ChangeLog 93 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690
  1. ToDo/Notes:
  2. - Find and fix bugs.
  3. - The only places in the kernel where a file is resized are
  4. ntfs_file_write*() and ntfs_truncate() for both of which i_mutex is
  5. held. Just have to be careful in read-/writepage and other helpers
  6. not running under i_mutex that we play nice. Also need to be careful
  7. with initialized_size extension in ntfs_file_write*() and writepage.
  8. UPDATE: The only things that need to be checked are the compressed
  9. write and the other attribute resize/write cases like index
  10. attributes, etc. For now none of these are implemented so are safe.
  11. - Implement filling in of holes in aops.c::ntfs_writepage() and its
  12. helpers.
  13. - Implement mft.c::sync_mft_mirror_umount(). We currently will just
  14. leave the volume dirty on umount if the final iput(vol->mft_ino)
  15. causes a write of any mirrored mft records due to the mft mirror
  16. inode having been discarded already. Whether this can actually ever
  17. happen is unclear however so it is worth waiting until someone hits
  18. the problem.
  19. 2.1.28 - Fix a deadlock.
  20. - Fix deadlock in fs/ntfs/inode.c::ntfs_put_inode(). Thanks to Sergey
  21. Vlasov for the report and detailed analysis of the deadlock. The fix
  22. involved getting rid of ntfs_put_inode() altogether and hence NTFS no
  23. longer has a ->put_inode super operation.
  24. 2.1.27 - Various bug fixes and cleanups.
  25. - Fix two compiler warnings on Alpha. Thanks to Andrew Morton for
  26. reporting them.
  27. - Fix an (innocent) off-by-one error in the runlist code.
  28. - Fix a buggette in an "should be impossible" case handling where we
  29. continued the attribute lookup loop instead of aborting it.
  30. - Use buffer_migrate_page() for the ->migratepage function of all ntfs
  31. address space operations.
  32. - Fix comparison of $MFT and $MFTMirr to not bail out when there are
  33. unused, invalid mft records which are the same in both $MFT and
  34. $MFTMirr.
  35. - Add support for sparse files which have a compression unit of 0.
  36. - Remove all the make_bad_inode() calls. This should only be called
  37. from read inode and new inode code paths.
  38. - Limit name length in fs/ntfs/unistr.c::ntfs_nlstoucs() to maximum
  39. allowed by NTFS, i.e. 255 Unicode characters, not including the
  40. terminating NULL (which is not stored on disk).
  41. - Improve comments on file attribute flags in fs/ntfs/layout.h.
  42. - Fix a bug in fs/ntfs/inode.c::ntfs_read_locked_index_inode() where we
  43. forgot to update a temporary variable so loading index inodes which
  44. have an index allocation attribute failed.
  45. - Add a missing call to flush_dcache_mft_record_page() in
  46. fs/ntfs/inode.c::ntfs_write_inode().
  47. - Handle the recently introduced -ENAMETOOLONG return value from
  48. fs/ntfs/unistr.c::ntfs_nlstoucs() in fs/ntfs/namei.c::ntfs_lookup().
  49. - Semaphore to mutex conversion. (Ingo Molnar)
  50. 2.1.26 - Minor bug fixes and updates.
  51. - Fix a potential overflow in file.c where a cast to s64 was missing in
  52. a left shift of a page index.
  53. - The struct inode has had its i_sem semaphore changed to a mutex named
  54. i_mutex.
  55. - We have struct kmem_cache now so use it instead of the typedef
  56. kmem_cache_t. (Pekka Enberg)
  57. - Implement support for sector sizes above 512 bytes (up to the maximum
  58. supported by NTFS which is 4096 bytes).
  59. - Do more detailed reporting of why we cannot mount read-write by
  60. special casing the VOLUME_MODIFIED_BY_CHKDSK flag.
  61. - Miscellaneous updates to layout.h.
  62. - Cope with attribute list attribute having invalid flags. Windows
  63. copes with this and even chkdsk does not detect or fix this so we
  64. have to cope with it, too. Thanks to Pawel Kot for reporting the
  65. problem.
  66. 2.1.25 - (Almost) fully implement write(2) and truncate(2).
