ChangeLog 83 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521
  1. ToDo/Notes:
  2. - Find and fix bugs.
  3. - In between ntfs_prepare/commit_write, need exclusion between
  4. simultaneous file extensions. This is given to us by holding i_sem
  5. on the inode. The only places in the kernel when a file is resized
  6. are prepare/commit write and truncate for both of which i_sem is
  7. held. Just have to be careful in readpage/writepage and all other
  8. helpers not running under i_sem that we play nice...
  9. Also need to be careful with initialized_size extention in
  10. ntfs_prepare_write. Basically, just be _very_ careful in this code...
  11. UPDATE: The only things that need to be checked are read/writepage
  12. which do not hold i_sem. Note writepage cannot change i_size but it
  13. needs to cope with a concurrent i_size change, just like readpage.
  14. Also both need to cope with concurrent changes to the other sizes,
  15. i.e. initialized/allocated/compressed size, as well.
  16. - Implement mft.c::sync_mft_mirror_umount(). We currently will just
  17. leave the volume dirty on umount if the final iput(vol->mft_ino)
  18. causes a write of any mirrored mft records due to the mft mirror
  19. inode having been discarded already. Whether this can actually ever
  20. happen is unclear however so it is worth waiting until someone hits
  21. the problem.
  22. - Enable the code for setting the NT4 compatibility flag when we start
  23. making NTFS 1.2 specific modifications.
  24. 2.1.24-WIP
  25. - Support journals ($LogFile) which have been modified by chkdsk. This
  26. means users can boot into Windows after we marked the volume dirty.
  27. The Windows boot will run chkdsk and then reboot. The user can then
  28. immediately boot into Linux rather than having to do a full Windows
  29. boot first before rebooting into Linux and we will recognize such a
  30. journal and empty it as it is clean by definition.
  31. - Support journals ($LogFile) with only one restart page as well as
  32. journals with two different restart pages. We sanity check both and
  33. either use the only sane one or the more recent one of the two in the
  34. case that both are valid.
  35. - Modify fs/ntfs/malloc.h::ntfs_malloc_nofs() to do the kmalloc() based
  36. allocations with __GFP_HIGHMEM, analogous to how the vmalloc() based
  37. allocations are done.
  38. - Add fs/ntfs/malloc.h::ntfs_malloc_nofs_nofail() which is analogous to
  39. ntfs_malloc_nofs() but it performs allocations with __GFP_NOFAIL and
  40. hence cannot fail.
  41. - Use ntfs_malloc_nofs_nofail() in the two critical regions in
  42. fs/ntfs/runlist.c::ntfs_runlists_merge(). This means we no longer
  43. need to panic() if the allocation fails as it now cannot fail.
  44. - Fix two nasty runlist merging bugs that had gone unnoticed so far.
  45. Thanks to Stefano Picerno for the bug report.
  46. - Remove two bogus BUG_ON()s from fs/ntfs/mft.c.
  47. 2.1.23 - Implement extension of resident files and make writing safe as well as
  48. many bug fixes, cleanups, and enhancements...
  49. - Add printk rate limiting for ntfs_warning() and ntfs_error() when
  50. compiled without debug. This avoids a possible denial of service
  51. attack. Thanks to Carl-Daniel Hailfinger from SuSE for pointing this
  52. out.
  53. - Fix compilation warnings on ia64. (Randy Dunlap)
  54. - Use i_size_{read,write}() instead of reading i_size by hand and cache
  55. the value where apropriate.
  56. - Add size_lock to the ntfs_inode structure. This is an rw spinlock
  57. and it locks against access to the inode sizes. Note, ->size_lock
  58. is also accessed from irq context so you must use the _irqsave and
  59. _irqrestore lock and unlock functions, respectively. Protect all
  60. accesses to allocated_size, initialized_size, and compressed_size.
  61. - Minor optimization to fs/ntfs/super.c::ntfs_statfs() and its helpers.
  62. - Implement extension of resident files in the regular file write code
  63. paths (fs/ntfs/aops.c::ntfs_{prepare,commit}_write()). At present
  64. this only works until the data attribute becomes too big for the mft
  65. record after which we abort the write returning -EOPNOTSUPP from
  66. ntfs_prepare_write().
  67. - Add disable_sparse mount option together with a per volume sparse
  68. enable bit which is set appropriately and a per inode sparse disable
  69. bit which is preset on some system file inodes as appropriate.
  70. - Enforce that sparse support is disabled on NTFS volumes pre 3.0.
  71. - Fix a bug in fs/ntfs/runlist.c::ntfs_mapping_pairs_decompress() in
  72. the creation of the unmapped runlist element for the base attribute
  73. extent.
  74. - Split ntfs_map_runlist() into ntfs_map_runlist() and a non-locking
  75. helper ntfs_map_runlist_nolock() which is used by ntfs_map_runlist().
  76. This allows us to map runlist fragments with the runlist lock already
  77. held without having to drop and reacquire it around the call. Adapt
  78. all callers.
  79. - Change ntfs_find_vcn() to ntfs_find_vcn_nolock() which takes a locked
  80. runlist. This allows us to find runlist elements with the runlist
  81. lock already held without having to drop and reacquire it around the
  82. call. Adapt all callers.
  83. - Change time to u64 in time.h::ntfs2utc() as it otherwise generates a
  84. warning in the do_div() call on sparc32. Thanks to Meelis Roos for
  85. the report and analysis of the warning.
  86. - Fix a nasty runlist merge bug when merging two holes.
  87. - Set the ntfs_inode->allocated_size to the real allocated size in the
  88. mft record for resident attributes (fs/ntfs/inode.c).
  89. - Small readability cleanup to use "a" instead of "ctx->attr"
  90. everywhere (fs/ntfs/inode.c).
  91. - Make fs/ntfs/namei.c::ntfs_get_{parent,dentry} static and move the
  92. definition of ntfs_export_ops from fs/ntfs/super.c to namei.c. Also,
  93. declare ntfs_export_ops in fs/ntfs/ntfs.h.
  94. - Correct sparse file handling. The compressed values need to be
  95. checked and set in the ntfs inode as done for compressed files and
  96. the compressed size needs to be used for vfs inode->i_blocks instead
  97. of the allocated size, again, as done for compressed files.
  98. - Add AT_EA in addition to AT_DATA to whitelist for being allowed to be
  99. non-resident in fs/ntfs/attrib.c::ntfs_attr_can_be_non_resident().
  100. - Add fs/ntfs/attrib.c::ntfs_attr_vcn_to_lcn_nolock() used by the new
  101. write code.
  102. - Fix bug in fs/ntfs/attrib.c::ntfs_find_vcn_nolock() where after
  103. dropping the read lock and taking the write lock we were not checking
  104. whether someone else did not already do the work we wanted to do.
  105. - Rename fs/ntfs/attrib.c::ntfs_find_vcn_nolock() to
  106. ntfs_attr_find_vcn_nolock() and update all callers.
  107. - Add fs/ntfs/attrib.[hc]::ntfs_attr_make_non_resident().
  108. - Fix sign of various error return values to be negative in
  109. fs/ntfs/lcnalloc.c.
  110. - Modify ->readpage and ->writepage (fs/ntfs/aops.c) so they detect and
  111. handle the case where an attribute is converted from resident to
  112. non-resident by a concurrent file write.
  113. - Remove checks for NULL before calling kfree() since kfree() does the
  114. checking itself. (Jesper Juhl)
  115. - Some utilities modify the boot sector but do not update the checksum.
  116. Thus, relax the checking in fs/ntfs/super.c::is_boot_sector_ntfs() to
  117. only emit a warning when the checksum is incorrect rather than
  118. refusing the mount. Thanks to Bernd Casimir for pointing this
  119. problem out.
  120. - Update attribute definition handling.
  121. - Add NTFS_MAX_CLUSTER_SIZE and NTFS_MAX_PAGES_PER_CLUSTER constants.
  122. - Use NTFS_MAX_CLUSTER_SIZE in super.c instead of hard coding 0x10000.
  123. - Use MAX_BUF_PER_PAGE instead of variable sized array allocation for
  124. better code generation and one less sparse warning in fs/ntfs/aops.c.
  125. - Remove spurious void pointer casts from fs/ntfs/. (Pekka Enberg)
  126. - Use C99 style structure initialization after memory allocation where
  127. possible (fs/ntfs/{attrib.c,index.c,super.c}). Thanks to Al Viro and
  128. Pekka Enberg.
  129. - Stamp the transaction log ($UsnJrnl), aka user space journal, if it
  130. is active on the volume and we are mounting read-write or remounting
  131. from read-only to read-write.
  132. - Fix a bug in address space operations error recovery code paths where
  133. if the runlist was not mapped at all and a mapping error occured we
  134. would leave the runlist locked on exit to the function so that the
  135. next access to the same file would try to take the lock and deadlock.
  136. - Detect the case when Windows has been suspended to disk on the volume
  137. to be mounted and if this is the case do not allow (re)mounting
  138. read-write. This is done by parsing hiberfil.sys if present.
  139. - Fix several occurences of a bug where we would perform 'var & ~const'
  140. with a 64-bit variable and a int, i.e. 32-bit, constant. This causes
  141. the higher order 32-bits of the 64-bit variable to be zeroed. To fix
  142. this cast the 'const' to the same 64-bit type as 'var'.
  143. - Change the runlist terminator of the newly allocated cluster(s) to
  144. LCN_ENOENT in ntfs_attr_make_non_resident(). Otherwise the runlist
  145. code gets confused.
  146. - Add an extra parameter @last_vcn to ntfs_get_size_for_mapping_pairs()
  147. and ntfs_mapping_pairs_build() to allow the runlist encoding to be
  148. partial which is desirable when filling holes in sparse attributes.
  149. Update all callers.
  150. - Change ntfs_map_runlist_nolock() to only decompress the mapping pairs
  151. if the requested vcn is inside it. Otherwise we get into problems
  152. when we try to map an out of bounds vcn because we then try to map
  153. the already mapped runlist fragment which causes
  154. ntfs_mapping_pairs_decompress() to fail and return error. Update
  155. ntfs_attr_find_vcn_nolock() accordingly.
  156. - Fix a nasty deadlock that appeared in recent kernels.
  157. The situation: VFS inode X on a mounted ntfs volume is dirty. For
  158. same inode X, the ntfs_inode is dirty and thus corresponding on-disk
  159. inode, i.e. mft record, which is in a dirty PAGE_CACHE_PAGE belonging
  160. to the table of inodes, i.e. $MFT, inode 0.
  161. What happens:
  162. Process 1: sys_sync()/umount()/whatever... calls
  163. __sync_single_inode() for $MFT -> do_writepages() -> write_page for
  164. the dirty page containing the on-disk inode X, the page is now locked
  165. -> ntfs_write_mst_block() which clears PageUptodate() on the page to
  166. prevent anyone else getting hold of it whilst it does the write out.
  167. This is necessary as the on-disk inode needs "fixups" applied before
  168. the write to disk which are removed again after the write and
  169. PageUptodate is then set again. It then analyses the page looking
  170. for dirty on-disk inodes and when it finds one it calls
  171. ntfs_may_write_mft_record() to see if it is safe to write this
  172. on-disk inode. This then calls ilookup5() to check if the
  173. corresponding VFS inode is in icache(). This in turn calls ifind()
  174. which waits on the inode lock via wait_on_inode whilst holding the
  175. global inode_lock.
