ~EXT4_MOUNT_##opt
  931. #define set_opt(sb, opt) EXT4_SB(sb)->s_mount_opt |= \
  932. EXT4_MOUNT_##opt
  933. #define test_opt(sb, opt) (EXT4_SB(sb)->s_mount_opt & \
  934. EXT4_MOUNT_##opt)
  935. #define clear_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 &= \
  936. ~EXT4_MOUNT2_##opt
  937. #define set_opt2(sb, opt) EXT4_SB(sb)->s_mount_opt2 |= \
  938. EXT4_MOUNT2_##opt
  939. #define test_opt2(sb, opt) (EXT4_SB(sb)->s_mount_opt2 & \
  940. EXT4_MOUNT2_##opt)
  941. #define ext4_test_and_set_bit __test_and_set_bit_le
  942. #define ext4_set_bit __set_bit_le
  943. #define ext4_set_bit_atomic ext2_set_bit_atomic
  944. #define ext4_test_and_clear_bit __test_and_clear_bit_le
  945. #define ext4_clear_bit __clear_bit_le
  946. #define ext4_clear_bit_atomic ext2_clear_bit_atomic
  947. #define ext4_test_bit test_bit_le
  948. #define ext4_find_next_zero_bit find_next_zero_bit_le
  949. #define ext4_find_next_bit find_next_bit_le
  950. extern void ext4_set_bits(void *bm, int cur, int len);
  951. /*
  952. * Maximal mount counts between two filesystem checks
  953. */
  954. #define EXT4_DFL_MAX_MNT_COUNT 20 /* Allow 20 mounts */
  955. #define EXT4_DFL_CHECKINTERVAL 0 /* Don't use interval check */
  956. /*
  957. * Behaviour when detecting errors
  958. */
  959. #define EXT4_ERRORS_CONTINUE 1 /* Continue execution */
  960. #define EXT4_ERRORS_RO 2 /* Remount fs read-only */
  961. #define EXT4_ERRORS_PANIC 3 /* Panic */
  962. #define EXT4_ERRORS_DEFAULT EXT4_ERRORS_CONTINUE
  963. /* Metadata checksum algorithm codes */
  964. #define EXT4_CRC32C_CHKSUM 1
  965. /*
  966. * Structure of the super block
  967. */
  968. struct ext4_super_block {
  969. /*00*/ __le32 s_inodes_count; /* Inodes count */
  970. __le32 s_blocks_count_lo; /* Blocks count */
  971. __le32 s_r_blocks_count_lo; /* Reserved blocks count */
  972. __le32 s_free_blocks_count_lo; /* Free blocks count */
  973. /*10*/ __le32 s_free_inodes_count; /* Free inodes count */
  974. __le32 s_first_data_block; /* First Data Block */
  975. __le32 s_log_block_size; /* Block size */
  976. __le32 s_log_cluster_size; /* Allocation cluster size */
  977. /*20*/ __le32 s_blocks_per_group; /* # Blocks per group */
  978. __le32 s_clusters_per_group; /* # Clusters per group */
  979. __le32 s_inodes_per_group; /* # Inodes per group */
  980. __le32 s_mtime; /* Mount time */
  981. /*30*/ __le32 s_wtime; /* Write time */
  982. __le16 s_mnt_count; /* Mount count */
  983. __le16 s_max_mnt_count; /* Maximal mount count */
  984. __le16 s_magic; /* Magic signature */
  985. __le16 s_state; /* File system state */
  986. __le16 s_errors; /* Behaviour when detecting errors */
  987. __le16 s_minor_rev_level; /* minor revision level */
  988. /*40*/ __le32 s_lastcheck; /* time of last check */
  989. __le32 s_checkinterval; /* max. time between checks */
  990. __le32 s_creator_os; /* OS */
  991. __le32 s_rev_level; /* Revision level */
  992. /*50*/ __le16 s_def_resuid; /* Default uid for reserved blocks */
  993. __le16 s_def_resgid; /* Default gid for reserved blocks */
  994. /*
  995. * These fields are for EXT4_DYNAMIC_REV superblocks only.
  996. *
  997. * Note: the difference between the compatible feature set and
  998. * the incompatible feature set is that if there is a bit set
  999. * in the incompatible feature set that the kernel doesn't
  1000. * know about, it should refuse to mount the filesystem.
  1001. *
  1002. * e2fsck's requirements are more strict; if it doesn't know
  1003. * about a feature in either the compatible or incompatible
  1004. * feature set, it must abort and not try to meddle with
  1005. * things it doesn't understand...
  1006. */
  1007. __le32 s_first_ino; /* First non-reserved inode */
  1008. __le16 s_inode_size; /* size of inode structure */
  1009. __le16 s_block_group_nr; /* block group # of this superblock */
  1010. __le32 s_feature_compat; /* compatible feature set */
  1011. /*60*/ __le32 s_feature_incompat; /* incompatible feature set */
  1012. __le32 s_feature_ro_compat; /* readonly-compatible feature set */
  1013. /*68*/ __u8 s_uuid[16]; /* 128-bit uuid for volume */
  1014. /*78*/ char s_volume_name[16]; /* volume name */
  1015. /*88*/ char s_last_mounted[64]; /* directory where last mounted */
  1016. /*C8*/ __le32 s_algorithm_usage_bitmap; /* For compression */
  1017. /*
  1018. * Performance hints. Directory preallocation should only
  1019. * happen if the EXT4_FEATURE_COMPAT_DIR_PREALLOC flag is on.
  1020. */
  1021. __u8 s_prealloc_blocks; /* Nr of blocks to try to preallocate*/
  1022. __u8 s_prealloc_dir_blocks; /* Nr to preallocate for dirs */
  1023. __le16 s_reserved_gdt_blocks; /* Per group desc for online growth */
  1024. /*
  1025. * Journaling support valid if EXT4_FEATURE_COMPAT_HAS_JOURNAL set.
  1026. */
  1027. /*D0*/ __u8 s_journal_uuid[16]; /* uuid of journal superblock */
  1028. /*E0*/ __le32 s_journal_inum; /* inode number of journal file */
  1029. __le32 s_journal_dev; /* device number of journal file */
  1030. __le32 s_last_orphan; /* start of list of inodes to delete */
  1031. __le32 s_hash_seed[4]; /* HTREE hash seed */
  1032. __u8 s_def_hash_version; /* Default hash version to use */
  1033. __u8 s_jnl_backup_type;
  1034. __le16 s_desc_size; /* size of group descriptor */
  1035. /*100*/ __le32 s_default_mount_opts;
  1036. __le32 s_first_meta_bg; /* First metablock block group */
  1037. __le32 s_mkfs_time; /* When the filesystem was created */
  1038. __le32 s_jnl_blocks[17]; /* Backup of the journal inode */
  1039. /* 64bit support valid if EXT4_FEATURE_COMPAT_64BIT */
  1040. /*150*/ __le32 s_blocks_count_hi; /* Blocks count */
  1041. __le32 s_r_blocks_count_hi; /* Reserved blocks count */
  1042. __le32 s_free_blocks_count_hi; /* Free blocks count */
  1043. __le16 s_min_extra_isize; /* All inodes have at least # bytes */
  1044. __le16 s_want_extra_isize; /* New inodes should reserve # bytes */
  1045. __le32 s_flags; /* Miscellaneous flags */
  1046. __le16 s_raid_stride; /* RAID stride */
  1047. __le16 s_mmp_update_interval; /* # seconds to wait in MMP checking */
  1048. __le64 s_mmp_block; /* Block for multi-mount protection */
  1049. __le32 s_raid_stripe_width; /* blocks on all data disks (N*stride)*/
  1050. __u8 s_log_groups_per_flex; /* FLEX_BG group size */
  1051. __u8 s_checksum_type; /* metadata checksum algorithm used */
  1052. __le16 s_reserved_pad;
  1053. __le64 s_kbytes_written; /* nr of lifetime kilobytes written */
  1054. __le32 s_snapshot_inum; /* Inode number of active snapshot */
  1055. __le32 s_snapshot_id; /* sequential ID of active snapshot */
  1056. __le64 s_snapshot_r_blocks_count; /* reserved blocks for active
  1057. snapshot's future use */
  1058. __le32 s_snapshot_list; /* inode number of the head of the
  1059. on-disk snapshot list */
  1060. #define EXT4_S_ERR_START offsetof(struct ext4_super_block, s_error_count)
  1061. __le32 s_error_count; /* number of fs errors */
  1062. __le32 s_first_error_time; /* first time an error happened */
  1063. __le32 s_first_error_ino; /* inode involved in first error */
  1064. __le64 s_first_error_block; /* block involved of first error */
  1065. __u8 s_first_error_func[32]; /* function where the error happened */
  1066. __le32 s_first_error_line; /* line number where error happened */