  67. - Change ntfs_map_runlist_nolock(), ntfs_attr_find_vcn_nolock() and
  68. {__,}ntfs_cluster_free() to also take an optional attribute search
  69. context as argument. This allows calling these functions with the
  70. mft record mapped. Update all callers.
  71. - Fix potential deadlock in ntfs_mft_data_extend_allocation_nolock()
  72. error handling by passing in the active search context when calling
  73. ntfs_cluster_free().
  74. - Change ntfs_cluster_alloc() to take an extra boolean parameter
  75. specifying whether the cluster are being allocated to extend an
  76. attribute or to fill a hole.
  77. - Change ntfs_attr_make_non_resident() to call ntfs_cluster_alloc()
  78. with @is_extension set to TRUE and remove the runlist terminator
  79. fixup code as this is now done by ntfs_cluster_alloc().
  80. - Change ntfs_attr_make_non_resident to take the attribute value size
  81. as an extra parameter. This is needed since we need to know the size
  82. before we can map the mft record and our callers always know it. The
  83. reason we cannot simply read the size from the vfs inode i_size is
  84. that this is not necessarily uptodate. This happens when
  85. ntfs_attr_make_non_resident() is called in the ->truncate call path.
  86. - Fix ntfs_attr_make_non_resident() to update the vfs inode i_blocks
  87. which is zero for a resident attribute but should no longer be zero
  88. once the attribute is non-resident as it then has real clusters
  89. allocated.
  90. - Add fs/ntfs/attrib.[hc]::ntfs_attr_extend_allocation(), a function to
  91. extend the allocation of an attributes. Optionally, the data size,
  92. but not the initialized size can be extended, too.
  93. - Implement fs/ntfs/inode.[hc]::ntfs_truncate(). It only supports
  94. uncompressed and unencrypted files and it never creates sparse files
  95. at least for the moment (making a file sparse requires us to modify
  96. its directory entries and we do not support directory operations at
  97. the moment). Also, support for highly fragmented files, i.e. ones
  98. whose data attribute is split across multiple extents, is severly
  99. limited. When such a case is encountered, EOPNOTSUPP is returned.
  100. - Enable ATTR_SIZE attribute changes in ntfs_setattr(). This completes
  101. the initial implementation of file truncation. Now both open(2)ing
  102. a file with the O_TRUNC flag and the {,f}truncate(2) system calls
  103. will resize a file appropriately. The limitations are that only
  104. uncompressed and unencrypted files are supported. Also, there is
  105. only very limited support for highly fragmented files (the ones whose
  106. $DATA attribute is split into multiple attribute extents).
  107. - In attrib.c::ntfs_attr_set() call balance_dirty_pages_ratelimited()
  108. and cond_resched() in the main loop as we could be dirtying a lot of
  109. pages and this ensures we play nice with the VM and the system as a
  110. whole.
  111. - Implement file operations ->write, ->aio_write, ->writev for regular
  112. files. This replaces the old use of generic_file_write(), et al and
  113. the address space operations ->prepare_write and ->commit_write.
  114. This means that both sparse and non-sparse (unencrypted and
  115. uncompressed) files can now be extended using the normal write(2)
  116. code path. There are two limitations at present and these are that
  117. we never create sparse files and that we only have limited support
  118. for highly fragmented files, i.e. ones whose data attribute is split
  119. across multiple extents. When such a case is encountered,
  120. EOPNOTSUPP is returned.
  121. - $EA attributes can be both resident and non-resident.
  122. - Use %z for size_t to fix compilation warnings. (Andrew Morton)
  123. - Fix compilation warnings with gcc-4.0.2 on SUSE 10.0.
  124. - Document extended attribute ($EA) NEED_EA flag. (Based on libntfs
  125. patch by Yura Pakhuchiy.)
  126. 2.1.24 - Lots of bug fixes and support more clean journal states.
  127. - Support journals ($LogFile) which have been modified by chkdsk. This
  128. means users can boot into Windows after we marked the volume dirty.