  176. Process 2: pdflush results in a call to __sync_single_inode for the
  177. same VFS inode X on the ntfs volume. This locks the inode (I_LOCK)
  178. then calls write-inode -> ntfs_write_inode -> map_mft_record() ->
  179. read_cache_page() for the page (in page cache of table of inodes
  180. $MFT, inode 0) containing the on-disk inode. This page has
  181. PageUptodate() clear because of Process 1 (see above) so
  182. read_cache_page() blocks when it tries to take the page lock for the
  183. page so it can call ntfs_read_page().
  184. Thus Process 1 is holding the page lock on the page containing the
  185. on-disk inode X and it is waiting on the inode X to be unlocked in
  186. ifind() so it can write the page out and then unlock the page.
  187. And Process 2 is holding the inode lock on inode X and is waiting for
  188. the page to be unlocked so it can call ntfs_readpage() or discover
  189. that Process 1 set PageUptodate() again and use the page.
  190. Thus we have a deadlock due to ifind() waiting on the inode lock.
  191. The solution: The fix is to use the newly introduced
  192. ilookup5_nowait() which does not wait on the inode's lock and hence
  193. avoids the deadlock. This is safe as we do not care about the VFS
  194. inode and only use the fact that it is in the VFS inode cache and the
  195. fact that the vfs and ntfs inodes are one struct in memory to find
  196. the ntfs inode in memory if present. Also, the ntfs inode has its
  197. own locking so it does not matter if the vfs inode is locked.
  198. - Fix bug in mft record writing where we forgot to set the device in
  199. the buffers when mapping them after the VM had discarded them.
  200. Thanks to Martin MOKREJŠ for the bug report.
  201. 2.1.22 - Many bug and race fixes and error handling improvements.
  202. - Improve error handling in fs/ntfs/inode.c::ntfs_truncate().
  203. - Change fs/ntfs/inode.c::ntfs_truncate() to return an error code
  204. instead of void and provide a helper ntfs_truncate_vfs() for the
  205. vfs ->truncate method.
  206. - Add a new ntfs inode flag NInoTruncateFailed() and modify
  207. fs/ntfs/inode.c::ntfs_truncate() to set and clear it appropriately.
  208. - Fix min_size and max_size definitions in ATTR_DEF structure in
  209. fs/ntfs/layout.h to be signed.
  210. - Add attribute definition handling helpers to fs/ntfs/attrib.[hc]:
  211. ntfs_attr_size_bounds_check(), ntfs_attr_can_be_non_resident(), and
  212. ntfs_attr_can_be_resident(), which in turn use the new private helper
  213. ntfs_attr_find_in_attrdef().
  214. - In fs/ntfs/aops.c::mark_ntfs_record_dirty(), take the
  215. mapping->private_lock around the dirtying of the buffer heads
  216. analagous to the way it is done in __set_page_dirty_buffers().
  217. - Ensure the mft record size does not exceed the PAGE_CACHE_SIZE at
  218. mount time as this cannot work with the current implementation.
  219. - Check for location of attribute name and improve error handling in
  220. general in fs/ntfs/inode.c::ntfs_read_locked_inode() and friends.
  221. - In fs/ntfs/aops.c::ntfs_writepage(), if the page is fully outside
  222. i_size, i.e. race with truncate, invalidate the buffers on the page
  223. so that they become freeable and hence the page does not leak.
  224. - Remove unused function fs/ntfs/runlist.c::ntfs_rl_merge(). (Adrian
  225. Bunk)
  226. - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_find() that resulted in
  227. a NULL pointer dereference in the error code path when a corrupt
  228. attribute was found. (Thanks to Domen Puncer for the bug report.)
  229. - Add MODULE_VERSION() to fs/ntfs/super.c.
  230. - Make several functions and variables static. (Adrian Bunk)
  231. - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() so it allocates
  232. buffers for the page if they are not present and then marks the
  233. buffers belonging to the ntfs record dirty. This causes the buffers
  234. to become busy and hence they are safe from removal until the page
  235. has been written out.
  236. - Fix stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find() in the
  237. error handling code path that resulted in a BUG() due to trying to
  238. unmap an extent mft record when the mapping of it had failed and it
  239. thus was not mapped. (Thanks to Ken MacFerrin for the bug report.)
  240. - Drop the runlist lock after the vcn has been read in
  241. fs/ntfs/lcnalloc.c::__ntfs_cluster_free().
  242. - Rewrite handling of multi sector transfer errors. We now do not set
  243. PageError() when such errors are detected in the async i/o handler
  244. fs/ntfs/aops.c::ntfs_end_buffer_async_read(). All users of mst
  245. protected attributes now check the magic of each ntfs record as they
  246. use it and act appropriately. This has the effect of making errors
  247. granular per ntfs record rather than per page which solves the case
  248. where we cannot access any of the ntfs records in a page when a
  249. single one of them had an mst error. (Thanks to Ken MacFerrin for
  250. the bug report.)
  251. - Fix error handling in fs/ntfs/quota.c::ntfs_mark_quotas_out_of_date()
  252. where we failed to release i_sem on the $Quota/$Q attribute inode.
  253. - Fix bug in handling of bad inodes in fs/ntfs/namei.c::ntfs_lookup().
  254. - Add mapping of unmapped buffers to all remaining code paths, i.e.
  255. fs/ntfs/aops.c::ntfs_write_mst_block(), mft.c::ntfs_sync_mft_mirror(),
  256. and write_mft_record_nolock(). From now on we require that the
  257. complete runlist for the mft mirror is always mapped into memory.
  258. - Add creation of buffers to fs/ntfs/mft.c::ntfs_sync_mft_mirror().
  259. - Improve error handling in fs/ntfs/aops.c::ntfs_{read,write}_block().
  260. - Cleanup fs/ntfs/aops.c::ntfs_{read,write}page() since we know that a
  261. resident attribute will be smaller than a page which makes the code
  262. simpler. Also make the code more tolerant to concurrent ->truncate.
  263. 2.1.21 - Fix some races and bugs, rewrite mft write code, add mft allocator.
  264. - Implement extent mft record deallocation
  265. fs/ntfs/mft.c::ntfs_extent_mft_record_free().
  266. - Splitt runlist related functions off from attrib.[hc] to runlist.[hc].
  267. - Add vol->mft_data_pos and initialize it at mount time.
  268. - Rename init_runlist() to ntfs_init_runlist(), ntfs_vcn_to_lcn() to
  269. ntfs_rl_vcn_to_lcn(), decompress_mapping_pairs() to
  270. ntfs_mapping_pairs_decompress(), ntfs_merge_runlists() to
  271. ntfs_runlists_merge() and adapt all callers.
  272. - Add fs/ntfs/runlist.[hc]::ntfs_get_nr_significant_bytes(),
  273. ntfs_get_size_for_mapping_pairs(), ntfs_write_significant_bytes(),
  274. and ntfs_mapping_pairs_build(), adapted from libntfs.
  275. - Make fs/ntfs/lcnalloc.c::ntfs_cluster_free_from_rl_nolock() not
  276. static and add a declaration for it to lcnalloc.h.
  277. - Add fs/ntfs/lcnalloc.h::ntfs_cluster_free_from_rl() which is a static
  278. inline wrapper for ntfs_cluster_free_from_rl_nolock() which takes the
  279. cluster bitmap lock for the duration of the call.
  280. - Add fs/ntfs/attrib.[hc]::ntfs_attr_record_resize().
  281. - Implement the equivalent of memset() for an ntfs attribute in
  282. fs/ntfs/attrib.[hc]::ntfs_attr_set() and switch
  283. fs/ntfs/logfile.c::ntfs_empty_logfile() to using it.
  284. - Remove unnecessary casts from LCN_* constants.
  285. - Implement fs/ntfs/runlist.c::ntfs_rl_truncate_nolock().
  286. - Add MFT_RECORD_OLD as a copy of MFT_RECORD in fs/ntfs/layout.h and
  287. change MFT_RECORD to contain the NTFS 3.1+ specific fields.
  288. - Add a helper function fs/ntfs/aops.c::mark_ntfs_record_dirty() which
  289. marks all buffers belonging to an ntfs record dirty, followed by
  290. marking the page the ntfs record is in dirty and also marking the vfs
  291. inode containing the ntfs record dirty (I_DIRTY_PAGES).
  292. - Switch fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to using the
  293. new helper fs/ntfs/aops.c::mark_ntfs_record_dirty() and remove the no
  294. longer needed fs/ntfs/index.[hc]::__ntfs_index_entry_mark_dirty().
  295. - Move ntfs_{un,}map_page() from ntfs.h to aops.h and fix resulting
  296. include errors.
  297. - Move the typedefs for runlist_element and runlist from types.h to
  298. runlist.h and fix resulting include errors.
  299. - Remove unused {__,}format_mft_record() from fs/ntfs/mft.c.
  300. - Modify fs/ntfs/mft.c::__mark_mft_record_dirty() to use the helper
  301. mark_ntfs_record_dirty() which also changes the behaviour in that we
  302. now set the buffers belonging to the mft record dirty as well as the
  303. page itself.
  304. - Update fs/ntfs/mft.c::write_mft_record_nolock() and sync_mft_mirror()
  305. to cope with the fact that there now are dirty buffers in mft pages.
  306. - Update fs/ntfs/inode.c::ntfs_write_inode() to also use the helper
  307. mark_ntfs_record_dirty() and thus to set the buffers belonging to the
  308. mft record dirty as well as the page itself.
  309. - Fix compiler warnings on x86-64 in fs/ntfs/dir.c. (Randy Dunlap,
  310. slightly modified by me)
  311. - Add fs/ntfs/mft.c::try_map_mft_record() which fails with -EALREADY if
  312. the mft record is already locked and otherwise behaves the same way
  313. as fs/ntfs/mft.c::map_mft_record().
  314. - Modify fs/ntfs/mft.c::write_mft_record_nolock() so that it only
  315. writes the mft record if the buffers belonging to it are dirty.
  316. Otherwise we assume that it was written out by other means already.
  317. - Attempting to write outside initialized size is _not_ a bug so remove
  318. the bug check from fs/ntfs/aops.c::ntfs_write_mst_block(). It is in
  319. fact required to write outside initialized size when preparing to
  320. extend the initialized size.
  321. - Map the page instead of using page_address() before writing to it in
  322. fs/ntfs/aops.c::ntfs_mft_writepage().
  323. - Provide exclusion between opening an inode / mapping an mft record
  324. and accessing the mft record in fs/ntfs/mft.c::ntfs_mft_writepage()
  325. by setting the page not uptodate throughout ntfs_mft_writepage().
  326. - Clear the page uptodate flag in fs/ntfs/aops.c::ntfs_write_mst_block()
  327. to ensure noone can see the page whilst the mst fixups are applied.
  328. - Add the helper fs/ntfs/mft.c::ntfs_may_write_mft_record() which
  329. checks if an mft record may be written out safely obtaining any
  330. necessary locks in the process. This is used by
  331. fs/ntfs/aops.c::ntfs_write_mst_block().
  332. - Modify fs/ntfs/aops.c::ntfs_write_mst_block() to also work for
  333. writing mft records and improve its error handling in the process.
  334. Now if any of the records in the page fail to be written out, all
  335. other records will be written out instead of aborting completely.
  336. - Remove ntfs_mft_aops and update all users to use ntfs_mst_aops.
  337. - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to set the
  338. ntfs_mst_aops for all inodes which are NInoMstProtected() and
  339. ntfs_aops for all other inodes.