  1067. __le32 s_last_error_time; /* most recent time of an error */
  1068. __le32 s_last_error_ino; /* inode involved in last error */
  1069. __le32 s_last_error_line; /* line number where error happened */
  1070. __le64 s_last_error_block; /* block involved of last error */
  1071. __u8 s_last_error_func[32]; /* function where the error happened */
  1072. #define EXT4_S_ERR_END offsetof(struct ext4_super_block, s_mount_opts)
  1073. __u8 s_mount_opts[64];
  1074. __le32 s_usr_quota_inum; /* inode for tracking user quota */
  1075. __le32 s_grp_quota_inum; /* inode for tracking group quota */
  1076. __le32 s_overhead_clusters; /* overhead blocks/clusters in fs */
  1077. __le32 s_reserved[108]; /* Padding to the end of the block */
  1078. __le32 s_checksum; /* crc32c(superblock) */
  1079. };
  1080. #define EXT4_S_ERR_LEN (EXT4_S_ERR_END - EXT4_S_ERR_START)
  1081. #ifdef __KERNEL__
  1082. /*
  1083. * run-time mount flags
  1084. */
  1085. #define EXT4_MF_MNTDIR_SAMPLED 0x0001
  1086. #define EXT4_MF_FS_ABORTED 0x0002 /* Fatal error detected */
  1087. /*
  1088. * fourth extended-fs super-block data in memory
  1089. */
  1090. struct ext4_sb_info {
  1091. unsigned long s_desc_size; /* Size of a group descriptor in bytes */
  1092. unsigned long s_inodes_per_block;/* Number of inodes per block */
  1093. unsigned long s_blocks_per_group;/* Number of blocks in a group */
  1094. unsigned long s_clusters_per_group; /* Number of clusters in a group */
  1095. unsigned long s_inodes_per_group;/* Number of inodes in a group */
  1096. unsigned long s_itb_per_group; /* Number of inode table blocks per group */
  1097. unsigned long s_gdb_count; /* Number of group descriptor blocks */
  1098. unsigned long s_desc_per_block; /* Number of group descriptors per block */
  1099. ext4_group_t s_groups_count; /* Number of groups in the fs */
  1100. ext4_group_t s_blockfile_groups;/* Groups acceptable for non-extent files */
  1101. unsigned long s_overhead; /* # of fs overhead clusters */
  1102. unsigned int s_cluster_ratio; /* Number of blocks per cluster */
  1103. unsigned int s_cluster_bits; /* log2 of s_cluster_ratio */
  1104. loff_t s_bitmap_maxbytes; /* max bytes for bitmap files */
  1105. struct buffer_head * s_sbh; /* Buffer containing the super block */
  1106. struct ext4_super_block *s_es; /* Pointer to the super block in the buffer */
  1107. struct buffer_head **s_group_desc;
  1108. unsigned int s_mount_opt;
  1109. unsigned int s_mount_opt2;
  1110. unsigned int s_mount_flags;
  1111. unsigned int s_def_mount_opt;
  1112. ext4_fsblk_t s_sb_block;
  1113. atomic64_t s_r_blocks_count;
  1114. uid_t s_resuid;
  1115. gid_t s_resgid;
  1116. unsigned short s_mount_state;
  1117. unsigned short s_pad;
  1118. int s_addr_per_block_bits;
  1119. int s_desc_per_block_bits;
  1120. int s_inode_size;
  1121. int s_first_ino;
  1122. unsigned int s_inode_readahead_blks;
  1123. unsigned int s_inode_goal;
  1124. spinlock_t s_next_gen_lock;
  1125. u32 s_next_generation;
  1126. u32 s_hash_seed[4];
  1127. int s_def_hash_version;
  1128. int s_hash_unsigned; /* 3 if hash should be signed, 0 if not */
  1129. struct percpu_counter s_freeclusters_counter;
  1130. struct percpu_counter s_freeinodes_counter;
  1131. struct percpu_counter s_dirs_counter;
  1132. struct percpu_counter s_dirtyclusters_counter;
  1133. struct blockgroup_lock *s_blockgroup_lock;
  1134. struct proc_dir_entry *s_proc;
  1135. struct kobject s_kobj;
  1136. struct completion s_kobj_unregister;
  1137. /* Journaling */
  1138. struct journal_s *s_journal;
  1139. struct list_head s_orphan;
  1140. struct mutex s_orphan_lock;
  1141. unsigned long s_resize_flags; /* Flags indicating if there
  1142. is a resizer */
  1143. unsigned long s_commit_interval;
  1144. u32 s_max_batch_time;
  1145. u32 s_min_batch_time;
  1146. struct block_device *journal_bdev;
  1147. #ifdef CONFIG_QUOTA
  1148. char *s_qf_names[MAXQUOTAS]; /* Names of quota files with journalled quota */
  1149. int s_jquota_fmt; /* Format of quota to use */
  1150. #endif
  1151. unsigned int s_want_extra_isize; /* New inodes should reserve # bytes */
  1152. struct rb_root system_blks;
  1153. #ifdef EXTENTS_STATS
  1154. /* ext4 extents stats */
  1155. unsigned long s_ext_min;
  1156. unsigned long s_ext_max;
  1157. unsigned long s_depth_max;
  1158. spinlock_t s_ext_stats_lock;
  1159. unsigned long s_ext_blocks;
  1160. unsigned long s_ext_extents;
  1161. #endif
  1162. /* for buddy allocator */
  1163. struct ext4_group_info ***s_group_info;
  1164. struct inode *s_buddy_cache;
  1165. spinlock_t s_md_lock;
  1166. unsigned short *s_mb_offsets;
  1167. unsigned int *s_mb_maxs;
  1168. /* tunables */
  1169. unsigned long s_stripe;
  1170. unsigned int s_mb_stream_request;
  1171. unsigned int s_mb_max_to_scan;
  1172. unsigned int s_mb_min_to_scan;
  1173. unsigned int s_mb_stats;
  1174. unsigned int s_mb_order2_reqs;
  1175. unsigned int s_mb_group_prealloc;
  1176. unsigned int s_max_writeback_mb_bump;
  1177. /* where last allocation was done - for stream allocation */
  1178. unsigned long s_mb_last_group;
  1179. unsigned long s_mb_last_start;
  1180. /* stats for buddy allocator */
  1181. atomic_t s_bal_reqs; /* number of reqs with len > 1 */
  1182. atomic_t s_bal_success; /* we found long enough chunks */
  1183. atomic_t s_bal_allocated; /* in blocks */
  1184. atomic_t s_bal_ex_scanned; /* total extents scanned */
  1185. atomic_t s_bal_goals; /* goal hits */
  1186. atomic_t s_bal_breaks; /* too long searches */
  1187. atomic_t s_bal_2orders; /* 2^order hits */
  1188. spinlock_t s_bal_lock;
  1189. unsigned long s_mb_buddies_generated;
  1190. unsigned long long s_mb_generation_time;
  1191. atomic_t s_mb_lost_chunks;
  1192. atomic_t s_mb_preallocated;
  1193. atomic_t s_mb_discarded;
  1194. atomic_t s_lock_busy;
  1195. /* locality groups */
  1196. struct ext4_locality_group __percpu *s_locality_groups;
  1197. /* for write statistics */
  1198. unsigned long s_sectors_written_start;
  1199. u64 s_kbytes_written;
  1200. unsigned int s_log_groups_per_flex;
  1201. struct flex_groups *s_flex_groups;
  1202. /* workqueue for dio unwritten */
  1203. struct workqueue_struct *dio_unwritten_wq;
  1204. /* timer for periodic error stats printing */
  1205. struct timer_list s_err_report;
  1206. /* Lazy inode table initialization info */
  1207. struct ext4_li_request *s_li_request;
  1208. /* Wait multiplier for lazy initialization thread */
  1209. unsigned int s_li_wait_mult;
  1210. /* Kernel thread for multiple mount protection */
  1211. struct task_struct *s_mmp_tsk;
  1212. /* record the last minlen when FITRIM is called. */
  1213. atomic_t s_last_trim_minblks;
  1214. /* Reference to checksum algorithm driver via cryptoapi */
  1215. struct crypto_shash *s_chksum_driver;
  1216. /* Precomputed FS UUID checksum for seeding other checksums */
  1217. __u32 s_csum_seed;
  1218. };
  1219. static inline struct ext4_sb_info *EXT4_SB(struct super_block *sb)
  1220. {
  1221. return sb->s_fs_info;
  1222. }
  1223. static inline struct ext4_inode_info *EXT4_I(struct inode *inode)
  1224. {
  1225. return container_of(inode, struct ext4_inode_info, vfs_inode);
  1226. }
  1227. static inline struct timespec ext4_current_time(struct inode *inode)
  1228. {
  1229. return (inode->i_sb->s_time_gran < NSEC_PER_SEC) ?