  129. The Windows boot will run chkdsk and then reboot. The user can then
  130. immediately boot into Linux rather than having to do a full Windows
  131. boot first before rebooting into Linux and we will recognize such a
  132. journal and empty it as it is clean by definition. Note, this only
  133. works if chkdsk left the journal in an obviously clean state.
  134. - Support journals ($LogFile) with only one restart page as well as
  135. journals with two different restart pages. We sanity check both and
  136. either use the only sane one or the more recent one of the two in the
  137. case that both are valid.
  138. - Add fs/ntfs/malloc.h::ntfs_malloc_nofs_nofail() which is analogous to
  139. ntfs_malloc_nofs() but it performs allocations with __GFP_NOFAIL and
  140. hence cannot fail.
  141. - Use ntfs_malloc_nofs_nofail() in the two critical regions in
  142. fs/ntfs/runlist.c::ntfs_runlists_merge(). This means we no longer
  143. need to panic() if the allocation fails as it now cannot fail.
  144. - Fix two nasty runlist merging bugs that had gone unnoticed so far.
  145. Thanks to Stefano Picerno for the bug report.
  146. - Remove two bogus BUG_ON()s from fs/ntfs/mft.c.
  147. - Fix handling of valid but empty mapping pairs array in
  148. fs/ntfs/runlist.c::ntfs_mapping_pairs_decompress().
  149. - Report unrepresentable inodes during ntfs_readdir() as KERN_WARNING
  150. messages and include the inode number. Thanks to Yura Pakhuchiy for
  151. pointing this out.
  152. - Change ntfs_rl_truncate_nolock() to throw away the runlist if the new
  153. length is zero.
  154. - Add runlist.[hc]::ntfs_rl_punch_nolock() which punches a caller
  155. specified hole into a runlist.
  156. - Fix a bug in fs/ntfs/index.c::ntfs_index_lookup(). When the returned
  157. index entry is in the index root, we forgot to set the @ir pointer in
  158. the index context. Thanks to Yura Pakhuchiy for finding this bug.
  159. - Remove bogus setting of PageError in ntfs_read_compressed_block().
  160. - Add fs/ntfs/attrib.[hc]::ntfs_resident_attr_value_resize().
  161. - Fix a bug in ntfs_map_runlist_nolock() where we forgot to protect
  162. access to the allocated size in the ntfs inode with the size lock.
  163. - Fix ntfs_attr_vcn_to_lcn_nolock() and ntfs_attr_find_vcn_nolock() to
  164. return LCN_ENOENT when there is no runlist and the allocated size is
  165. zero.
  166. - Fix load_attribute_list() to handle the case of a NULL runlist.
  167. - Fix handling of sparse attributes in ntfs_attr_make_non_resident().
  168. - Add BUG() checks to ntfs_attr_make_non_resident() and ntfs_attr_set()
  169. to ensure that these functions are never called for compressed or
  170. encrypted attributes.
  171. - Fix cluster (de)allocators to work when the runlist is NULL and more
  172. importantly to take a locked runlist rather than them locking it
  173. which leads to lock reversal.
  174. - Truncate {a,c,m}time to the ntfs supported time granularity when
  175. updating the times in the inode in ntfs_setattr().
  176. - Fixup handling of sparse, compressed, and encrypted attributes in
  177. fs/ntfs/inode.c::ntfs_read_locked_{,attr_,index_}inode(),
  178. fs/ntfs/aops.c::ntfs_{read,write}page().
  179. - Make ntfs_write_block() not instantiate sparse blocks if they contain
  180. only zeroes.
  181. - Optimize fs/ntfs/aops.c::ntfs_write_block() by extending the page
  182. lock protection over the buffer submission for i/o which allows the
  183. removal of the get_bh()/put_bh() pairs for each buffer.
  184. - Fix fs/ntfs/aops.c::ntfs_{read,write}_block() to handle the case
  185. where a concurrent truncate has truncated the runlist under our feet.
  186. - Fix page_has_buffers()/page_buffers() handling in fs/ntfs/aops.c.
  187. - In fs/ntfs/aops.c::ntfs_end_buffer_async_read(), use a bit spin lock
  188. in the first buffer head instead of a driver global spin lock to
  189. improve scalability.