  340. - Rename fs/ntfs/mft.c::sync_mft_mirror{,_umount}() to
  341. ntfs_sync_mft_mirror{,_umount}() and change their parameters so they
  342. no longer require an ntfs inode to be present. Update all callers.
  343. - Cleanup the error handling in fs/ntfs/mft.c::ntfs_sync_mft_mirror().
  344. - Clear the page uptodate flag in fs/ntfs/mft.c::ntfs_sync_mft_mirror()
  345. to ensure noone can see the page whilst the mst fixups are applied.
  346. - Remove the no longer needed fs/ntfs/mft.c::ntfs_mft_writepage() and
  347. fs/ntfs/mft.c::try_map_mft_record().
  348. - Fix callers of fs/ntfs/aops.c::mark_ntfs_record_dirty() to call it
  349. with the ntfs inode which contains the page rather than the ntfs
  350. inode the mft record of which is in the page.
  351. - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by moving the
  352. index inode bitmap inode release code from there to
  353. fs/ntfs/inode.c::ntfs_clear_big_inode(). (Thanks to Christoph
  354. Hellwig for spotting this.)
  355. - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by taking the
  356. inode semaphore around the code that sets ni->itype.index.bmp_ino to
  357. NULL and reorganize the code to optimize it a bit. (Thanks to
  358. Christoph Hellwig for spotting this.)
  359. - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() to no longer take the
  360. ntfs inode as a parameter as this is confusing and misleading and the
  361. needed ntfs inode is available via NTFS_I(page->mapping->host).
  362. Adapt all callers to this change.
  363. - Modify fs/ntfs/mft.c::write_mft_record_nolock() and
  364. fs/ntfs/aops.c::ntfs_write_mst_block() to only check the dirty state
  365. of the first buffer in a record and to take this as the ntfs record
  366. dirty state. We cannot look at the dirty state for subsequent
  367. buffers because we might be racing with
  368. fs/ntfs/aops.c::mark_ntfs_record_dirty().
  369. - Move the static inline ntfs_init_big_inode() from fs/ntfs/inode.c to
  370. inode.h and make fs/ntfs/inode.c::__ntfs_init_inode() non-static and
  371. add a declaration for it to inode.h. Fix some compilation issues
  372. that resulted due to #includes and header file interdependencies.
  373. - Simplify setup of i_mode in fs/ntfs/inode.c::ntfs_read_locked_inode().
  374. - Add helpers fs/ntfs/layout.h::MK_MREF() and MK_LE_MREF().
  375. - Modify fs/ntfs/mft.c::map_extent_mft_record() to only verify the mft
  376. record sequence number if it is specified (i.e. not zero).
  377. - Add fs/ntfs/mft.[hc]::ntfs_mft_record_alloc() and various helper
  378. functions used by it.
  379. - Update Documentation/filesystems/ntfs.txt with instructions on how to
  380. use the Device-Mapper driver with NTFS ftdisk/LDM raid. This removes
  381. the linear raid problem with the Software RAID / MD driver when one
  382. or more of the devices has an odd number of sectors.
  383. 2.1.20 - Fix two stupid bugs introduced in 2.1.18 release.
  384. - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_reinit_search_ctx()
  385. where we did not clear ctx->al_entry but it was still set due to
  386. changes in ntfs_attr_lookup() and ntfs_external_attr_find() in
  387. particular.
  388. - Fix another stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find()
  389. where we forgot to unmap the extent mft record when we had finished
  390. enumerating an attribute which caused a bug check to trigger when the
  391. VFS calls ->clear_inode.
  392. 2.1.19 - Many cleanups, improvements, and a minor bug fix.
  393. - Update ->setattr (fs/ntfs/inode.c::ntfs_setattr()) to refuse to
  394. change the uid, gid, and mode of an inode as we do not support NTFS
  395. ACLs yet.
  396. - Remove BKL use from ntfs_setattr() syncing up with the rest of the
  397. kernel.
  398. - Get rid of the ugly transparent union in fs/ntfs/dir.c::ntfs_readdir()
  399. and ntfs_filldir() as per suggestion from Al Viro.
  400. - Change '\0' and L'\0' to simply 0 as per advice from Linus Torvalds.
  401. - Update ->truncate (fs/ntfs/inode.c::ntfs_truncate()) to check if the
  402. inode size has changed and to only output an error if so.
  403. - Rename fs/ntfs/attrib.h::attribute_value_length() to ntfs_attr_size().
  404. - Add le{16,32,64} as well as sle{16,32,64} data types to
  405. fs/ntfs/types.h.
  406. - Change ntfschar to be le16 instead of u16 in fs/ntfs/types.h.
  407. - Add le versions of VCN, LCN, and LSN called leVCN, leLCN, and leLSN,
  408. respectively, to fs/ntfs/types.h.
  409. - Update endianness conversion macros in fs/ntfs/endian.h to use the
  410. new types as appropriate.
  411. - Do proper type casting when using sle64_to_cpup() in fs/ntfs/dir.c
  412. and index.c.
  413. - Add leMFT_REF data type to fs/ntfs/layout.h.
  414. - Update all NTFS header files with the new little endian data types.
  415. Affected files are fs/ntfs/layout.h, logfile.h, and time.h.
  416. - Do proper type casting when using ntfs_is_*_recordp() in
  417. fs/ntfs/logfile.c, mft.c, and super.c.
  418. - Fix all the sparse bitwise warnings. Had to change all the typedef
  419. enums storing little endian values to simple enums plus a typedef for
  420. the datatype to make sparse happy.
  421. - Fix a bug found by the new sparse bitwise warnings where the default
  422. upcase table was defined as a pointer to wchar_t rather than ntfschar
  423. in fs/ntfs/ntfs.h and super.c.
  424. - Change {const_,}cpu_to_le{16,32}(0) to just 0 as suggested by Al Viro.
  425. 2.1.18 - Fix scheduling latencies at mount time as well as an endianness bug.
  426. - Remove vol->nr_mft_records as it was pretty meaningless and optimize
  427. the calculation of total/free inodes as used by statfs().
  428. - Fix scheduling latencies in ntfs_fill_super() by dropping the BKL
  429. because the code itself is using the ntfs_lock semaphore which
  430. provides safe locking. (Ingo Molnar)
  431. - Fix a potential bug in fs/ntfs/mft.c::map_extent_mft_record() that
  432. could occur in the future for when we start closing/freeing extent
  433. inodes if we don't set base_ni->ext.extent_ntfs_inos to NULL after
  434. we free it.
  435. - Rename {find,lookup}_attr() to ntfs_attr_{find,lookup}() as well as
  436. find_external_attr() to ntfs_external_attr_find() to cleanup the
  437. namespace a bit and to be more consistent with libntfs.
  438. - Rename {{re,}init,get,put}_attr_search_ctx() to
  439. ntfs_attr_{{re,}init,get,put}_search_ctx() as well as the type
  440. attr_search_context to ntfs_attr_search_ctx.
  441. - Force use of ntfs_attr_find() in ntfs_attr_lookup() when searching
  442. for the attribute list attribute itself.
  443. - Fix endianness bug in ntfs_external_attr_find().
  444. - Change ntfs_{external_,}attr_find() to return 0 on success, -ENOENT
  445. if the attribute is not found, and -EIO on real error. In the case
  446. of -ENOENT, the search context is updated to describe the attribute
  447. before which the attribute being searched for would need to be
  448. inserted if such an action were to be desired and in the case of
  449. ntfs_external_attr_find() the search context is also updated to
  450. indicate the attribute list entry before which the attribute list
  451. entry of the attribute being searched for would need to be inserted
  452. if such an action were to be desired. Also make ntfs_find_attr()
  453. static and remove its prototype from attrib.h as it is not used
  454. anywhere other than attrib.c. Update ntfs_attr_lookup() and all
  455. callers of ntfs_{external,}attr_{find,lookup}() for the new return
  456. values.
  457. - Minor cleanup of fs/ntfs/inode.c::ntfs_init_locked_inode().
  458. 2.1.17 - Fix bugs in mount time error code paths and other updates.
  459. - Implement bitmap modification code (fs/ntfs/bitmap.[hc]). This
  460. includes functions to set/clear a single bit or a run of bits.
  461. - Add fs/ntfs/attrib.[hc]::ntfs_find_vcn() which returns the locked
  462. runlist element containing a particular vcn. It also takes care of
  463. mapping any needed runlist fragments.
  464. - Implement cluster (de-)allocation code (fs/ntfs/lcnalloc.[hc]).
  465. - Load attribute definition table from $AttrDef at mount time.
  466. - Fix bugs in mount time error code paths involving (de)allocation of
  467. the default and volume upcase tables.
  468. - Remove ntfs_nr_mounts as it is no longer used.
  469. 2.1.16 - Implement access time updates, file sync, async io, and read/writev.
  470. - Add support for readv/writev and aio_read/aio_write (fs/ntfs/file.c).
  471. This is done by setting the appropriate file operations pointers to
  472. the generic helper functions provided by mm/filemap.c.
  473. - Implement fsync, fdatasync, and msync both for files (fs/ntfs/file.c)
  474. and directories (fs/ntfs/dir.c).
  475. - Add support for {a,m,c}time updates to inode.c::ntfs_write_inode().
  476. Note, except for the root directory and any other system files opened
  477. by the user, the system files will not have their access times
  478. updated as they are only accessed at the inode level an hence the
  479. file level functions which cause the times to be updated are never
  480. invoked.
  481. 2.1.15 - Invalidate quotas when (re)mounting read-write.
  482. - Add new element itype.index.collation_rule to the ntfs inode
  483. structure and set it appropriately in ntfs_read_locked_inode().
  484. - Implement a new inode type "index" to allow efficient access to the
  485. indices found in various system files and adapt inode handling
  486. accordingly (fs/ntfs/inode.[hc]). An index inode is essentially an
  487. attribute inode (NInoAttr() is true) with an attribute type of
  488. AT_INDEX_ALLOCATION. As such, it is no longer allowed to call
  489. ntfs_attr_iget() with an attribute type of AT_INDEX_ALLOCATION as
  490. there would be no way to distinguish between normal attribute inodes
  491. and index inodes. The function to obtain an index inode is
  492. ntfs_index_iget() and it uses the helper function
  493. ntfs_read_locked_index_inode(). Note, we do not overload
  494. ntfs_attr_iget() as indices consist of multiple attributes so using
  495. ntfs_attr_iget() to obtain an index inode would be confusing.
  496. - Ensure that there is no overflow when doing page->index <<
  497. PAGE_CACHE_SHIFT by casting page->index to s64 in fs/ntfs/aops.c.
  498. - Use atomic kmap instead of kmap() in fs/ntfs/aops.c::ntfs_read_page()
  499. and ntfs_read_block().
  500. - Use case sensitive attribute lookups instead of case insensitive ones.
  501. - Lock all page cache pages belonging to mst protected attributes while
  502. accessing them to ensure we never see corrupt data while the page is
  503. under writeout.
  504. - Add framework for generic ntfs collation (fs/ntfs/collation.[hc]).
  505. We have ntfs_is_collation_rule_supported() to check if the collation
  506. rule you want to use is supported and ntfs_collation() which actually
  507. collates two data items. We currently only support COLLATION_BINARY
  508. and COLLATION_NTOFS_ULONG but support for other collation rules will
  509. be added as the need arises.
  510. - Add a new type, ntfs_index_context, to allow retrieval of an index
  511. entry using the corresponding index key. To get an index context,
  512. use ntfs_index_ctx_get() and to release it, use ntfs_index_ctx_put().