  1230. current_fs_time(inode->i_sb) : CURRENT_TIME_SEC;
  1231. }
  1232. static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
  1233. {
  1234. return ino == EXT4_ROOT_INO ||
  1235. (ino >= EXT4_FIRST_INO(sb) &&
  1236. ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
  1237. }
  1238. static inline void ext4_set_io_unwritten_flag(struct inode *inode,
  1239. struct ext4_io_end *io_end)
  1240. {
  1241. if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
  1242. io_end->flag |= EXT4_IO_END_UNWRITTEN;
  1243. atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten);
  1244. }
  1245. }
  1246. /*
  1247. * Inode dynamic state flags
  1248. */
  1249. enum {
  1250. EXT4_STATE_JDATA, /* journaled data exists */
  1251. EXT4_STATE_NEW, /* inode is newly created */
  1252. EXT4_STATE_XATTR, /* has in-inode xattrs */
  1253. EXT4_STATE_NO_EXPAND, /* No space for expansion */
  1254. EXT4_STATE_DA_ALLOC_CLOSE, /* Alloc DA blks on close */
  1255. EXT4_STATE_EXT_MIGRATE, /* Inode is migrating */
  1256. EXT4_STATE_DIO_UNWRITTEN, /* need convert on dio done*/
  1257. EXT4_STATE_NEWENTRY, /* File just added to dir */
  1258. EXT4_STATE_DELALLOC_RESERVED, /* blks already reserved for delalloc */
  1259. };
  1260. #define EXT4_INODE_BIT_FNS(name, field, offset) \
  1261. static inline int ext4_test_inode_##name(struct inode *inode, int bit) \
  1262. { \
  1263. return test_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
  1264. } \
  1265. static inline void ext4_set_inode_##name(struct inode *inode, int bit) \
  1266. { \
  1267. set_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
  1268. } \
  1269. static inline void ext4_clear_inode_##name(struct inode *inode, int bit) \
  1270. { \
  1271. clear_bit(bit + (offset), &EXT4_I(inode)->i_##field); \
  1272. }
  1273. EXT4_INODE_BIT_FNS(flag, flags, 0)
  1274. #if (BITS_PER_LONG < 64)
  1275. EXT4_INODE_BIT_FNS(state, state_flags, 0)
  1276. static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
  1277. {
  1278. (ei)->i_state_flags = 0;
  1279. }
  1280. #else
  1281. EXT4_INODE_BIT_FNS(state, flags, 32)
  1282. static inline void ext4_clear_state_flags(struct ext4_inode_info *ei)
  1283. {
  1284. /* We depend on the fact that callers will set i_flags */
  1285. }
  1286. #endif
  1287. #else
  1288. /* Assume that user mode programs are passing in an ext4fs superblock, not
  1289. * a kernel struct super_block. This will allow us to call the feature-test
  1290. * macros from user land. */
  1291. #define EXT4_SB(sb) (sb)
  1292. #endif
  1293. #define NEXT_ORPHAN(inode) EXT4_I(inode)->i_dtime
  1294. /*
  1295. * Codes for operating systems
  1296. */
  1297. #define EXT4_OS_LINUX 0
  1298. #define EXT4_OS_HURD 1
  1299. #define EXT4_OS_MASIX 2
  1300. #define EXT4_OS_FREEBSD 3
  1301. #define EXT4_OS_LITES 4
  1302. /*
  1303. * Revision levels
  1304. */
  1305. #define EXT4_GOOD_OLD_REV 0 /* The good old (original) format */
  1306. #define EXT4_DYNAMIC_REV 1 /* V2 format w/ dynamic inode sizes */
  1307. #define EXT4_CURRENT_REV EXT4_GOOD_OLD_REV
  1308. #define EXT4_MAX_SUPP_REV EXT4_DYNAMIC_REV
  1309. #define EXT4_GOOD_OLD_INODE_SIZE 128
  1310. /*
  1311. * Feature set definitions
  1312. */
  1313. #define EXT4_HAS_COMPAT_FEATURE(sb,mask) \
  1314. ((EXT4_SB(sb)->s_es->s_feature_compat & cpu_to_le32(mask)) != 0)
  1315. #define EXT4_HAS_RO_COMPAT_FEATURE(sb,mask) \
  1316. ((EXT4_SB(sb)->s_es->s_feature_ro_compat & cpu_to_le32(mask)) != 0)
  1317. #define EXT4_HAS_INCOMPAT_FEATURE(sb,mask) \
  1318. ((EXT4_SB(sb)->s_es->s_feature_incompat & cpu_to_le32(mask)) != 0)
  1319. #define EXT4_SET_COMPAT_FEATURE(sb,mask) \
  1320. EXT4_SB(sb)->s_es->s_feature_compat |= cpu_to_le32(mask)
  1321. #define EXT4_SET_RO_COMPAT_FEATURE(sb,mask) \
  1322. EXT4_SB(sb)->s_es->s_feature_ro_compat |= cpu_to_le32(mask)
  1323. #define EXT4_SET_INCOMPAT_FEATURE(sb,mask) \
  1324. EXT4_SB(sb)->s_es->s_feature_incompat |= cpu_to_le32(mask)
  1325. #define EXT4_CLEAR_COMPAT_FEATURE(sb,mask) \
  1326. EXT4_SB(sb)->s_es->s_feature_compat &= ~cpu_to_le32(mask)
  1327. #define EXT4_CLEAR_RO_COMPAT_FEATURE(sb,mask) \
  1328. EXT4_SB(sb)->s_es->s_feature_ro_compat &= ~cpu_to_le32(mask)
  1329. #define EXT4_CLEAR_INCOMPAT_FEATURE(sb,mask) \
  1330. EXT4_SB(sb)->s_es->s_feature_incompat &= ~cpu_to_le32(mask)
  1331. #define EXT4_FEATURE_COMPAT_DIR_PREALLOC 0x0001
  1332. #define EXT4_FEATURE_COMPAT_IMAGIC_INODES 0x0002
  1333. #define EXT4_FEATURE_COMPAT_HAS_JOURNAL 0x0004
  1334. #define EXT4_FEATURE_COMPAT_EXT_ATTR 0x0008
  1335. #define EXT4_FEATURE_COMPAT_RESIZE_INODE 0x0010
  1336. #define EXT4_FEATURE_COMPAT_DIR_INDEX 0x0020
  1337. #define EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER 0x0001
  1338. #define EXT4_FEATURE_RO_COMPAT_LARGE_FILE 0x0002
  1339. #define EXT4_FEATURE_RO_COMPAT_BTREE_DIR 0x0004
  1340. #define EXT4_FEATURE_RO_COMPAT_HUGE_FILE 0x0008
  1341. #define EXT4_FEATURE_RO_COMPAT_GDT_CSUM 0x0010
  1342. #define EXT4_FEATURE_RO_COMPAT_DIR_NLINK 0x0020
  1343. #define EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE 0x0040
  1344. #define EXT4_FEATURE_RO_COMPAT_QUOTA 0x0100
  1345. #define EXT4_FEATURE_RO_COMPAT_BIGALLOC 0x0200
  1346. /*
  1347. * METADATA_CSUM also enables group descriptor checksums (GDT_CSUM). When
  1348. * METADATA_CSUM is set, group descriptor checksums use the same algorithm as
  1349. * all other data structures' checksums. However, the METADATA_CSUM and
  1350. * GDT_CSUM bits are mutually exclusive.
  1351. */
  1352. #define EXT4_FEATURE_RO_COMPAT_METADATA_CSUM 0x0400
  1353. #define EXT4_FEATURE_INCOMPAT_COMPRESSION 0x0001
  1354. #define EXT4_FEATURE_INCOMPAT_FILETYPE 0x0002
  1355. #define EXT4_FEATURE_INCOMPAT_RECOVER 0x0004 /* Needs recovery */
  1356. #define EXT4_FEATURE_INCOMPAT_JOURNAL_DEV 0x0008 /* Journal device */
  1357. #define EXT4_FEATURE_INCOMPAT_META_BG 0x0010
  1358. #define EXT4_FEATURE_INCOMPAT_EXTENTS 0x0040 /* extents support */
  1359. #define EXT4_FEATURE_INCOMPAT_64BIT 0x0080
  1360. #define EXT4_FEATURE_INCOMPAT_MMP 0x0100
  1361. #define EXT4_FEATURE_INCOMPAT_FLEX_BG 0x0200
  1362. #define EXT4_FEATURE_INCOMPAT_EA_INODE 0x0400 /* EA in inode */
  1363. #define EXT4_FEATURE_INCOMPAT_DIRDATA 0x1000 /* data in dirent */
  1364. #define EXT4_FEATURE_INCOMPAT_BG_USE_META_CSUM 0x2000 /* use crc32c for bg */
  1365. #define EXT4_FEATURE_INCOMPAT_LARGEDIR 0x4000 /* >2GB or 3-lvl htree */
  1366. #define EXT4_FEATURE_INCOMPAT_INLINEDATA 0x8000 /* data in inode */
  1367. #define EXT2_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
  1368. #define EXT2_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
  1369. EXT4_FEATURE_INCOMPAT_META_BG)
  1370. #define EXT2_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
  1371. EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
  1372. EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
  1373. #define EXT3_FEATURE_COMPAT_SUPP EXT4_FEATURE_COMPAT_EXT_ATTR
  1374. #define EXT3_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
  1375. EXT4_FEATURE_INCOMPAT_RECOVER| \
  1376. EXT4_FEATURE_INCOMPAT_META_BG)
  1377. #define EXT3_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
  1378. EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
  1379. EXT4_FEATURE_RO_COMPAT_BTREE_DIR)
  1380. #define EXT4_FEATURE_COMPAT_SUPP EXT2_FEATURE_COMPAT_EXT_ATTR
  1381. #define EXT4_FEATURE_INCOMPAT_SUPP (EXT4_FEATURE_INCOMPAT_FILETYPE| \
  1382. EXT4_FEATURE_INCOMPAT_RECOVER| \
  1383. EXT4_FEATURE_INCOMPAT_META_BG| \
  1384. EXT4_FEATURE_INCOMPAT_EXTENTS| \
  1385. EXT4_FEATURE_INCOMPAT_64BIT| \
  1386. EXT4_FEATURE_INCOMPAT_FLEX_BG| \
  1387. EXT4_FEATURE_INCOMPAT_MMP)
  1388. #define EXT4_FEATURE_RO_COMPAT_SUPP (EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER| \
  1389. EXT4_FEATURE_RO_COMPAT_LARGE_FILE| \
  1390. EXT4_FEATURE_RO_COMPAT_GDT_CSUM| \
  1391. EXT4_FEATURE_RO_COMPAT_DIR_NLINK | \
  1392. EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE | \
  1393. EXT4_FEATURE_RO_COMPAT_BTREE_DIR |\
  1394. EXT4_FEATURE_RO_COMPAT_HUGE_FILE |\
  1395. EXT4_FEATURE_RO_COMPAT_BIGALLOC)
  1396. /*
  1397. * Default values for user and/or group using reserved blocks
  1398. */
  1399. #define EXT4_DEF_RESUID 0
  1400. #define EXT4_DEF_RESGID 0
  1401. #define EXT4_DEF_INODE_READAHEAD_BLKS 32
  1402. /*
  1403. * Default mount options
  1404. */
  1405. #define EXT4_DEFM_DEBUG 0x0001
  1406. #define EXT4_DEFM_BSDGROUPS 0x0002
  1407. #define EXT4_DEFM_XATTR_USER 0x0004
  1408. #define EXT4_DEFM_ACL 0x0008
  1409. #define EXT4_DEFM_UID16 0x0010
  1410. #define EXT4_DEFM_JMODE 0x0060
  1411. #define EXT4_DEFM_JMODE_DATA 0x0020
  1412. #define EXT4_DEFM_JMODE_ORDERED 0x0040
  1413. #define EXT4_DEFM_JMODE_WBACK 0x0060
  1414. #define EXT4_DEFM_NOBARRIER 0x0100
  1415. #define EXT4_DEFM_BLOCK_VALIDITY 0x0200
  1416. #define EXT4_DEFM_DISCARD 0x0400
  1417. #define EXT4_DEFM_NODELALLOC 0x0800
  1418. /*
  1419. * Default journal batch times
  1420. */
  1421. #define EXT4_DEF_MIN_BATCH_TIME 0
  1422. #define EXT4_DEF_MAX_BATCH_TIME 15000 /* 15ms */
  1423. /*
  1424. * Minimum number of groups in a flexgroup before we separate out
  1425. * directories into the first block group of a flexgroup
  1426. */
  1427. #define EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME 4
  1428. /*
  1429. * Structure of a directory entry
  1430. */
  1431. #define EXT4_NAME_LEN 255
  1432. struct ext4_dir_entry {
  1433. __le32 inode; /* Inode number */
  1434. __le16 rec_len; /* Directory entry length */
  1435. __le16 name_len; /* Name length */
  1436. char name[EXT4_NAME_LEN]; /* File name */
  1437. };
  1438. /*
  1439. * The new version of the directory entry. Since EXT4 structures are
  1440. * stored in intel byte order, and the name_len field could never be
  1441. * bigger than 255 chars, it's safe to reclaim the extra byte for the
  1442. * file_type field.