  190. - Minor fix to error handling and error message display in
  191. fs/ntfs/aops.c::ntfs_prepare_nonresident_write().
  192. - Change the mount options {u,f,d}mask to always parse the number as
  193. an octal number to conform to how chmod(1) works, too. Thanks to
  194. Giuseppe Bilotta and Horst von Brand for pointing out the errors of
  195. my ways.
  196. - Fix various bugs in the runlist merging code. (Based on libntfs
  197. changes by Richard Russon.)
  198. - Fix sparse warnings that have crept in over time.
  199. - Change ntfs_cluster_free() to require a write locked runlist on entry
  200. since we otherwise get into a lock reversal deadlock if a read locked
  201. runlist is passed in. In the process also change it to take an ntfs
  202. inode instead of a vfs inode as parameter.
  203. - Fix the definition of the CHKD ntfs record magic. It had an off by
  204. two error causing it to be CHKB instead of CHKD.
  205. - Fix a stupid bug in __ntfs_bitmap_set_bits_in_run() which caused the
  206. count to become negative and hence we had a wild memset() scribbling
  207. all over the system's ram.
  208. 2.1.23 - Implement extension of resident files and make writing safe as well as
  209. many bug fixes, cleanups, and enhancements...
  210. - Add printk rate limiting for ntfs_warning() and ntfs_error() when
  211. compiled without debug. This avoids a possible denial of service
  212. attack. Thanks to Carl-Daniel Hailfinger from SuSE for pointing this
  213. out.
  214. - Fix compilation warnings on ia64. (Randy Dunlap)
  215. - Use i_size_{read,write}() instead of reading i_size by hand and cache
  216. the value where apropriate.
  217. - Add size_lock to the ntfs_inode structure. This is an rw spinlock
  218. and it locks against access to the inode sizes. Note, ->size_lock
  219. is also accessed from irq context so you must use the _irqsave and
  220. _irqrestore lock and unlock functions, respectively. Protect all
  221. accesses to allocated_size, initialized_size, and compressed_size.
  222. - Minor optimization to fs/ntfs/super.c::ntfs_statfs() and its helpers.
  223. - Implement extension of resident files in the regular file write code
  224. paths (fs/ntfs/aops.c::ntfs_{prepare,commit}_write()). At present
  225. this only works until the data attribute becomes too big for the mft
  226. record after which we abort the write returning -EOPNOTSUPP from
  227. ntfs_prepare_write().
  228. - Add disable_sparse mount option together with a per volume sparse
  229. enable bit which is set appropriately and a per inode sparse disable
  230. bit which is preset on some system file inodes as appropriate.
  231. - Enforce that sparse support is disabled on NTFS volumes pre 3.0.
  232. - Fix a bug in fs/ntfs/runlist.c::ntfs_mapping_pairs_decompress() in
  233. the creation of the unmapped runlist element for the base attribute
  234. extent.
  235. - Split ntfs_map_runlist() into ntfs_map_runlist() and a non-locking
  236. helper ntfs_map_runlist_nolock() which is used by ntfs_map_runlist().
  237. This allows us to map runlist fragments with the runlist lock already
  238. held without having to drop and reacquire it around the call. Adapt
  239. all callers.
  240. - Change ntfs_find_vcn() to ntfs_find_vcn_nolock() which takes a locked
  241. runlist. This allows us to find runlist elements with the runlist
  242. lock already held without having to drop and reacquire it around the
  243. call. Adapt all callers.
  244. - Change time to u64 in time.h::ntfs2utc() as it otherwise generates a
  245. warning in the do_div() call on sparc32. Thanks to Meelis Roos for
  246. the report and analysis of the warning.
  247. - Fix a nasty runlist merge bug when merging two holes.
  248. - Set the ntfs_inode->allocated_size to the real allocated size in the
  249. mft record for resident attributes (fs/ntfs/inode.c).
  250. - Small readability cleanup to use "a" instead of "ctx->attr"
  251. everywhere (fs/ntfs/inode.c).
  252. - Make fs/ntfs/namei.c::ntfs_get_{parent,dentry} static and move the
  253. definition of ntfs_export_ops from fs/ntfs/super.c to namei.c. Also,
  254. declare ntfs_export_ops in fs/ntfs/ntfs.h.