  513. This also adds a new slab cache for the index contexts. To lookup a
  514. key in an index inode, use ntfs_index_lookup(). After modifying an
  515. index entry, call ntfs_index_entry_flush_dcache_page() followed by
  516. ntfs_index_entry_mark_dirty() to ensure the changes are written out
  517. to disk. For details see fs/ntfs/index.[hc]. Note, at present, if
  518. an index entry is in the index allocation attribute rather than the
  519. index root attribute it will not be written out (you will get a
  520. warning message about discarded changes instead).
  521. - Load the quota file ($Quota) and check if quota tracking is enabled
  522. and if so, mark the quotas out of date. This causes windows to
  523. rescan the volume on boot and update all quota entries.
  524. - Add a set_page_dirty address space operation for ntfs_m[fs]t_aops.
  525. It is simply set to __set_page_dirty_nobuffers() to make sure that
  526. running set_page_dirty() on a page containing mft/ntfs records will
  527. not affect the dirty state of the page buffers.
  528. - Add fs/ntfs/index.c::__ntfs_index_entry_mark_dirty() which sets all
  529. buffers that are inside the ntfs record in the page dirty after which
  530. it sets the page dirty. This allows ->writepage to only write the
  531. dirty index records rather than having to write all the records in
  532. the page. Modify fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to
  533. use this rather than __set_page_dirty_nobuffers().
  534. - Implement fs/ntfs/aops.c::ntfs_write_mst_block() which enables the
  535. writing of page cache pages belonging to mst protected attributes
  536. like the index allocation attribute in directory indices and other
  537. indices like $Quota/$Q, etc. This means that the quota is now marked
  538. out of date on all volumes rather than only on ones where the quota
  539. defaults entry is in the index root attribute of the $Quota/$Q index.
  540. 2.1.14 - Fix an NFSd caused deadlock reported by several users.
  541. - Modify fs/ntfs/ntfs_readdir() to copy the index root attribute value
  542. to a buffer so that we can put the search context and unmap the mft
  543. record before calling the filldir() callback. We need to do this
  544. because of NFSd which calls ->lookup() from its filldir callback()
  545. and this causes NTFS to deadlock as ntfs_lookup() maps the mft record
  546. of the directory and since ntfs_readdir() has got it mapped already
  547. ntfs_lookup() deadlocks.
  548. 2.1.13 - Enable overwriting of resident files and housekeeping of system files.
  549. - Implement writing of mft records (fs/ntfs/mft.[hc]), which includes
  550. keeping the mft mirror in sync with the mft when mirrored mft records
  551. are written. The functions are write_mft_record{,_nolock}(). The
  552. implementation is quite rudimentary for now with lots of things not
  553. implemented yet but I am not sure any of them can actually occur so
  554. I will wait for people to hit each one and only then implement it.
  555. - Commit open system inodes at umount time. This should make it
  556. virtually impossible for sync_mft_mirror_umount() to ever be needed.
  557. - Implement ->write_inode (fs/ntfs/inode.c::ntfs_write_inode()) for the
  558. ntfs super operations. This gives us inode writing via the VFS inode
  559. dirty code paths. Note: Access time updates are not implemented yet.
  560. - Implement fs/ntfs/mft.[hc]::{,__}mark_mft_record_dirty() and make
  561. fs/ntfs/aops.c::ntfs_writepage() and ntfs_commit_write() use it, thus
  562. finally enabling resident file overwrite! (-8 This also includes a
  563. placeholder for ->writepage (ntfs_mft_writepage()), which for now
  564. just redirties the page and returns. Also, at umount time, we for
  565. now throw away all mft data page cache pages after the last call to
  566. ntfs_commit_inode() in the hope that all inodes will have been
  567. written out by then and hence no dirty (meta)data will be lost. We
  568. also check for this case and emit an error message telling the user
  569. to run chkdsk.
  570. - Use set_page_writeback() and end_page_writeback() in the resident
  571. attribute code path of fs/ntfs/aops.c::ntfs_writepage() otherwise
  572. the radix-tree tag PAGECACHE_TAG_DIRTY remains set even though the
  573. page is clean.
  574. - Implement ntfs_mft_writepage() so it now checks if any of the mft
  575. records in the page are dirty and if so redirties the page and
  576. returns. Otherwise it just returns (after doing set_page_writeback(),
  577. unlock_page(), end_page_writeback() or the radix-tree tag
  578. PAGECACHE_TAG_DIRTY remains set even though the page is clean), thus
  579. alowing the VM to do with the page as it pleases. Also, at umount
  580. time, now only throw away dirty mft (meta)data pages if dirty inodes
  581. are present and ask the user to email us if they see this happening.
  582. - Add functions ntfs_{clear,set}_volume_flags(), to modify the volume
  583. information flags (fs/ntfs/super.c).
  584. - Mark the volume dirty when (re)mounting read-write and mark it clean
  585. when unmounting or remounting read-only. If any volume errors are
  586. found, the volume is left marked dirty to force chkdsk to run.
  587. - Add code to set the NT4 compatibility flag when (re)mounting
  588. read-write for newer NTFS versions but leave it commented out for now
  589. since we do not make any modifications that are NTFS 1.2 specific yet
  590. and since setting this flag breaks Captive-NTFS which is not nice.
  591. This code must be enabled once we start writing NTFS 1.2 specific
  592. changes otherwise Windows NTFS driver might crash / cause corruption.
  593. 2.1.12 - Fix the second fix to the decompression engine and some cleanups.
  594. - Add a new address space operations struct, ntfs_mst_aops, for mst
  595. protected attributes. This is because the default ntfs_aops do not
  596. make sense with mst protected data and were they to write anything to
  597. such an attribute they would cause data corruption so we provide
  598. ntfs_mst_aops which does not have any write related operations set.
  599. - Cleanup dirty ntfs inode handling (fs/ntfs/inode.[hc]) which also
  600. includes an adapted ntfs_commit_inode() and an implementation of
  601. ntfs_write_inode() which for now just cleans dirty inodes without
  602. writing them (it does emit a warning that this is happening).
  603. - Undo the second decompression engine fix (see 2.1.9 release ChangeLog
  604. entry) as it was only fixing a theoretical bug but at the same time
  605. it badly broke the handling of sparse and uncompressed compression
  606. blocks.
  607. 2.1.11 - Driver internal cleanups.
  608. - Only build logfile.o if building the driver with read-write support.
  609. - Really final white space cleanups.
  610. - Use generic_ffs() instead of ffs() in logfile.c which allows the
  611. log_page_size variable to be optimized by gcc into a constant.
  612. - Rename uchar_t to ntfschar everywhere as uchar_t is unsigned 1-byte
  613. char as defined by POSIX and as found on some systems.
  614. 2.1.10 - Force read-only (re)mounting of volumes with unsupported volume flags.
  615. - Finish off the white space cleanups (remove trailing spaces, etc).
  616. - Clean up ntfs_fill_super() and ntfs_read_inode_mount() by removing
  617. the kludges around the first iget(). Instead of (re)setting ->s_op
  618. we have the $MFT inode set up by explicit new_inode() / set ->i_ino /
  619. insert_inode_hash() / call ntfs_read_inode_mount() directly. This
  620. kills the need for second super_operations and allows to return error
  621. from ntfs_read_inode_mount() without resorting to ugly "poisoning"
  622. tricks. (Al Viro)
  623. - Force read-only (re)mounting if any of the following bits are set in
  624. the volume information flags:
  625. VOLUME_IS_DIRTY, VOLUME_RESIZE_LOG_FILE,
  626. VOLUME_UPGRADE_ON_MOUNT, VOLUME_DELETE_USN_UNDERWAY,
  627. VOLUME_REPAIR_OBJECT_ID, VOLUME_MODIFIED_BY_CHKDSK
  628. To make this easier we define VOLUME_MUST_MOUNT_RO_MASK with all the
  629. above bits set so the test is made easy.
  630. 2.1.9 - Fix two bugs in decompression engine.
  631. - Fix a bug where we would not always detect that we have reached the
  632. end of a compression block because we were ending at minus one byte
  633. which is effectively the same as being at the end. The fix is to
  634. check whether the uncompressed buffer has been fully filled and if so
  635. we assume we have reached the end of the compression block. A big
  636. thank you to Marcin Gibuła for the bug report, the assistance in
  637. tracking down the bug and testing the fix.
  638. - Fix a possible bug where when a compressed read is truncated to the
  639. end of the file, the offset inside the last page was not truncated.
  640. 2.1.8 - Handle $MFT mirror and $LogFile, improve time handling, and cleanups.
  641. - Use get_bh() instead of manual atomic_inc() in fs/ntfs/compress.c.
  642. - Modify fs/ntfs/time.c::ntfs2utc(), get_current_ntfs_time(), and
  643. utc2ntfs() to work with struct timespec instead of time_t on the
  644. Linux UTC time side thus preserving the full precision of the NTFS
  645. time and only loosing up to 99 nano-seconds in the Linux UTC time.
  646. - Move fs/ntfs/time.c to fs/ntfs/time.h and make the time functions
  647. static inline.
  648. - Remove unused ntfs_dirty_inode().
  649. - Cleanup super operations declaration in fs/ntfs/super.c.
  650. - Wrap flush_dcache_mft_record_page() in #ifdef NTFS_RW.
  651. - Add NInoTestSetFoo() and NInoTestClearFoo() macro magic to
  652. fs/ntfs/inode.h and use it to declare NInoTest{Set,Clear}Dirty.
  653. - Move typedefs for ntfs_attr and test_t from fs/ntfs/inode.c to
  654. fs/ntfs/inode.h so they can be used elsewhere.
  655. - Determine the mft mirror size as the number of mirrored mft records
  656. and store it in ntfs_volume->mftmirr_size (fs/ntfs/super.c).
  657. - Load the mft mirror at mount time and compare the mft records stored
  658. in it to the ones in the mft. Force a read-only mount if the two do
  659. not match (fs/ntfs/super.c).
  660. - Fix type casting related warnings on 64-bit architectures. Thanks
  661. to Meelis Roos for reporting them.
  662. - Move %L to %ll as %L is floating point and %ll is integer which is
  663. what we want.
  664. - Read the journal ($LogFile) and determine if the volume has been
  665. shutdown cleanly and force a read-only mount if not (fs/ntfs/super.c
  666. and fs/ntfs/logfile.c). This is a little bit of a crude check in
  667. that we only look at the restart areas and not at the actual log
  668. records so that there will be a very small number of cases where we
  669. think that a volume is dirty when in fact it is clean. This should
  670. only affect volumes that have not been shutdown cleanly and did not
  671. have any pending, non-check-pointed i/o.
  672. - If the $LogFile indicates a clean shutdown and a read-write (re)mount
  673. is requested, empty $LogFile by overwriting it with 0xff bytes to
  674. ensure that Windows cannot cause data corruption by replaying a stale
  675. journal after Linux has written to the volume.
  676. 2.1.7 - Enable NFS exporting of mounted NTFS volumes.
  677. - Set i_generation in the VFS inode from the seq_no of the NTFS inode.
  678. - Make ntfs_lookup() NFS export safe, i.e. use d_splice_alias(), etc.
  679. - Implement ->get_dentry() in fs/ntfs/namei.c::ntfs_get_dentry() as the
  680. default doesn't allow inode number 0 which is a valid inode on NTFS
  681. and even if it did allow that it uses iget() instead of ntfs_iget()
  682. which makes it useless for us.