  1443. */
  1444. struct ext4_dir_entry_2 {
  1445. __le32 inode; /* Inode number */
  1446. __le16 rec_len; /* Directory entry length */
  1447. __u8 name_len; /* Name length */
  1448. __u8 file_type;
  1449. char name[EXT4_NAME_LEN]; /* File name */
  1450. };
  1451. /*
  1452. * This is a bogus directory entry at the end of each leaf block that
  1453. * records checksums.
  1454. */
  1455. struct ext4_dir_entry_tail {
  1456. __le32 det_reserved_zero1; /* Pretend to be unused */
  1457. __le16 det_rec_len; /* 12 */
  1458. __u8 det_reserved_zero2; /* Zero name length */
  1459. __u8 det_reserved_ft; /* 0xDE, fake file type */
  1460. __le32 det_checksum; /* crc32c(uuid+inum+dirblock) */
  1461. };
  1462. /*
  1463. * Ext4 directory file types. Only the low 3 bits are used. The
  1464. * other bits are reserved for now.
  1465. */
  1466. #define EXT4_FT_UNKNOWN 0
  1467. #define EXT4_FT_REG_FILE 1
  1468. #define EXT4_FT_DIR 2
  1469. #define EXT4_FT_CHRDEV 3
  1470. #define EXT4_FT_BLKDEV 4
  1471. #define EXT4_FT_FIFO 5
  1472. #define EXT4_FT_SOCK 6
  1473. #define EXT4_FT_SYMLINK 7
  1474. #define EXT4_FT_MAX 8
  1475. #define EXT4_FT_DIR_CSUM 0xDE
  1476. /*
  1477. * EXT4_DIR_PAD defines the directory entries boundaries
  1478. *
  1479. * NOTE: It must be a multiple of 4
  1480. */
  1481. #define EXT4_DIR_PAD 4
  1482. #define EXT4_DIR_ROUND (EXT4_DIR_PAD - 1)
  1483. #define EXT4_DIR_REC_LEN(name_len) (((name_len) + 8 + EXT4_DIR_ROUND) & \
  1484. ~EXT4_DIR_ROUND)
  1485. #define EXT4_MAX_REC_LEN ((1<<16)-1)
  1486. /*
  1487. * If we ever get support for fs block sizes > page_size, we'll need
  1488. * to remove the #if statements in the next two functions...
  1489. */
  1490. static inline unsigned int
  1491. ext4_rec_len_from_disk(__le16 dlen, unsigned blocksize)
  1492. {
  1493. unsigned len = le16_to_cpu(dlen);
  1494. #if (PAGE_CACHE_SIZE >= 65536)
  1495. if (len == EXT4_MAX_REC_LEN || len == 0)
  1496. return blocksize;
  1497. return (len & 65532) | ((len & 3) << 16);
  1498. #else
  1499. return len;
  1500. #endif
  1501. }
  1502. static inline __le16 ext4_rec_len_to_disk(unsigned len, unsigned blocksize)
  1503. {
  1504. if ((len > blocksize) || (blocksize > (1 << 18)) || (len & 3))
  1505. BUG();
  1506. #if (PAGE_CACHE_SIZE >= 65536)
  1507. if (len < 65536)
  1508. return cpu_to_le16(len);
  1509. if (len == blocksize) {
  1510. if (blocksize == 65536)
  1511. return cpu_to_le16(EXT4_MAX_REC_LEN);
  1512. else
  1513. return cpu_to_le16(0);
  1514. }
  1515. return cpu_to_le16((len & 65532) | ((len >> 16) & 3));
  1516. #else
  1517. return cpu_to_le16(len);
  1518. #endif
  1519. }
  1520. /*
  1521. * Hash Tree Directory indexing
  1522. * (c) Daniel Phillips, 2001
  1523. */
  1524. #define is_dx(dir) (EXT4_HAS_COMPAT_FEATURE(dir->i_sb, \
  1525. EXT4_FEATURE_COMPAT_DIR_INDEX) && \
  1526. ext4_test_inode_flag((dir), EXT4_INODE_INDEX))
  1527. #define EXT4_DIR_LINK_MAX(dir) (!is_dx(dir) && (dir)->i_nlink >= EXT4_LINK_MAX)
  1528. #define EXT4_DIR_LINK_EMPTY(dir) ((dir)->i_nlink == 2 || (dir)->i_nlink == 1)
  1529. /* Legal values for the dx_root hash_version field: */
  1530. #define DX_HASH_LEGACY 0
  1531. #define DX_HASH_HALF_MD4 1
  1532. #define DX_HASH_TEA 2
  1533. #define DX_HASH_LEGACY_UNSIGNED 3
  1534. #define DX_HASH_HALF_MD4_UNSIGNED 4
  1535. #define DX_HASH_TEA_UNSIGNED 5
  1536. static inline u32 ext4_chksum(struct ext4_sb_info *sbi, u32 crc,
  1537. const void *address, unsigned int length)
  1538. {
  1539. struct {
  1540. struct shash_desc shash;
  1541. char ctx[crypto_shash_descsize(sbi->s_chksum_driver)];
  1542. } desc;
  1543. int err;
  1544. desc.shash.tfm = sbi->s_chksum_driver;
  1545. desc.shash.flags = 0;
  1546. *(u32 *)desc.ctx = crc;
  1547. err = crypto_shash_update(&desc.shash, address, length);
  1548. BUG_ON(err);
  1549. return *(u32 *)desc.ctx;
  1550. }
  1551. #ifdef __KERNEL__
  1552. /* hash info structure used by the directory hash */
  1553. struct dx_hash_info
  1554. {
  1555. u32 hash;
  1556. u32 minor_hash;
  1557. int hash_version;
  1558. u32 *seed;
  1559. };
  1560. /* 32 and 64 bit signed EOF for dx directories */
  1561. #define EXT4_HTREE_EOF_32BIT ((1UL << (32 - 1)) - 1)
  1562. #define EXT4_HTREE_EOF_64BIT ((1ULL << (64 - 1)) - 1)
  1563. /*
  1564. * Control parameters used by ext4_htree_next_block
  1565. */
  1566. #define HASH_NB_ALWAYS 1
  1567. /*
  1568. * Describe an inode's exact location on disk and in memory
  1569. */
  1570. struct ext4_iloc
  1571. {
  1572. struct buffer_head *bh;
  1573. unsigned long offset;
  1574. ext4_group_t block_group;
  1575. };
  1576. static inline struct ext4_inode *ext4_raw_inode(struct ext4_iloc *iloc)
  1577. {
  1578. return (struct ext4_inode *) (iloc->bh->b_data + iloc->offset);
  1579. }
  1580. /*
  1581. * This structure is stuffed into the struct file's private_data field
  1582. * for directories. It is where we put information so that we can do
  1583. * readdir operations in hash tree order.
  1584. */
  1585. struct dir_private_info {
  1586. struct rb_root root;
  1587. struct rb_node *curr_node;
  1588. struct fname *extra_fname;
  1589. loff_t last_pos;
  1590. __u32 curr_hash;
  1591. __u32 curr_minor_hash;
  1592. __u32 next_hash;
  1593. };
  1594. /* calculate the first block number of the group */
  1595. static inline ext4_fsblk_t
  1596. ext4_group_first_block_no(struct super_block *sb, ext4_group_t group_no)
  1597. {
  1598. return group_no * (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
  1599. le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block);
  1600. }
  1601. /*
  1602. * Special error return code only used by dx_probe() and its callers.
  1603. */
  1604. #define ERR_BAD_DX_DIR (-(MAX_ERRNO - 1))
  1605. void ext4_get_group_no_and_offset(struct super_block *sb, ext4_fsblk_t blocknr,
  1606. ext4_group_t *blockgrpp, ext4_grpblk_t *offsetp);
  1607. /*
  1608. * Timeout and state flag for lazy initialization inode thread.
  1609. */
  1610. #define EXT4_DEF_LI_WAIT_MULT 10
  1611. #define EXT4_DEF_LI_MAX_START_DELAY 5
  1612. #define EXT4_LAZYINIT_QUIT 0x0001
  1613. #define EXT4_LAZYINIT_RUNNING 0x0002
  1614. /*
  1615. * Lazy inode table initialization info
  1616. */
  1617. struct ext4_lazy_init {
  1618. unsigned long li_state;
  1619. struct list_head li_request_list;
  1620. struct mutex li_list_mtx;
  1621. };
  1622. struct ext4_li_request {
  1623. struct super_block *lr_super;
  1624. struct ext4_sb_info *lr_sbi;
  1625. ext4_group_t lr_next_group;
  1626. struct list_head lr_request;
  1627. unsigned long lr_next_sched;
  1628. unsigned long lr_timeout;
  1629. };
  1630. struct ext4_features {
  1631. struct kobject f_kobj;
  1632. struct completion f_kobj_unregister;
  1633. };
  1634. /*
  1635. * This structure will be used for multiple mount protection. It will be
  1636. * written into the block number saved in the s_mmp_block field in the
  1637. * superblock. Programs that check MMP should assume that if
  1638. * SEQ_FSCK (or any unknown code above SEQ_MAX) is present then it is NOT safe
  1639. * to use the filesystem, regardless of how old the timestamp is.
  1640. */
  1641. #define EXT4_MMP_MAGIC 0x004D4D50U /* ASCII for MMP */
  1642. #define EXT4_MMP_SEQ_CLEAN 0xFF4D4D50U /* mmp_seq value for clean unmount */
  1643. #define EXT4_MMP_SEQ_FSCK 0xE24D4D50U /* mmp_seq value when being fscked */
  1644. #define EXT4_MMP_SEQ_MAX 0xE24D4D4FU /* maximum valid mmp_seq value */
  1645. struct mmp_struct {
  1646. __le32 mmp_magic; /* Magic number for MMP */
  1647. __le32 mmp_seq; /* Sequence no. updated periodically */
  1648. /*
  1649. * mmp_time, mmp_nodename & mmp_bdevname are only used for information
  1650. * purposes and do not affect the correctness of the algorithm
  1651. */
  1652. __le64 mmp_time; /* Time last updated */
  1653. char mmp_nodename[64]; /* Node which last updated MMP block */
  1654. char mmp_bdevname[32]; /* Bdev which last updated MMP block */
  1655. /*
  1656. * mmp_check_interval is used to verify if the MMP block has been
  1657. * updated on the block device. The value is updated based on the
  1658. * maximum time to write the MMP block during an update cycle.
  1659. */
  1660. __le16 mmp_check_interval;
  1661. __le16 mmp_pad1;
  1662. __le32 mmp_pad2[226];
  1663. __le32 mmp_checksum; /* crc32c(uuid+mmp_block) */
  1664. };
  1665. /* arguments passed to the mmp thread */
  1666. struct mmpd_data {
  1667. struct buffer_head *bh; /* bh from initial read_mmp_block() */
  1668. struct super_block *sb; /* super block of the fs */
  1669. };
  1670. /*
  1671. * Check interval multiplier
  1672. * The MMP block is written every update interval and initially checked every
  1673. * update interval x the multiplier (the value is then adapted based on the
  1674. * write latency). The reason is that writes can be delayed under load and we
  1675. * don't want readers to incorrectly assume that the filesystem is no longer
  1676. * in use.
  1677. */
  1678. #define EXT4_MMP_CHECK_MULT 2UL
  1679. /*
  1680. * Minimum interval for MMP checking in seconds.
  1681. */
  1682. #define EXT4_MMP_MIN_CHECK_INTERVAL 5UL
  1683. /*
  1684. * Maximum interval for MMP checking in seconds.
  1685. */
  1686. #define EXT4_MMP_MAX_CHECK_INTERVAL 300UL
  1687. /*
  1688. * Function prototypes
  1689. */
  1690. /*
  1691. * Ok, these declarations are also in <linux/kernel.h> but none of the
  1692. * ext4 source programs needs to include it so they are duplicated here.
  1693. */
  1694. # define NORET_TYPE /**/
  1695. # define ATTRIB_NORET __attribute__((noreturn))
  1696. # define NORET_AND noreturn,
  1697. /* bitmap.c */
  1698. extern unsigned int ext4_count_free(char *bitmap, unsigned numchars);
  1699. void ext4_inode_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
  1700. struct ext4_group_desc *gdp,
  1701. struct buffer_head *bh, int sz);
  1702. int ext4_inode_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
  1703. struct ext4_group_desc *gdp,
  1704. struct buffer_head *bh, int sz);
  1705. void ext4_block_bitmap_csum_set(struct super_block *sb, ext4_group_t group,
  1706. struct ext4_group_desc *gdp,
  1707. struct buffer_head *bh);
  1708. int ext4_block_bitmap_csum_verify(struct super_block *sb, ext4_group_t group,
  1709. struct ext4_group_desc *gdp,
  1710. struct buffer_head *bh);
  1711. /* balloc.c */
  1712. extern void ext4_validate_block_bitmap(struct super_block *sb,
  1713. struct ext4_group_desc *desc,
  1714. unsigned int block_group,
  1715. struct buffer_head *bh);
  1716. extern unsigned int ext4_block_group(struct super_block *sb,
  1717. ext4_fsblk_t blocknr);
  1718. extern ext4_grpblk_t ext4_block_group_offset(struct super_block *sb,
  1719. ext4_fsblk_t blocknr);
  1720. extern int ext4_bg_has_super(struct super_block *sb, ext4_group_t group);
  1721. extern unsigned long ext4_bg_num_gdb(struct super_block *sb,
  1722. ext4_group_t group);
  1723. extern ext4_fsblk_t ext4_new_meta_blocks(handle_t *handle, struct inode *inode,
  1724. ext4_fsblk_t goal,
  1725. unsigned int flags,
  1726. unsigned long *count,
  1727. int *errp);
  1728. extern int ext4_claim_free_clusters(struct ext4_sb_info *sbi,
  1729. s64 nclusters, unsigned int flags);
  1730. extern ext4_fsblk_t ext4_count_free_clusters(struct super_block *);
  1731. extern void ext4_check_blocks_bitmap(struct super_block *);
  1732. extern struct ext4_group_desc * ext4_get_group_desc(struct super_block * sb,
  1733. ext4_group_t block_group,
  1734. struct buffer_head ** bh);
  1735. extern int ext4_should_retry_alloc(struct super_block *sb, int *retries);
  1736. extern struct buffer_head *ext4_read_block_bitmap_nowait(struct super_block *sb,
  1737. ext4_group_t block_group);
  1738. extern int ext4_wait_block_bitmap(struct super_block *sb,
  1739. ext4_group_t block_group,
  1740. struct buffer_head *bh);
  1741. extern struct buffer_head *ext4_read_block_bitmap(struct super_block *sb,
  1742. ext4_group_t block_group);
  1743. extern unsigned ext4_free_clusters_after_init(struct super_block *sb,
  1744. ext4_group_t block_group,
  1745. struct ext4_group_desc *gdp);
  1746. extern unsigned ext4_num_overhead_clusters(struct super_block *sb,
  1747. ext4_group_t block_group,
  1748. struct ext4_group_desc *gdp);
  1749. ext4_fsblk_t ext4_inode_to_goal_block(struct inode *);
  1750. /* dir.c */
  1751. extern int __ext4_check_dir_entry(const char *, unsigned int, struct inode *,
  1752. struct file *,
  1753. struct ext4_dir_entry_2 *,
  1754. struct buffer_head *, unsigned int);
  1755. #define ext4_check_dir_entry(dir, filp, de, bh, offset) \
  1756. unlikely(__ext4_check_dir_entry(__func__, __LINE__, (dir), (filp), \
  1757. (de), (bh), (offset)))
  1758. extern int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
  1759. __u32 minor_hash,
  1760. struct ext4_dir_entry_2 *dirent);