  683. - Implement ->get_parent() in fs/ntfs/namei.c::ntfs_get_parent() as the
  684. default just returns -EACCES which is not very useful.
  685. - Define export operations (->s_export_op) for NTFS (ntfs_export_ops)
  686. and set them up in the super block at mount time (super.c) this
  687. allows mounted NTFS volumes to be exported via NFS.
  688. - Add missing return -EOPNOTSUPP; in
  689. fs/ntfs/aops.c::ntfs_commit_nonresident_write().
  690. - Enforce no atime and no dir atime updates at mount/remount time as
  691. they are not implemented yet anyway.
  692. - Move a few assignments in fs/ntfs/attrib.c::load_attribute_list() to
  693. after a NULL check. Thanks to Dave Jones for pointing this out.
  694. 2.1.6 - Fix minor bug in handling of compressed directories.
  695. - Fix bug in handling of compressed directories. A compressed
  696. directory is not really compressed so when we set the ->i_blocks
  697. field of a compressed directory inode we were setting it from the
  698. non-existing field ni->itype.compressed.size which gave random
  699. results... For directories we now always use ni->allocated_size.
  700. 2.1.5 - Fix minor bug in attribute list attribute handling.
  701. - Fix bug in attribute list handling. Actually it is not as much a bug
  702. as too much protection in that we were not allowing attribute lists
  703. which waste space on disk while Windows XP clearly allows it and in
  704. fact creates such attribute lists so our driver was failing.
  705. - Update NTFS documentation ready for 2.6 kernel release.
  706. 2.1.4 - Reduce compiler requirements.
  707. - Remove all uses of unnamed structs and unions in the driver to make
  708. old and newer gcc versions happy. Makes it a bit uglier IMO but at
  709. least people will stop hassling me about it.
  710. 2.1.3 - Important bug fixes in corner cases.
  711. - super.c::parse_ntfs_boot_sector(): Correct the check for 64-bit
  712. clusters. (Philipp Thomas)
  713. - attrib.c::load_attribute_list(): Fix bug when initialized_size is a
  714. multiple of the block_size but not the cluster size. (Szabolcs
  715. Szakacsits <szaka@sienet.hu>)
  716. 2.1.2 - Important bug fixes aleviating the hangs in statfs.
  717. - Fix buggy free cluster and free inode determination logic.
  718. 2.1.1 - Minor updates.
  719. - Add handling for initialized_size != data_size in compressed files.
  720. - Reduce function local stack usage from 0x3d4 bytes to just noise in
  721. fs/ntfs/upcase.c. (Randy Dunlap <rddunlap@osdl.ord>)
  722. - Remove compiler warnings for newer gcc.
  723. - Pages are no longer kmapped by mm/filemap.c::generic_file_write()
  724. around calls to ->{prepare,commit}_write. Adapt NTFS appropriately
  725. in fs/ntfs/aops.c::ntfs_prepare_nonresident_write() by using
  726. kmap_atomic(KM_USER0).
  727. 2.1.0 - First steps towards write support: implement file overwrite.
  728. - Add configuration option for developmental write support with an
  729. appropriately scary configuration help text.
  730. - Initial implementation of fs/ntfs/aops.c::ntfs_writepage() and its
  731. helper fs/ntfs/aops.c::ntfs_write_block(). This enables mmap(2) based
  732. overwriting of existing files on ntfs. Note: Resident files are
  733. only written into memory, and not written out to disk at present, so
  734. avoid writing to files smaller than about 1kiB.
  735. - Initial implementation of fs/ntfs/aops.c::ntfs_prepare_write(), its
  736. helper fs/ntfs/aops.c::ntfs_prepare_nonresident_write() and their
  737. counterparts, fs/ntfs/aops.c::ntfs_commit_write(), and
  738. fs/ntfs/aops.c::ntfs_commit_nonresident_write(), respectively. Also,
  739. add generic_file_write() to the ntfs file operations (fs/ntfs/file.c).
  740. This enables write(2) based overwriting of existing files on ntfs.
  741. Note: As with mmap(2) based overwriting, resident files are only
  742. written into memory, and not written out to disk at present, so avoid
  743. writing to files smaller than about 1kiB.
  744. - Implement ->truncate (fs/ntfs/inode.c::ntfs_truncate()) and
  745. ->setattr() (fs/ntfs/inode.c::ntfs_setattr()) inode operations for
  746. files with the purpose of intercepting and aborting all i_size
  747. changes which we do not support yet. ntfs_truncate() actually only
  748. emits a warning message but AFAICS our interception of i_size changes
  749. elsewhere means ntfs_truncate() never gets called for i_size changes.
  750. It is only called from generic_file_write() when we fail in
  751. ntfs_prepare_{,nonresident_}write() in order to discard any
  752. instantiated buffers beyond i_size. Thus i_size is not actually
  753. changed so our warning message is enough. Unfortunately it is not
  754. possible to easily determine if i_size is being changed or not hence
  755. we just emit an appropriately worded error message.
  756. 2.0.25 - Small bug fixes and cleanups.
  757. - Unlock the page in an out of memory error code path in
  758. fs/ntfs/aops.c::ntfs_read_block().
  759. - If fs/ntfs/aops.c::ntfs_read_page() is called on an uptodate page,
  760. just unlock the page and return. (This can happen due to ->writepage
  761. clearing PageUptodate() during write out of MstProtected()
  762. attributes.
  763. - Remove leaked write code again.
  764. 2.0.24 - Cleanups.
  765. - Treat BUG_ON() as ASSERT() not VERIFY(), i.e. do not use side effects
  766. inside BUG_ON(). (Adam J. Richter)
  767. - Split logical OR expressions inside BUG_ON() into individual BUG_ON()
  768. calls for improved debugging. (Adam J. Richter)
  769. - Add errors flag to the ntfs volume state, accessed via
  770. NVol{,Set,Clear}Errors(vol).
  771. - Do not allow read-write remounts of read-only volumes with errors.
  772. - Clarify comment for ntfs file operation sendfile which was added by
  773. Christoph Hellwig a while ago (just using generic_file_sendfile())
  774. to say that ntfs ->sendfile is only used for the case where the
  775. source data is on the ntfs partition and the destination is
  776. somewhere else, i.e. nothing we need to concern ourselves with.
  777. - Add generic_file_write() as our ntfs file write operation.
  778. 2.0.23 - Major bug fixes (races, deadlocks, non-i386 architectures).
  779. - Massive internal locking changes to mft record locking. Fixes lock
  780. recursion and replaces the mrec_lock read/write semaphore with a
  781. mutex. Also removes the now superfluous mft_count. This fixes several
  782. race conditions and deadlocks, especially in the future write code.
  783. - Fix ntfs over loopback for compressed files by adding an
  784. optimization barrier. (gcc was screwing up otherwise ?)
  785. - Miscellaneous cleanups all over the code and a fix or two in error
  786. handling code paths.
  787. Thanks go to Christoph Hellwig for pointing out the following two:
  788. - Remove now unused function fs/ntfs/malloc.h::vmalloc_nofs().
  789. - Fix ntfs_free() for ia64 and parisc by checking for VMALLOC_END, too.
  790. 2.0.22 - Cleanups, mainly to ntfs_readdir(), and use C99 initializers.
  791. - Change fs/ntfs/dir.c::ntfs_reddir() to only read/write ->f_pos once
  792. at entry/exit respectively.
  793. - Use C99 initializers for structures.
  794. - Remove unused variable blocks from fs/ntfs/aops.c::ntfs_read_block().
  795. 2.0.21 - Check for, and refuse to work with too large files/directories/volumes.
  796. - Limit volume size at mount time to 2TiB on architectures where
  797. unsigned long is 32-bits (fs/ntfs/super.c::parse_ntfs_boot_sector()).
  798. This is the most we can do without overflowing the 32-bit limit of
  799. the block device size imposed on us by sb_bread() and sb_getblk()
  800. for the time being.
  801. - Limit file/directory size at open() time to 16TiB on architectures
  802. where unsigned long is 32-bits (fs/ntfs/file.c::ntfs_file_open() and
  803. fs/ntfs/dir.c::ntfs_dir_open()). This is the most we can do without
  804. overflowing the page cache page index.
  805. 2.0.20 - Support non-resident directory index bitmaps, fix page leak in readdir.
  806. - Move the directory index bitmap to use an attribute inode instead of
  807. having special fields for it inside the ntfs inode structure. This
  808. means that the index bitmaps now use the page cache for i/o, too,
  809. and also as a side effect we get support for non-resident index
  810. bitmaps for free.
  811. - Simplify/cleanup error handling in fs/ntfs/dir.c::ntfs_readdir() and
  812. fix a page leak that manifested itself in some cases.
  813. - Add fs/ntfs/inode.c::ntfs_put_inode(), which we need to release the
  814. index bitmap inode on the final iput().
  815. 2.0.19 - Fix race condition, improvements, and optimizations in i/o interface.
  816. - Apply block optimization added to fs/ntfs/aops.c::ntfs_read_block()
  817. to fs/ntfs/compress.c::ntfs_file_read_compressed_block() as well.
  818. - Drop the "file" from ntfs_file_read_compressed_block().
  819. - Rename fs/ntfs/aops.c::ntfs_enb_buffer_read_async() to
  820. ntfs_end_buffer_async_read() (more like the fs/buffer.c counterpart).
  821. - Update ntfs_end_buffer_async_read() with the improved logic from
  822. its updated counterpart fs/buffer.c::end_buffer_async_read(). Apply
  823. further logic improvements to better determine when we set PageError.
  824. - Update submission of buffers in fs/ntfs/aops.c::ntfs_read_block() to
  825. check for the buffers being uptodate first in line with the updated
  826. fs/buffer.c::block_read_full_page(). This plugs a small race
  827. condition.
  828. 2.0.18 - Fix race condition in reading of compressed files.
  829. - There was a narrow window between checking a buffer head for being
  830. uptodate and locking it in ntfs_file_read_compressed_block(). We now
  831. lock the buffer and then check whether it is uptodate or not.
  832. 2.0.17 - Cleanups and optimizations - shrinking the ToDo list.
  833. - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to return an error
  834. code and update callers, i.e. ntfs_iget(), to pass that error code
  835. up instead of just using -EIO.
  836. - Modifications to super.c to ensure that both mount and remount
  837. cannot set any write related options when the driver is compiled
  838. read-only.
  839. - Optimize block resolution in fs/ntfs/aops.c::ntfs_read_block() to
  840. cache the current runlist element. This should improve performance
  841. when reading very large and/or very fragmented data.
  842. 2.0.16 - Convert access to $MFT/$BITMAP to attribute inode API.
  843. - Fix a stupid bug introduced in 2.0.15 where we were unmapping the
  844. wrong inode in fs/ntfs/inode.c::ntfs_attr_iget().
  845. - Fix debugging check in fs/ntfs/aops.c::ntfs_read_block().
  846. - Convert $MFT/$BITMAP access to attribute inode API and remove all
  847. remnants of the ugly mftbmp address space and operations hack. This
  848. means we finally have only one readpage function as well as only one
  849. async io completion handler. Yey! The mft bitmap is now just an
  850. attribute inode and is accessed from vol->mftbmp_ino just as if it
  851. were a normal file. Fake inodes rule. (-:
  852. 2.0.15 - Fake inodes based attribute i/o via the pagecache, fixes and cleanups.
  853. - Fix silly bug in fs/ntfs/super.c::parse_options() which was causing
  854. remounts to fail when the partition had an entry in /etc/fstab and
  855. the entry specified the nls= option.