  1761. extern void ext4_htree_free_dir_info(struct dir_private_info *p);
  1762. /* fsync.c */
  1763. extern int ext4_sync_file(struct file *, loff_t, loff_t, int);
  1764. extern int ext4_flush_completed_IO(struct inode *);
  1765. /* hash.c */
  1766. extern int ext4fs_dirhash(const char *name, int len, struct
  1767. dx_hash_info *hinfo);
  1768. /* ialloc.c */
  1769. extern struct inode *__ext4_new_inode(handle_t *, struct inode *, umode_t,
  1770. const struct qstr *qstr, __u32 goal,
  1771. uid_t *owner, int nblocks);
  1772. #define ext4_new_inode(handle, dir, mode, qstr, goal, owner) \
  1773. __ext4_new_inode((handle), (dir), (mode), (qstr), (goal), (owner), 0)
  1774. #define ext4_new_inode_start_handle(dir, mode, qstr, goal, owner, nblocks) \
  1775. __ext4_new_inode(NULL, (dir), (mode), (qstr), (goal), (owner), (nblocks))
  1776. extern void ext4_free_inode(handle_t *, struct inode *);
  1777. extern struct inode * ext4_orphan_get(struct super_block *, unsigned long);
  1778. extern unsigned long ext4_count_free_inodes(struct super_block *);
  1779. extern unsigned long ext4_count_dirs(struct super_block *);
  1780. extern void ext4_check_inodes_bitmap(struct super_block *);
  1781. extern void ext4_mark_bitmap_end(int start_bit, int end_bit, char *bitmap);
  1782. extern int ext4_init_inode_table(struct super_block *sb,
  1783. ext4_group_t group, int barrier);
  1784. extern void ext4_end_bitmap_read(struct buffer_head *bh, int uptodate);
  1785. /* mballoc.c */
  1786. extern long ext4_mb_stats;
  1787. extern long ext4_mb_max_to_scan;
  1788. extern int ext4_mb_init(struct super_block *, int);
  1789. extern int ext4_mb_release(struct super_block *);
  1790. extern ext4_fsblk_t ext4_mb_new_blocks(handle_t *,
  1791. struct ext4_allocation_request *, int *);
  1792. extern int ext4_mb_reserve_blocks(struct super_block *, int);
  1793. extern void ext4_discard_preallocations(struct inode *);
  1794. extern int __init ext4_init_mballoc(void);
  1795. extern void ext4_exit_mballoc(void);
  1796. extern void ext4_free_blocks(handle_t *handle, struct inode *inode,
  1797. struct buffer_head *bh, ext4_fsblk_t block,
  1798. unsigned long count, int flags);
  1799. extern int ext4_mb_add_groupinfo(struct super_block *sb,
  1800. ext4_group_t i, struct ext4_group_desc *desc);
  1801. extern int ext4_group_add_blocks(handle_t *handle, struct super_block *sb,
  1802. ext4_fsblk_t block, unsigned long count);
  1803. extern int ext4_trim_fs(struct super_block *, struct fstrim_range *,
  1804. unsigned long blkdev_flags);
  1805. /* inode.c */
  1806. struct buffer_head *ext4_getblk(handle_t *, struct inode *,
  1807. ext4_lblk_t, int, int *);
  1808. struct buffer_head *ext4_bread(handle_t *, struct inode *,
  1809. ext4_lblk_t, int, int *);
  1810. int ext4_get_block(struct inode *inode, sector_t iblock,
  1811. struct buffer_head *bh_result, int create);
  1812. extern struct inode *ext4_iget(struct super_block *, unsigned long);
  1813. extern struct inode *ext4_iget_normal(struct super_block *, unsigned long);
  1814. extern int ext4_write_inode(struct inode *, struct writeback_control *);
  1815. extern int ext4_setattr(struct dentry *, struct iattr *);
  1816. extern int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
  1817. struct kstat *stat);
  1818. extern void ext4_evict_inode(struct inode *);
  1819. extern void ext4_clear_inode(struct inode *);
  1820. extern int ext4_sync_inode(handle_t *, struct inode *);
  1821. extern void ext4_dirty_inode(struct inode *, int);
  1822. extern int ext4_change_inode_journal_flag(struct inode *, int);
  1823. extern int ext4_get_inode_loc(struct inode *, struct ext4_iloc *);
  1824. extern int ext4_can_truncate(struct inode *inode);
  1825. extern void ext4_truncate(struct inode *);
  1826. extern int ext4_punch_hole(struct file *file, loff_t offset, loff_t length);
  1827. extern int ext4_truncate_restart_trans(handle_t *, struct inode *, int nblocks);
  1828. extern void ext4_set_inode_flags(struct inode *);
  1829. extern void ext4_get_inode_flags(struct ext4_inode_info *);
  1830. extern int ext4_alloc_da_blocks(struct inode *inode);
  1831. extern void ext4_set_aops(struct inode *inode);
  1832. extern int ext4_writepage_trans_blocks(struct inode *);
  1833. extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
  1834. extern int ext4_discard_partial_page_buffers(handle_t *handle,
  1835. struct address_space *mapping, loff_t from,
  1836. loff_t length, int flags);
  1837. extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
  1838. extern qsize_t *ext4_get_reserved_space(struct inode *inode);
  1839. extern void ext4_da_update_reserve_space(struct inode *inode,
  1840. int used, int quota_claim);
  1841. /* indirect.c */
  1842. extern int ext4_ind_map_blocks(handle_t *handle, struct inode *inode,
  1843. struct ext4_map_blocks *map, int flags);
  1844. extern ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
  1845. const struct iovec *iov, loff_t offset,
  1846. unsigned long nr_segs);
  1847. extern int ext4_ind_calc_metadata_amount(struct inode *inode, sector_t lblock);
  1848. extern int ext4_ind_trans_blocks(struct inode *inode, int nrblocks, int chunk);
  1849. extern void ext4_ind_truncate(struct inode *inode);
  1850. /* ioctl.c */
  1851. extern long ext4_ioctl(struct file *, unsigned int, unsigned long);
  1852. extern long ext4_compat_ioctl(struct file *, unsigned int, unsigned long);
  1853. /* migrate.c */
  1854. extern int ext4_ext_migrate(struct inode *);
  1855. /* namei.c */
  1856. extern int ext4_orphan_add(handle_t *, struct inode *);
  1857. extern int ext4_orphan_del(handle_t *, struct inode *);
  1858. extern int ext4_htree_fill_tree(struct file *dir_file, __u32 start_hash,
  1859. __u32 start_minor_hash, __u32 *next_hash);
  1860. /* resize.c */
  1861. extern int ext4_group_add(struct super_block *sb,
  1862. struct ext4_new_group_data *input);
  1863. extern int ext4_group_extend(struct super_block *sb,
  1864. struct ext4_super_block *es,
  1865. ext4_fsblk_t n_blocks_count);
  1866. extern int ext4_resize_fs(struct super_block *sb, ext4_fsblk_t n_blocks_count);
  1867. /* super.c */
  1868. extern int ext4_calculate_overhead(struct super_block *sb);
  1869. extern int ext4_superblock_csum_verify(struct super_block *sb,
  1870. struct ext4_super_block *es);
  1871. extern void ext4_superblock_csum_set(struct super_block *sb,
  1872. struct ext4_super_block *es);
  1873. extern void *ext4_kvmalloc(size_t size, gfp_t flags);
  1874. extern void *ext4_kvzalloc(size_t size, gfp_t flags);
  1875. extern void ext4_kvfree(void *ptr);
  1876. extern __printf(4, 5)
  1877. void __ext4_error(struct super_block *, const char *, unsigned int,
  1878. const char *, ...);
  1879. #define ext4_error(sb, message...) __ext4_error(sb, __func__, \
  1880. __LINE__, ## message)
  1881. extern __printf(5, 6)
  1882. void ext4_error_inode(struct inode *, const char *, unsigned int, ext4_fsblk_t,
  1883. const char *, ...);
  1884. extern __printf(5, 6)
  1885. void ext4_error_file(struct file *, const char *, unsigned int, ext4_fsblk_t,
  1886. const char *, ...);
  1887. extern void __ext4_std_error(struct super_block *, const char *,
  1888. unsigned int, int);
  1889. extern __printf(4, 5)
  1890. void __ext4_abort(struct super_block *, const char *, unsigned int,
  1891. const char *, ...);
  1892. #define ext4_abort(sb, message...) __ext4_abort(sb, __func__, \
  1893. __LINE__, ## message)
  1894. extern __printf(4, 5)
  1895. void __ext4_warning(struct super_block *, const char *, unsigned int,
  1896. const char *, ...);
  1897. #define ext4_warning(sb, message...) __ext4_warning(sb, __func__, \
  1898. __LINE__, ## message)
  1899. extern __printf(3, 4)
  1900. void ext4_msg(struct super_block *, const char *, const char *, ...);
  1901. extern void __dump_mmp_msg(struct super_block *, struct mmp_struct *mmp,
  1902. const char *, unsigned int, const char *);
  1903. #define dump_mmp_msg(sb, mmp, msg) __dump_mmp_msg(sb, mmp, __func__, \
  1904. __LINE__, msg)
  1905. extern __printf(7, 8)
  1906. void __ext4_grp_locked_error(const char *, unsigned int,
  1907. struct super_block *, ext4_group_t,
  1908. unsigned long, ext4_fsblk_t,
  1909. const char *, ...);
  1910. #define ext4_grp_locked_error(sb, grp, message...) \
  1911. __ext4_grp_locked_error(__func__, __LINE__, (sb), (grp), ## message)
  1912. extern void ext4_update_dynamic_rev(struct super_block *sb);
  1913. extern int ext4_update_compat_feature(handle_t *handle, struct super_block *sb,
  1914. __u32 compat);
  1915. extern int ext4_update_rocompat_feature(handle_t *handle,
  1916. struct super_block *sb, __u32 rocompat);
  1917. extern int ext4_update_incompat_feature(handle_t *handle,
  1918. struct super_block *sb, __u32 incompat);
  1919. extern ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
  1920. struct ext4_group_desc *bg);
  1921. extern ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
  1922. struct ext4_group_desc *bg);
  1923. extern ext4_fsblk_t ext4_inode_table(struct super_block *sb,
  1924. struct ext4_group_desc *bg);
  1925. extern __u32 ext4_free_group_clusters(struct super_block *sb,
  1926. struct ext4_group_desc *bg);
  1927. extern __u32 ext4_free_inodes_count(struct super_block *sb,
  1928. struct ext4_group_desc *bg);
  1929. extern __u32 ext4_used_dirs_count(struct super_block *sb,
  1930. struct ext4_group_desc *bg);
  1931. extern __u32 ext4_itable_unused_count(struct super_block *sb,
  1932. struct ext4_group_desc *bg);
  1933. extern void ext4_block_bitmap_set(struct super_block *sb,
  1934. struct ext4_group_desc *bg, ext4_fsblk_t blk);
  1935. extern void ext4_inode_bitmap_set(struct super_block *sb,
  1936. struct ext4_group_desc *bg, ext4_fsblk_t blk);
  1937. extern void ext4_inode_table_set(struct super_block *sb,
  1938. struct ext4_group_desc *bg, ext4_fsblk_t blk);
  1939. extern void ext4_free_group_clusters_set(struct super_block *sb,
  1940. struct ext4_group_desc *bg,
  1941. __u32 count);
  1942. extern void ext4_free_inodes_set(struct super_block *sb,
  1943. struct ext4_group_desc *bg, __u32 count);
  1944. extern void ext4_used_dirs_set(struct super_block *sb,
  1945. struct ext4_group_desc *bg, __u32 count);
  1946. extern void ext4_itable_unused_set(struct super_block *sb,
  1947. struct ext4_group_desc *bg, __u32 count);
  1948. extern int ext4_group_desc_csum_verify(struct super_block *sb, __u32 group,
  1949. struct ext4_group_desc *gdp);
  1950. extern void ext4_group_desc_csum_set(struct super_block *sb, __u32 group,
  1951. struct ext4_group_desc *gdp);
  1952. extern void print_iloc_info(struct super_block *sb,
  1953. struct ext4_iloc iloc);
  1954. extern void print_bh(struct super_block *sb,
  1955. struct buffer_head *bh, int start, int len);
  1956. extern void print_block_data(struct super_block *sb, sector_t blocknr,
  1957. unsigned char *data_to_dump, int start, int len);
  1958. static inline int ext4_has_group_desc_csum(struct super_block *sb)
  1959. {
  1960. return EXT4_HAS_RO_COMPAT_FEATURE(sb,
  1961. EXT4_FEATURE_RO_COMPAT_GDT_CSUM |
  1962. EXT4_FEATURE_RO_COMPAT_METADATA_CSUM);
  1963. }
  1964. static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
  1965. {
  1966. return ((ext4_fsblk_t)le32_to_cpu(es->s_blocks_count_hi) << 32) |
  1967. le32_to_cpu(es->s_blocks_count_lo);
  1968. }
  1969. static inline ext4_fsblk_t ext4_r_blocks_count(struct ext4_super_block *es)
  1970. {
  1971. return ((ext4_fsblk_t)le32_to_cpu(es->s_r_blocks_count_hi) << 32) |
  1972. le32_to_cpu(es->s_r_blocks_count_lo);
  1973. }
  1974. static inline ext4_fsblk_t ext4_free_blocks_count(struct ext4_super_block *es)
  1975. {
  1976. return ((ext4_fsblk_t)le32_to_cpu(es->s_free_blocks_count_hi) << 32) |
  1977. le32_to_cpu(es->s_free_blocks_count_lo);
  1978. }
  1979. static inline void ext4_blocks_count_set(struct ext4_super_block *es,
  1980. ext4_fsblk_t blk)
  1981. {
  1982. es->s_blocks_count_lo = cpu_to_le32((u32)blk);
  1983. es->s_blocks_count_hi = cpu_to_le32(blk >> 32);
  1984. }
  1985. static inline void ext4_free_blocks_count_set(struct ext4_super_block *es,
  1986. ext4_fsblk_t blk)
  1987. {
  1988. es->s_free_blocks_count_lo = cpu_to_le32((u32)blk);
  1989. es->s_free_blocks_count_hi = cpu_to_le32(blk >> 32);
  1990. }
  1991. static inline void ext4_r_blocks_count_set(struct ext4_super_block *es,
  1992. ext4_fsblk_t blk)
  1993. {
  1994. es->s_r_blocks_count_lo = cpu_to_le32((u32)blk);
  1995. es->s_r_blocks_count_hi = cpu_to_le32(blk >> 32);
  1996. }
  1997. static inline loff_t ext4_isize(struct ext4_inode *raw_inode)
  1998. {
  1999. if (S_ISREG(le16_to_cpu(raw_inode->i_mode)))
  2000. return ((loff_t)le32_to_cpu(raw_inode->i_size_high) << 32) |
  2001. le32_to_cpu(raw_inode->i_size_lo);
  2002. else
  2003. return (loff_t) le32_to_cpu(raw_inode->i_size_lo);
  2004. }
  2005. static inline void ext4_isize_set(struct ext4_inode *raw_inode, loff_t i_size)
  2006. {
  2007. raw_inode->i_size_lo = cpu_to_le32(i_size);
  2008. raw_inode->i_size_high = cpu_to_le32(i_size >> 32);
  2009. }
  2010. static inline
  2011. struct ext4_group_info *ext4_get_group_info(struct super_block *sb,
  2012. ext4_group_t group)
  2013. {
  2014. struct ext4_group_info ***grp_info;
  2015. long indexv, indexh;
  2016. grp_info = EXT4_SB(sb)->s_group_info;
  2017. indexv = group >> (EXT4_DESC_PER_BLOCK_BITS(sb));
  2018. indexh = group & ((EXT4_DESC_PER_BLOCK(sb)) - 1);
  2019. return grp_info[indexv][indexh];
  2020. }
  2021. /*
  2022. * Reading s_groups_count requires using smp_rmb() afterwards. See
  2023. * the locking protocol documented in the comments of ext4_group_add()
  2024. * in resize.c
  2025. */
  2026. static inline ext4_group_t ext4_get_groups_count(struct super_block *sb)
  2027. {
  2028. ext4_group_t ngroups = EXT4_SB(sb)->s_groups_count;
  2029. smp_rmb();
  2030. return ngroups;
  2031. }
  2032. static inline ext4_group_t ext4_flex_group(struct ext4_sb_info *sbi,
  2033. ext4_group_t block_group)
  2034. {
  2035. return block_group >> sbi->s_log_groups_per_flex;
  2036. }
  2037. static inline unsigned int ext4_flex_bg_size(struct ext4_sb_info *sbi)
  2038. {
  2039. return 1 << sbi->s_log_groups_per_flex;
  2040. }
  2041. #define ext4_std_error(sb, errno) \
  2042. do { \
  2043. if ((errno)) \
  2044. __ext4_std_error((sb), __func__, __LINE__, (errno)); \
  2045. } while (0)
  2046. #ifdef CONFIG_SMP
  2047. /* Each CPU can accumulate percpu_counter_batch clusters in their local
  2048. * counters. So we need to make sure we have free clusters more
  2049. * than percpu_counter_batch * nr_cpu_ids. Also add a window of 4 times.
  2050. */
  2051. #define EXT4_FREECLUSTERS_WATERMARK (4 * (percpu_counter_batch * nr_cpu_ids))
  2052. #else
  2053. #define EXT4_FREECLUSTERS_WATERMARK 0
  2054. #endif
  2055. static inline void ext4_update_i_disksize(struct inode *inode, loff_t newsize)
  2056. {
  2057. /*
  2058. * XXX: replace with spinlock if seen contended -bzzz
  2059. */
  2060. down_write(&EXT4_I(inode)->i_data_sem);
  2061. if (newsize > EXT4_I(inode)->i_disksize)
  2062. EXT4_I(inode)->i_disksize = newsize;
  2063. up_write(&EXT4_I(inode)->i_data_sem);
  2064. return ;
  2065. }
  2066. struct ext4_group_info {
  2067. unsigned long bb_state;
  2068. struct rb_root bb_free_root;
  2069. ext4_grpblk_t bb_first_free; /* first free block */
  2070. ext4_grpblk_t bb_free; /* total free blocks */
  2071. ext4_grpblk_t bb_fragments; /* nr of freespace fragments */
  2072. ext4_grpblk_t bb_largest_free_order;/* order of largest frag in BG */
  2073. struct list_head bb_prealloc_list;
  2074. #ifdef DOUBLE_CHECK
  2075. void *bb_bitmap;
  2076. #endif
  2077. struct rw_semaphore alloc_sem;
  2078. ext4_grpblk_t bb_counters[]; /* Nr of free power-of-two-block
  2079. * regions, index is order.