  856. - Apply same macro magic used in fs/ntfs/inode.h to fs/ntfs/volume.h to
  857. expand all the helper functions NVolFoo(), NVolSetFoo(), and
  858. NVolClearFoo().
  859. - Move copyright statement from driver initialisation message to
  860. module description (fs/super.c). This makes the initialisation
  861. message fit on one line and fits in better with rest of kernel.
  862. - Update fs/ntfs/attrib.c::map_run_list() to work on both real and
  863. attribute inodes, and both for files and directories.
  864. - Implement fake attribute inodes allowing all attribute i/o to go via
  865. the page cache and to use all the normal vfs/mm functionality:
  866. - Add ntfs_attr_iget() and its helper ntfs_read_locked_attr_inode()
  867. to fs/ntfs/inode.c.
  868. - Add needed cleanup code to ntfs_clear_big_inode().
  869. - Merge address space operations for files and directories (aops.c),
  870. now just have ntfs_aops:
  871. - Rename:
  872. end_buffer_read_attr_async() -> ntfs_end_buffer_read_async(),
  873. ntfs_attr_read_block() -> ntfs_read_block(),
  874. ntfs_file_read_page() -> ntfs_readpage().
  875. - Rewrite fs/ntfs/aops.c::ntfs_readpage() to work on both real and
  876. attribute inodes, and both for files and directories.
  877. - Remove obsolete fs/ntfs/aops.c::ntfs_mst_readpage().
  878. 2.0.14 - Run list merging code cleanup, minor locking changes, typo fixes.
  879. - Change fs/ntfs/super.c::ntfs_statfs() to not rely on BKL by moving
  880. the locking out of super.c::get_nr_free_mft_records() and taking and
  881. dropping the mftbmp_lock rw_semaphore in ntfs_statfs() itself.
  882. - Bring attribute runlist merging code (fs/ntfs/attrib.c) in sync with
  883. current userspace ntfs library code. This means that if a merge
  884. fails the original runlists are always left unmodified instead of
  885. being silently corrupted.
  886. - Misc typo fixes.
  887. 2.0.13 - Use iget5_locked() in preparation for fake inodes and small cleanups.
  888. - Remove nr_mft_bits and the now superfluous union with nr_mft_records
  889. from ntfs_volume structure.
  890. - Remove nr_lcn_bits and the now superfluous union with nr_clusters
  891. from ntfs_volume structure.
  892. - Use iget5_locked() and friends instead of conventional iget(). Wrap
  893. the call in fs/ntfs/inode.c::ntfs_iget() and update callers of iget()
  894. to use ntfs_iget(). Leave only one iget() call at mount time so we
  895. don't need an ntfs_iget_mount().
  896. - Change fs/ntfs/inode.c::ntfs_new_extent_inode() to take mft_no as an
  897. additional argument.
  898. 2.0.12 - Initial cleanup of address space operations following 2.0.11 changes.
  899. - Merge fs/ntfs/aops.c::end_buffer_read_mst_async() and
  900. fs/ntfs/aops.c::end_buffer_read_file_async() into one function
  901. fs/ntfs/aops.c::end_buffer_read_attr_async() using NInoMstProtected()
  902. to determine whether to apply mst fixups or not.
  903. - Above change allows merging fs/ntfs/aops.c::ntfs_file_read_block()
  904. and fs/ntfs/aops.c::ntfs_mst_readpage() into one function
  905. fs/ntfs/aops.c::ntfs_attr_read_block(). Also, create a tiny wrapper
  906. fs/ntfs/aops.c::ntfs_mst_readpage() to transform the parameters from
  907. the VFS readpage function prototype to the ntfs_attr_read_block()
  908. function prototype.
  909. 2.0.11 - Initial preparations for fake inode based attribute i/o.
  910. - Move definition of ntfs_inode_state_bits to fs/ntfs/inode.h and
  911. do some macro magic (adapted from include/linux/buffer_head.h) to
  912. expand all the helper functions NInoFoo(), NInoSetFoo(), and
  913. NInoClearFoo().
  914. - Add new flag to ntfs_inode_state_bits: NI_Sparse.
  915. - Add new fields to ntfs_inode structure to allow use of fake inodes
  916. for attribute i/o: type, name, name_len. Also add new state bits:
  917. NI_Attr, which, if set, indicates the inode is a fake inode, and
  918. NI_MstProtected, which, if set, indicates the attribute uses multi
  919. sector transfer protection, i.e. fixups need to be applied after
  920. reads and before/after writes.
  921. - Rename fs/ntfs/inode.c::ntfs_{new,clear,destroy}_inode() to
  922. ntfs_{new,clear,destroy}_extent_inode() and update callers.
  923. - Use ntfs_clear_extent_inode() in fs/ntfs/inode.c::__ntfs_clear_inode()
  924. instead of ntfs_destroy_extent_inode().
  925. - Cleanup memory deallocations in {__,}ntfs_clear_{,big_}inode().
  926. - Make all operations on ntfs inode state bits use the NIno* functions.
  927. - Set up the new ntfs inode fields and state bits in
  928. fs/ntfs/inode.c::ntfs_read_inode() and add appropriate cleanup of
  929. allocated memory to __ntfs_clear_inode().
  930. - Cleanup ntfs_inode structure a bit for better ordering of elements
  931. w.r.t. their size to allow better packing of the structure in memory.
  932. 2.0.10 - There can only be 2^32 - 1 inodes on an NTFS volume.
  933. - Add check at mount time to verify that the number of inodes on the
  934. volume does not exceed 2^32 - 1, which is the maximum allowed for
  935. NTFS according to Microsoft.
  936. - Change mft_no member of ntfs_inode structure to be unsigned long.
  937. Update all users. This makes ntfs_inode->mft_no just a copy of struct
  938. inode->i_ino. But we can't just always use struct inode->i_ino and
  939. remove mft_no because extent inodes do not have an attached struct
  940. inode.
  941. 2.0.9 - Decompression engine now uses a single buffer and other cleanups.
  942. - Change decompression engine to use a single buffer protected by a
  943. spin lock instead of per-CPU buffers. (Rusty Russell)
  944. - Do not update cb_pos when handling a partial final page during
  945. decompression of a sparse compression block, as the value is later
  946. reset without being read/used. (Rusty Russell)
  947. - Switch to using the new KM_BIO_SRC_IRQ for atomic kmap()s. (Andrew
  948. Morton)
  949. - Change buffer size in ntfs_readdir()/ntfs_filldir() to use
  950. NLS_MAX_CHARSET_SIZE which makes the buffers almost 1kiB each but
  951. it also makes everything safer so it is a good thing.
  952. - Miscellaneous minor cleanups to comments.
  953. 2.0.8 - Major updates for handling of case sensitivity and dcache aliasing.
  954. Big thanks go to Al Viro and other inhabitants of #kernel for investing
  955. their time to discuss the case sensitivity and dcache aliasing issues.
  956. - Remove unused source file fs/ntfs/attraops.c.
  957. - Remove show_inodes mount option(s), thus dropping support for
  958. displaying of short file names.
  959. - Remove deprecated mount option posix.
  960. - Restore show_sys_files mount option.
  961. - Add new mount option case_sensitive, to determine if the driver
  962. treats file names as case sensitive or not. If case sensitive, create
  963. file names in the POSIX namespace. Otherwise create file names in the
  964. LONG/WIN32 namespace. Note, files remain accessible via their short
  965. file name, if it exists.
  966. - Remove really dumb logic bug in boot sector recovery code.
  967. - Fix dcache aliasing issues wrt short/long file names via changes
  968. to fs/ntfs/dir.c::ntfs_lookup_inode_by_name() and
  969. fs/ntfs/namei.c::ntfs_lookup():
  970. - Add additional argument to ntfs_lookup_inode_by_name() in which we
  971. return information about the matching file name if the case is not
  972. matching or the match is a short file name. See comments above the
  973. function definition for details.
  974. - Change ntfs_lookup() to only create dcache entries for the correctly
  975. cased file name and only for the WIN32 namespace counterpart of DOS
  976. namespace file names. This ensures we have only one dentry per
  977. directory and also removes all dcache aliasing issues between short
  978. and long file names once we add write support. See comments above
  979. function for details.
  980. - Fix potential 1 byte overflow in fs/ntfs/unistr.c::ntfs_ucstonls().
  981. 2.0.7 - Minor cleanups and updates for changes in core kernel code.
  982. - Remove much of the NULL struct element initializers.
  983. - Various updates to make compatible with recent kernels.
  984. - Remove defines of MAX_BUF_PER_PAGE and include linux/buffer_head.h
  985. in fs/ntfs/ntfs.h instead.
  986. - Remove no longer needed KERNEL_VERSION checks. We are now in the
  987. kernel proper so they are no longer needed.
  988. 2.0.6 - Major bugfix to make compatible with other kernel changes.
  989. - Initialize the mftbmp address space properly now that there are more
  990. fields in the struct address_space. This was leading to hangs and
  991. oopses on umount since 2.5.12 because of changes to other parts of
  992. the kernel. We probably want a kernel generic init_address_space()
  993. function...
  994. - Drop BKL from ntfs_readdir() after consultation with Al Viro. The
  995. only caller of ->readdir() is vfs_readdir() which holds i_sem during
  996. the call, and i_sem is sufficient protection against changes in the
  997. directory inode (including ->i_size).
  998. - Use generic_file_llseek() for directories (as opposed to
  999. default_llseek()) as this downs i_sem instead of the BKL which is
  1000. what we now need for exclusion against ->f_pos changes considering we
  1001. no longer take the BKL in ntfs_readdir().
  1002. 2.0.5 - Major bugfix. Buffer overflow in extent inode handling.
  1003. - No need to set old blocksize in super.c::ntfs_fill_super() as the
  1004. VFS does so via invocation of deactivate_super() calling
  1005. fs->fill_super() calling block_kill_super() which does it.
  1006. - BKL moved from VFS into dir.c::ntfs_readdir(). (Linus Torvalds)
  1007. -> Do we really need it? I don't think so as we have exclusion on
  1008. the directory ntfs_inode rw_semaphore mrec_lock. We mmight have to
  1009. move the ->f_pos accesses under the mrec_lock though. Check this...
  1010. - Fix really, really, really stupid buffer overflow in extent inode
  1011. handling in mft.c::map_extent_mft_record().
  1012. 2.0.4 - Cleanups and updates for kernel 2.5.11.
  1013. - Add documentation on how to use the MD driver to be able to use NTFS
  1014. stripe and volume sets in Linux and generally cleanup documentation
  1015. a bit.
  1016. Remove all uses of kdev_t in favour of struct block_device *:
  1017. - Change compress.c::ntfs_file_read_compressed_block() to use
  1018. sb_getblk() instead of getblk().
  1019. - Change super.c::ntfs_fill_super() to use bdev_hardsect_size() instead
  1020. of get_hardsect_size().
  1021. - No need to get old blocksize in super.c::ntfs_fill_super() as
  1022. fs/super.c::get_sb_bdev() already does this.
  1023. - Set bh->b_bdev instead of bh->b_dev throughout aops.c.
  1024. 2.0.3 - Small bug fixes, cleanups, and performance improvements.
  1025. - Remove some dead code from mft.c.
  1026. - Optimize readpage and read_block functions throughout aops.c so that
  1027. only initialized blocks are read. Non-initialized ones have their
  1028. buffer head mapped, zeroed, and set up to date, without scheduling
  1029. any i/o. Thanks to Al Viro for advice on how to avoid the device i/o.