  2080. * bb_counters[3] = 5 means
  2081. * 5 free 8-block regions. */
  2082. };
  2083. #define EXT4_GROUP_INFO_NEED_INIT_BIT 0
  2084. #define EXT4_GROUP_INFO_WAS_TRIMMED_BIT 1
  2085. #define EXT4_MB_GRP_NEED_INIT(grp) \
  2086. (test_bit(EXT4_GROUP_INFO_NEED_INIT_BIT, &((grp)->bb_state)))
  2087. #define EXT4_MB_GRP_WAS_TRIMMED(grp) \
  2088. (test_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
  2089. #define EXT4_MB_GRP_SET_TRIMMED(grp) \
  2090. (set_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
  2091. #define EXT4_MB_GRP_CLEAR_TRIMMED(grp) \
  2092. (clear_bit(EXT4_GROUP_INFO_WAS_TRIMMED_BIT, &((grp)->bb_state)))
  2093. #define EXT4_MAX_CONTENTION 8
  2094. #define EXT4_CONTENTION_THRESHOLD 2
  2095. static inline spinlock_t *ext4_group_lock_ptr(struct super_block *sb,
  2096. ext4_group_t group)
  2097. {
  2098. return bgl_lock_ptr(EXT4_SB(sb)->s_blockgroup_lock, group);
  2099. }
  2100. /*
  2101. * Returns true if the filesystem is busy enough that attempts to
  2102. * access the block group locks has run into contention.
  2103. */
  2104. static inline int ext4_fs_is_busy(struct ext4_sb_info *sbi)
  2105. {
  2106. return (atomic_read(&sbi->s_lock_busy) > EXT4_CONTENTION_THRESHOLD);
  2107. }
  2108. static inline void ext4_lock_group(struct super_block *sb, ext4_group_t group)
  2109. {
  2110. spinlock_t *lock = ext4_group_lock_ptr(sb, group);
  2111. if (spin_trylock(lock))
  2112. /*
  2113. * We're able to grab the lock right away, so drop the
  2114. * lock contention counter.
  2115. */
  2116. atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, -1, 0);
  2117. else {
  2118. /*
  2119. * The lock is busy, so bump the contention counter,
  2120. * and then wait on the spin lock.
  2121. */
  2122. atomic_add_unless(&EXT4_SB(sb)->s_lock_busy, 1,
  2123. EXT4_MAX_CONTENTION);
  2124. spin_lock(lock);
  2125. }
  2126. }
  2127. static inline void ext4_unlock_group(struct super_block *sb,
  2128. ext4_group_t group)
  2129. {
  2130. spin_unlock(ext4_group_lock_ptr(sb, group));
  2131. }
  2132. static inline void ext4_mark_super_dirty(struct super_block *sb)
  2133. {
  2134. struct ext4_super_block *es = EXT4_SB(sb)->s_es;
  2135. ext4_superblock_csum_set(sb, es);
  2136. if (EXT4_SB(sb)->s_journal == NULL)
  2137. sb->s_dirt =1;
  2138. }
  2139. /*
  2140. * Block validity checking
  2141. */
  2142. #define ext4_check_indirect_blockref(inode, bh) \
  2143. ext4_check_blockref(__func__, __LINE__, inode, \
  2144. (__le32 *)(bh)->b_data, \
  2145. EXT4_ADDR_PER_BLOCK((inode)->i_sb))
  2146. #define ext4_ind_check_inode(inode) \
  2147. ext4_check_blockref(__func__, __LINE__, inode, \
  2148. EXT4_I(inode)->i_data, \
  2149. EXT4_NDIR_BLOCKS)
  2150. /*
  2151. * Inodes and files operations
  2152. */
  2153. /* dir.c */
  2154. extern const struct file_operations ext4_dir_operations;
  2155. /* file.c */
  2156. extern const struct inode_operations ext4_file_inode_operations;
  2157. extern const struct file_operations ext4_file_operations;
  2158. extern loff_t ext4_llseek(struct file *file, loff_t offset, int origin);
  2159. /* namei.c */
  2160. extern const struct inode_operations ext4_dir_inode_operations;
  2161. extern const struct inode_operations ext4_special_inode_operations;
  2162. extern struct dentry *ext4_get_parent(struct dentry *child);
  2163. /* symlink.c */
  2164. extern const struct inode_operations ext4_symlink_inode_operations;
  2165. extern const struct inode_operations ext4_fast_symlink_inode_operations;
  2166. /* block_validity */
  2167. extern void ext4_release_system_zone(struct super_block *sb);
  2168. extern int ext4_setup_system_zone(struct super_block *sb);
  2169. extern int __init ext4_init_system_zone(void);
  2170. extern void ext4_exit_system_zone(void);
  2171. extern int ext4_data_block_valid(struct ext4_sb_info *sbi,
  2172. ext4_fsblk_t start_blk,
  2173. unsigned int count);
  2174. extern int ext4_check_blockref(const char *, unsigned int,
  2175. struct inode *, __le32 *, unsigned int);
  2176. /* extents.c */
  2177. extern int ext4_ext_tree_init(handle_t *handle, struct inode *);
  2178. extern int ext4_ext_writepage_trans_blocks(struct inode *, int);
  2179. extern int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks,
  2180. int chunk);
  2181. extern int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
  2182. struct ext4_map_blocks *map, int flags);
  2183. extern void ext4_ext_truncate(struct inode *);
  2184. extern int ext4_ext_punch_hole(struct file *file, loff_t offset,
  2185. loff_t length);
  2186. extern void ext4_ext_init(struct super_block *);
  2187. extern void ext4_ext_release(struct super_block *);
  2188. extern long ext4_fallocate(struct file *file, int mode, loff_t offset,
  2189. loff_t len);
  2190. extern int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
  2191. ssize_t len);
  2192. extern int ext4_map_blocks(handle_t *handle, struct inode *inode,
  2193. struct ext4_map_blocks *map, int flags);
  2194. extern int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
  2195. __u64 start, __u64 len);
  2196. /* move_extent.c */
  2197. extern int ext4_move_extents(struct file *o_filp, struct file *d_filp,
  2198. __u64 start_orig, __u64 start_donor,
  2199. __u64 len, __u64 *moved_len);
  2200. /* page-io.c */
  2201. extern int __init ext4_init_pageio(void);
  2202. extern void ext4_exit_pageio(void);
  2203. extern void ext4_ioend_wait(struct inode *);
  2204. extern void ext4_free_io_end(ext4_io_end_t *io);
  2205. extern ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags);
  2206. extern int ext4_end_io_nolock(ext4_io_end_t *io);
  2207. extern void ext4_io_submit(struct ext4_io_submit *io);
  2208. extern int ext4_bio_write_page(struct ext4_io_submit *io,
  2209. struct page *page,
  2210. int len,
  2211. struct writeback_control *wbc);
  2212. /* mmp.c */
  2213. extern int ext4_multi_mount_protect(struct super_block *, ext4_fsblk_t);
  2214. /* BH_Uninit flag: blocks are allocated but uninitialized on disk */
  2215. enum ext4_state_bits {
  2216. BH_Uninit /* blocks are allocated but uninitialized on disk */
  2217. = BH_JBDPrivateStart,
  2218. BH_AllocFromCluster, /* allocated blocks were part of already
  2219. * allocated cluster. Note that this flag will
  2220. * never, ever appear in a buffer_head's state
  2221. * flag. See EXT4_MAP_FROM_CLUSTER to see where
  2222. * this is used. */
  2223. BH_Da_Mapped, /* Delayed allocated block that now has a mapping. This
  2224. * flag is set when ext4_map_blocks is called on a
  2225. * delayed allocated block to get its real mapping. */
  2226. };
  2227. BUFFER_FNS(Uninit, uninit)
  2228. TAS_BUFFER_FNS(Uninit, uninit)
  2229. BUFFER_FNS(Da_Mapped, da_mapped)
  2230. /*
  2231. * Add new method to test wether block and inode bitmaps are properly
  2232. * initialized. With uninit_bg reading the block from disk is not enough
  2233. * to mark the bitmap uptodate. We need to also zero-out the bitmap
  2234. */
  2235. #define BH_BITMAP_UPTODATE BH_JBDPrivateStart
  2236. static inline int bitmap_uptodate(struct buffer_head *bh)
  2237. {
  2238. return (buffer_uptodate(bh) &&
  2239. test_bit(BH_BITMAP_UPTODATE, &(bh)->b_state));
  2240. }
  2241. static inline void set_bitmap_uptodate(struct buffer_head *bh)
  2242. {
  2243. set_bit(BH_BITMAP_UPTODATE, &(bh)->b_state);
  2244. }
  2245. #define in_range(b, first, len) ((b) >= (first) && (b) <= (first) + (len) - 1)
  2246. /* For ioend & aio unwritten conversion wait queues */
  2247. #define EXT4_WQ_HASH_SZ 37
  2248. #define ext4_ioend_wq(v) (&ext4__ioend_wq[((unsigned long)(v)) %\
  2249. EXT4_WQ_HASH_SZ])
  2250. #define ext4_aio_mutex(v) (&ext4__aio_mutex[((unsigned long)(v)) %\
  2251. EXT4_WQ_HASH_SZ])
  2252. extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
  2253. extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
  2254. #define EXT4_RESIZING 0
  2255. extern int ext4_resize_begin(struct super_block *sb);
  2256. extern void ext4_resize_end(struct super_block *sb);
  2257. #endif /* __KERNEL__ */
  2258. #include "ext4_extents.h"
  2259. #endif /* _EXT4_H */