  1030. Thanks go to Andrew Morton for spotting the below:
  1031. - Fix buglet in allocate_compression_buffers() error code path.
  1032. - Call flush_dcache_page() after modifying page cache page contents in
  1033. ntfs_file_readpage().
  1034. - Check for existence of page buffers throughout aops.c before calling
  1035. create_empty_buffers(). This happens when an I/O error occurs and the
  1036. read is retried. (It also happens once writing is implemented so that
  1037. needed doing anyway but I had left it for later...)
  1038. - Don't BUG_ON() uptodate and/or mapped buffers throughout aops.c in
  1039. readpage and read_block functions. Reasoning same as above (i.e. I/O
  1040. error retries and future write code paths.)
  1041. 2.0.2 - Minor updates and cleanups.
  1042. - Cleanup: rename mst.c::__post_read_mst_fixup to post_write_mst_fixup
  1043. and cleanup the code a bit, removing the unused size parameter.
  1044. - Change default fmask to 0177 and update documentation.
  1045. - Change attrib.c::get_attr_search_ctx() to return the search context
  1046. directly instead of taking the address of a pointer. A return value
  1047. of NULL means the allocation failed. Updated all callers
  1048. appropriately.
  1049. - Update to 2.5.9 kernel (preserving backwards compatibility) by
  1050. replacing all occurences of page->buffers with page_buffers(page).
  1051. - Fix minor bugs in runlist merging, also minor cleanup.
  1052. - Updates to bootsector layout and mft mirror contents descriptions.
  1053. - Small bug fix in error detection in unistr.c and some cleanups.
  1054. - Grow name buffer allocations in unistr.c in aligned mutlipled of 64
  1055. bytes.
  1056. 2.0.1 - Minor updates.
  1057. - Make default umask correspond to documentation.
  1058. - Improve documentation.
  1059. - Set default mode to include execute bit. The {u,f,d}mask can be used
  1060. to take it away if desired. This allows binaries to be executed from
  1061. a mounted ntfs partition.
  1062. 2.0.0 - New version number. Remove TNG from the name. Now in the kernel.
  1063. - Add kill_super, just keeping up with the vfs changes in the kernel.
  1064. - Repeat some changes from tng-0.0.8 that somehow got lost on the way
  1065. from the CVS import into BitKeeper.
  1066. - Begin to implement proper handling of allocated_size vs
  1067. initialized_size vs data_size (i.e. i_size). Done are
  1068. mft.c::ntfs_mft_readpage(), aops.c::end_buffer_read_index_async(),
  1069. and attrib.c::load_attribute_list().
  1070. - Lock the runlist in attrib.c::load_attribute_list() while using it.
  1071. - Fix memory leak in ntfs_file_read_compressed_block() and generally
  1072. clean up compress.c a little, removing some uncommented/unused debug
  1073. code.
  1074. - Tidy up dir.c a little bit.
  1075. - Don't bother getting the runlist in inode.c::ntfs_read_inode().
  1076. - Merge mft.c::ntfs_mft_readpage() and aops.c::ntfs_index_readpage()
  1077. creating aops.c::ntfs_mst_readpage(), improving the handling of
  1078. holes and overflow in the process and implementing the correct
  1079. equivalent of ntfs_file_get_block() in ntfs_mst_readpage() itself.
  1080. I am aiming for correctness at the moment. Modularisation can come
  1081. later.
  1082. - Rename aops.c::end_buffer_read_index_async() to
  1083. end_buffer_read_mst_async() and optimize the overflow checking and
  1084. handling.
  1085. - Use the host of the mftbmp address space mapping to hold the ntfs
  1086. volume. This is needed so the async i/o completion handler can
  1087. retrieve a pointer to the volume. Hopefully this will not cause
  1088. problems elsewhere in the kernel... Otherwise will need to use a
  1089. fake inode.
  1090. - Complete implementation of proper handling of allocated_size vs
  1091. initialized_size vs data_size (i.e. i_size) in whole driver.
  1092. Basically aops.c is now completely rewritten.
  1093. - Change NTFS driver name to just NTFS and set version number to 2.0.0
  1094. to make a clear distinction from the old driver which is still on
  1095. version 1.1.22.
  1096. tng-0.0.8 - 08/03/2002 - Now using BitKeeper, http://linux-ntfs.bkbits.net/
  1097. - Replace bdevname(sb->s_dev) with sb->s_id.
  1098. - Remove now superfluous new-line characters in all callers of
  1099. ntfs_debug().
  1100. - Apply kludge in ntfs_read_inode(), setting i_nlink to 1 for
  1101. directories. Without this the "find" utility gets very upset which is
  1102. fair enough as Linux/Unix do not support directory hard links.
  1103. - Further runlist merging work. (Richard Russon)
  1104. - Backwards compatibility for gcc-2.95. (Richard Russon)
  1105. - Update to kernel 2.5.5-pre1 and rediff the now tiny patch.
  1106. - Convert to new filesystem declaration using ->ntfs_get_sb() and
  1107. replacing ntfs_read_super() with ntfs_fill_super().
  1108. - Set s_maxbytes to MAX_LFS_FILESIZE to avoid page cache page index
  1109. overflow on 32-bit architectures.
  1110. - Cleanup upcase loading code to use ntfs_(un)map_page().
  1111. - Disable/reenable preemtion in critical sections of compession engine.
  1112. - Replace device size determination in ntfs_fill_super() with
  1113. sb->s_bdev->bd_inode->i_size (in bytes) and remove now superfluous
  1114. function super.c::get_nr_blocks().
  1115. - Implement a mount time option (show_inodes) allowing choice of which
  1116. types of inode names readdir() returns and modify ntfs_filldir()
  1117. accordingly. There are several parameters to show_inodes:
  1118. system: system files
  1119. win32: long file names (including POSIX file names) [DEFAULT]
  1120. long: same as win32
  1121. dos: short file names only (excluding POSIX file names)
  1122. short: same as dos
  1123. posix: same as both win32 and dos
  1124. all: all file names
  1125. Note that the options are additive, i.e. specifying:
  1126. -o show_inodes=system,show_inodes=win32,show_inodes=dos
  1127. is the same as specifying:
  1128. -o show_inodes=all
  1129. Note that the "posix" and "all" options will show all directory
  1130. names, BUT the link count on each directory inode entry is set to 1,
  1131. due to Linux not supporting directory hard links. This may well
  1132. confuse some userspace applications, since the directory names will
  1133. have the same inode numbers. Thus it is NOT advisable to use the
  1134. "posix" or "all" options. We provide them only for completeness sake.
  1135. - Add copies of allocated_size, initialized_size, and compressed_size to
  1136. the ntfs inode structure and set them up in
  1137. inode.c::ntfs_read_inode(). These reflect the unnamed data attribute
  1138. for files and the index allocation attribute for directories.
  1139. - Add copies of allocated_size and initialized_size to ntfs inode for
  1140. $BITMAP attribute of large directories and set them up in
  1141. inode.c::ntfs_read_inode().
  1142. - Add copies of allocated_size and initialized_size to ntfs volume for
  1143. $BITMAP attribute of $MFT and set them up in
  1144. super.c::load_system_files().
  1145. - Parse deprecated ntfs driver options (iocharset, show_sys_files,
  1146. posix, and utf8) and tell user what the new options to use are. Note
  1147. we still do support them but they will be removed with kernel 2.7.x.
  1148. - Change all occurences of integer long long printf formatting to hex
  1149. as printk() will not support long long integer format if/when the
  1150. div64 patch goes into the kernel.
  1151. - Make slab caches have stable names and change the names to what they
  1152. were intended to be. These changes are required/made possible by the
  1153. new slab cache name handling which removes the length limitation by
  1154. requiring the caller of kmem_cache_create() to supply a stable name
  1155. which is then referenced but not copied.
  1156. - Rename run_list structure to run_list_element and create a new
  1157. run_list structure containing a pointer to a run_list_element
  1158. structure and a read/write semaphore. Adapt all users of runlists
  1159. to new scheme and take and release the lock as needed. This fixes a
  1160. nasty race as the run_list changes even when inodes are locked for
  1161. reading and even when the inode isn't locked at all, so we really
  1162. needed the serialization. We use a semaphore rather than a spinlock
  1163. as memory allocations can sleep and doing everything GFP_ATOMIC
  1164. would be silly.
  1165. - Cleanup read_inode() removing all code checking for lowest_vcn != 0.
  1166. This can never happen due to the nature of lookup_attr() and how we
  1167. support attribute lists. If it did happen it would imply the inode
  1168. being corrupt.
  1169. - Check for lowest_vcn != 0 in ntfs_read_inode() and mark the inode as
  1170. bad if found.
  1171. - Update to 2.5.6-pre2 changes in struct address_space.
  1172. - Use parent_ino() when accessing d_parent inode number in dir.c.
  1173. - Import Sourceforge CVS repository into BitKeeper repository:
  1174. http://linux-ntfs.bkbits.net/ntfs-tng-2.5
  1175. - Update fs/Makefile, fs/Config.help, fs/Config.in, and
  1176. Documentation/filesystems/ntfs.txt for NTFS TNG.
  1177. - Create kernel configuration option controlling whether debugging
  1178. is enabled or not.
  1179. - Add the required export of end_buffer_io_sync() from the patches
  1180. directory to the kernel code.
  1181. - Update inode.c::ntfs_show_options() with show_inodes mount option.
  1182. - Update errors mount option.
  1183. tng-0.0.7 - 13/02/2002 - The driver is now feature complete for read-only!
  1184. - Cleanup mft.c and it's debug/error output in particular. Fix a minor
  1185. bug in mapping of extent inodes. Update all the comments to fit all
  1186. the recent code changes.
  1187. - Modify vcn_to_lcn() to cope with entirely unmapped runlists.
  1188. - Cleanups in compress.c, mostly comments and folding help.
  1189. - Implement attrib.c::map_run_list() as a generic helper.
  1190. - Make compress.c::ntfs_file_read_compressed_block() use map_run_list()
  1191. thus making code shorter and enabling attribute list support.
  1192. - Cleanup incorrect use of [su]64 with %L printf format specifier in
  1193. all source files. Type casts to [unsigned] long long added to correct
  1194. the mismatches (important for architectures which have long long not
  1195. being 64 bits).
  1196. - Merge async io completion handlers for directory indexes and $MFT
  1197. data into one by setting the index_block_size{_bits} of the ntfs
  1198. inode for $MFT to the mft_record_size{_bits} of the ntfs_volume.
  1199. - Cleanup aops.c, update comments.
  1200. - Make ntfs_file_get_block() use map_run_list() so all files now
  1201. support attribute lists.
  1202. - Make ntfs_dir_readpage() almost verbatim copy of
  1203. block_read_full_page() by using ntfs_file_get_block() with only real
  1204. difference being the use of our own async io completion handler
  1205. rather than the default one, thus reducing the amount of code and
  1206. automatically enabling attribute list support for directory indices.
  1207. - Fix bug in load_attribute_list() - forgot to call brelse in error
  1208. code path.
  1209. - Change parameters to find_attr() and lookup_attr(). We no longer
  1210. pass in the upcase table and its length. These can be gotten from
  1211. ctx->ntfs_ino->vol->upcase{_len}. Update all callers.
  1212. - Cleanups in attrib.c.
  1213. - Implement merging of runlists, attrib.c::merge_run_lists() and its
  1214. helpers. (Richard Russon)
  1215. - Attribute lists part 2, attribute extents and multi part runlists:
  1216. enable proper support for LCN_RL_NOT_MAPPED and automatic mapping of
  1217. further runlist parts via attrib.c::map_run_list().
  1218. - Tiny endianness bug fix in decompress_mapping_pairs().
  1219. tng-0.0.6 - Encrypted directories, bug fixes, cleanups, debugging enhancements.
  1220. - Enable encrypted directories. (Their index root is marked encrypted
  1221. to indicate that new files in that directory should be created
  1222. encrypted.)
  1223. - Fix bug in NInoBmpNonResident() macro. (Cut and paste error.)
  1224. - Enable $Extend system directory. Most (if not all) extended system
  1225. files do not have unnamed data attributes so ntfs_read_inode() had to
  1226. special case them but that is ok, as the special casing recovery
  1227. happens inside an error code path so there is zero slow down in the
  1228. normal fast path. The special casing is done by introducing a new
  1229. function inode.c::ntfs_is_extended_system_file() which checks if any
  1230. of the hard links in the inode point to $Extend as being their parent
  1231. directory and if they do we assume this is an extended system file.
  1232. - Create a sysctl/proc interface to allow {dis,en}abling of debug output
  1233. when compiled with -DDEBUG. Default is debug messages to be disabled.
  1234. To enable them, one writes a non-zero value to /proc/sys/fs/ntfs-debug
  1235. (if /proc is enabled) or uses sysctl(2) to effect the same (if sysctl
  1236. interface is enabled). Inspired by old ntfs driver.
  1237. - Add debug_msgs insmod/kernel boot parameter to set whether debug
  1238. messages are {dis,en}abled. This is useful to enable debug messages
  1239. during ntfs initialization and is the only way to activate debugging
  1240. when the sysctl interface is not enabled.
  1241. - Cleanup debug output in various places.
  1242. - Remove all dollar signs ($) from the source (except comments) to
  1243. enable compilation on architectures whose gcc compiler does not
  1244. support dollar signs in the names of variables/constants. Attribute
  1245. types now start with AT_ instead of $ and $I30 is now just I30.
  1246. - Cleanup ntfs_lookup() and add consistency check of sequence numbers.
  1247. - Load complete runlist for $MFT/$BITMAP during mount and cleanup
  1248. access functions. This means we now cope with $MFT/$BITMAP being
  1249. spread accross several mft records.
  1250. - Disable modification of mft_zone_multiplier on remount. We can always
  1251. reenable this later on if we really want to, but we will need to make
  1252. sure we readjust the mft_zone size / layout accordingly.
  1253. tng-0.0.5 - Modernize for 2.5.x and further in line-ing with Al Viro's comments.
  1254. - Use sb_set_blocksize() instead of set_blocksize() and verify the
  1255. return value.
  1256. - Use sb_bread() instead of bread() throughout.
  1257. - Add index_vcn_size{_bits} to ntfs_inode structure to store the size
  1258. of a directory index block vcn. Apply resulting simplifications in
  1259. dir.c everywhere.
  1260. - Fix a small bug somewhere (but forgot what it was).
  1261. - Change ntfs_{debug,error,warning} to enable gcc to do type checking
  1262. on the printf-format parameter list and fix bugs reported by gcc
  1263. as a result. (Richard Russon)
  1264. - Move inode allocation strategy to Al's new stuff but maintain the
  1265. divorce of ntfs_inode from struct inode. To achieve this we have two
  1266. separate slab caches, one for big ntfs inodes containing a struct
  1267. inode and pure ntfs inodes and at the same time fix some faulty
  1268. error code paths in ntfs_read_inode().
  1269. - Show mount options in proc (inode.c::ntfs_show_options()).
  1270. tng-0.0.4 - Big changes, getting in line with Al Viro's comments.
  1271. - Modified (un)map_mft_record functions to be common for read and write
  1272. case. To specify which is which, added extra parameter at front of
  1273. parameter list. Pass either READ or WRITE to this, each has the
  1274. obvious meaning.
  1275. - General cleanups to allow for easier folding in vi.
  1276. - attrib.c::decompress_mapping_pairs() now accepts the old runlist
  1277. argument, and invokes attrib.c::merge_run_lists() to merge the old
  1278. and the new runlists.
  1279. - Removed attrib.c::find_first_attr().
  1280. - Implemented loading of attribute list and complete runlist for $MFT.
  1281. This means we now cope with $MFT being spread across several mft
  1282. records.
  1283. - Adapt to 2.5.2-pre9 and the changed create_empty_buffers() syntax.
  1284. - Adapt major/minor/kdev_t/[bk]devname stuff to new 2.5.x kernels.
  1285. - Make ntfs_volume be allocated via kmalloc() instead of using a slab
  1286. cache. There are too little ntfs_volume structures at any one time
  1287. to justify a private slab cache.
  1288. - Fix bogus kmap() use in async io completion. Now use kmap_atomic().
  1289. Use KM_BIO_IRQ on advice from IRC/kernel...
  1290. - Use ntfs_map_page() in map_mft_record() and create ->readpage method
  1291. for reading $MFT (ntfs_mft_readpage). In the process create dedicated
  1292. address space operations (ntfs_mft_aops) for $MFT inode mapping. Also
  1293. removed the now superfluous exports from the kernel core patch.
  1294. - Fix a bug where kfree() was used insted of ntfs_free().
  1295. - Change map_mft_record() to take ntfs_inode as argument instead of
  1296. vfs inode. Dito for unmap_mft_record(). Adapt all callers.
  1297. - Add pointer to ntfs_volume to ntfs_inode.
  1298. - Add mft record number and sequence number to ntfs_inode. Stop using
  1299. i_ino and i_generation for in-driver purposes.
  1300. - Implement attrib.c::merge_run_lists(). (Richard Russon)
  1301. - Remove use of proper inodes by extent inodes. Move i_ino and
  1302. i_generation to ntfs_inode to do this. Apply simplifications that
  1303. result and remove iget_no_wait(), etc.
  1304. - Pass ntfs_inode everywhere in the driver (used to be struct inode).
  1305. - Add reference counting in ntfs_inode for the ntfs inode itself and
  1306. for the mapped mft record.
  1307. - Extend mft record mapping so we can (un)map extent mft records (new
  1308. functions (un)map_extent_mft_record), and so mappings are reference
  1309. counted and don't have to happen twice if already mapped - just ref
  1310. count increases.
  1311. - Add -o iocharset as alias to -o nls for backwards compatibility.
  1312. - The latest core patch is now tiny. In fact just a single additional
  1313. export is necessary over the base kernel.
  1314. tng-0.0.3 - Cleanups, enhancements, bug fixes.
  1315. - Work on attrib.c::decompress_mapping_pairs() to detect base extents
  1316. and setup the runlist appropriately using knowledge provided by the
  1317. sizes in the base attribute record.
  1318. - Balance the get_/put_attr_search_ctx() calls so we don't leak memory
  1319. any more.
  1320. - Introduce ntfs_malloc_nofs() and ntfs_free() to allocate/free a single
  1321. page or use vmalloc depending on the amount of memory requested.
  1322. - Cleanup error output. The __FUNCTION__ "(): " is now added
  1323. automatically. Introduced a new header file debug.h to support this
  1324. and also moved ntfs_debug() function into it.
  1325. - Make reading of compressed files more intelligent and especially get
  1326. rid of the vmalloc_nofs() from readpage(). This now uses per CPU
  1327. buffers (allocated at first mount with cluster size <= 4kiB and
  1328. deallocated on last umount with cluster size <= 4kiB), and
  1329. asynchronous io for the compressed data using a list of buffer heads.
  1330. Er, we use synchronous io as async io only works on whole pages
  1331. covered by buffers and not on individual buffer heads...
  1332. - Bug fix for reading compressed files with sparse compression blocks.
  1333. tng-0.0.2 - Now handles larger/fragmented/compressed volumes/files/dirs.
  1334. - Fixed handling of directories when cluster size exceeds index block
  1335. size.
  1336. - Hide DOS only name space directory entries from readdir() but allow
  1337. them in lookup(). This should fix the problem that Linux doesn't
  1338. support directory hard links, while still allowing access to entries
  1339. via their short file name. This also has the benefit of mimicking
  1340. what Windows users are used to, so it is the ideal solution.
  1341. - Implemented sync_page everywhere so no more hangs in D state when
  1342. waiting for a page.
  1343. - Stop using bforget() in favour of brelse().
  1344. - Stop locking buffers unnecessarily.
  1345. - Implemented compressed files (inode->mapping contains uncompressed
  1346. data, raw compressed data is currently bread() into a vmalloc()ed
  1347. memory buffer).
  1348. - Enable compressed directories. (Their index root is marked compressed
  1349. to indicate that new files in that directory should be created
  1350. compressed.)
  1351. - Use vsnprintf rather than vsprintf in the ntfs_error and ntfs_warning
  1352. functions. (Thanks to Will Dyson for pointing this out.)
  1353. - Moved the ntfs_inode and ntfs_volume (the former ntfs_inode_info and
  1354. ntfs_sb_info) out of the common inode and super_block structures and
  1355. started using the generic_ip and generic_sbp pointers instead. This
  1356. makes ntfs entirely private with respect to the kernel tree.
  1357. - Detect compiler version and abort with error message if gcc less than
  1358. 2.96 is used.
  1359. - Fix bug in name comparison function in unistr.c.
  1360. - Implement attribute lists part 1, the infrastructure: search contexts
  1361. and operations, find_external_attr(), lookup_attr()) and make the
  1362. code use the infrastructure.
  1363. - Fix stupid buffer overflow bug that became apparent on larger run
  1364. list containing attributes.
  1365. - Fix bugs in readdir() that became apparent on larger directories.
  1366. The driver is now really useful and survives the test
  1367. find . -type f -exec md5sum "{}" \;
  1368. without any error messages on a over 1GiB sized partition with >16k
  1369. files on it, including compressed files and directories and many files
  1370. and directories with attribute lists.
  1371. tng-0.0.1 - The first useful version.
  1372. - Added ntfs_lookup().
  1373. - Added default upcase generation and handling.
  1374. - Added compile options to be shown on module init.
  1375. - Many bug fixes that were "hidden" before.
  1376. - Update to latest kernel.
  1377. - Added ntfs_readdir().
  1378. - Added file operations for mmap(), read(), open() and llseek(). We just
  1379. use the generic ones. The whole point of going through implementing
  1380. readpage() methods and where possible get_block() call backs is that
  1381. this allows us to make use of the generic high level methods provided
  1382. by the kernel.
  1383. The driver is now actually useful! Yey. (-: It undoubtedly has got bugs
  1384. though and it doesn't implement accesssing compressed files yet. Also,
  1385. accessing files with attribute list attributes is not implemented yet
  1386. either. But for small or simple filesystems it should work and allow
  1387. you to list directories, use stat on directory entries and the file
  1388. system, open, read, mmap and llseek around in files. A big mile stone
  1389. has been reached!
  1390. tng-0.0.0 - Initial version tag.
  1391. Initial driver implementation. The driver can mount and umount simple
  1392. NTFS filesystems (i.e. ones without attribute lists in the system
  1393. files). If the mount fails there might be problems in the error handling
  1394. code paths, so be warned. Otherwise it seems to be loading the system
  1395. files nicely and the mft record read mapping/unmapping seems to be
  1396. working nicely, too. Proof of inode metadata in the page cache and non-
  1397. resident file unnamed stream data in the page cache concepts is thus
  1398. complete.