  255. - Correct sparse file handling. The compressed values need to be
  256. checked and set in the ntfs inode as done for compressed files and
  257. the compressed size needs to be used for vfs inode->i_blocks instead
  258. of the allocated size, again, as done for compressed files.
  259. - Add AT_EA in addition to AT_DATA to whitelist for being allowed to be
  260. non-resident in fs/ntfs/attrib.c::ntfs_attr_can_be_non_resident().
  261. - Add fs/ntfs/attrib.c::ntfs_attr_vcn_to_lcn_nolock() used by the new
  262. write code.
  263. - Fix bug in fs/ntfs/attrib.c::ntfs_find_vcn_nolock() where after
  264. dropping the read lock and taking the write lock we were not checking
  265. whether someone else did not already do the work we wanted to do.
  266. - Rename fs/ntfs/attrib.c::ntfs_find_vcn_nolock() to
  267. ntfs_attr_find_vcn_nolock() and update all callers.
  268. - Add fs/ntfs/attrib.[hc]::ntfs_attr_make_non_resident().
  269. - Fix sign of various error return values to be negative in
  270. fs/ntfs/lcnalloc.c.
  271. - Modify ->readpage and ->writepage (fs/ntfs/aops.c) so they detect and
  272. handle the case where an attribute is converted from resident to
  273. non-resident by a concurrent file write.
  274. - Remove checks for NULL before calling kfree() since kfree() does the
  275. checking itself. (Jesper Juhl)
  276. - Some utilities modify the boot sector but do not update the checksum.
  277. Thus, relax the checking in fs/ntfs/super.c::is_boot_sector_ntfs() to
  278. only emit a warning when the checksum is incorrect rather than
  279. refusing the mount. Thanks to Bernd Casimir for pointing this
  280. problem out.
  281. - Update attribute definition handling.
  282. - Add NTFS_MAX_CLUSTER_SIZE and NTFS_MAX_PAGES_PER_CLUSTER constants.
  283. - Use NTFS_MAX_CLUSTER_SIZE in super.c instead of hard coding 0x10000.
  284. - Use MAX_BUF_PER_PAGE instead of variable sized array allocation for
  285. better code generation and one less sparse warning in fs/ntfs/aops.c.
  286. - Remove spurious void pointer casts from fs/ntfs/. (Pekka Enberg)
  287. - Use C99 style structure initialization after memory allocation where
  288. possible (fs/ntfs/{attrib.c,index.c,super.c}). Thanks to Al Viro and
  289. Pekka Enberg.
  290. - Stamp the transaction log ($UsnJrnl), aka user space journal, if it
  291. is active on the volume and we are mounting read-write or remounting
  292. from read-only to read-write.
  293. - Fix a bug in address space operations error recovery code paths where
  294. if the runlist was not mapped at all and a mapping error occured we
  295. would leave the runlist locked on exit to the function so that the
  296. next access to the same file would try to take the lock and deadlock.
  297. - Detect the case when Windows has been suspended to disk on the volume
  298. to be mounted and if this is the case do not allow (re)mounting
  299. read-write. This is done by parsing hiberfil.sys if present.
  300. - Fix several occurences of a bug where we would perform 'var & ~const'
  301. with a 64-bit variable and a int, i.e. 32-bit, constant. This causes
  302. the higher order 32-bits of the 64-bit variable to be zeroed. To fix
  303. this cast the 'const' to the same 64-bit type as 'var'.
  304. - Change the runlist terminator of the newly allocated cluster(s) to
  305. LCN_ENOENT in ntfs_attr_make_non_resident(). Otherwise the runlist
  306. code gets confused.
  307. - Add an extra parameter @last_vcn to ntfs_get_size_for_mapping_pairs()
  308. and ntfs_mapping_pairs_build() to allow the runlist encoding to be
  309. partial which is desirable when filling holes in sparse attributes.
  310. Update all callers.
  311. - Change ntfs_map_runlist_nolock() to only decompress the mapping pairs
  312. if the requested vcn is inside it. Otherwise we get into problems
  313. when we try to map an out of bounds vcn because we then try to map
  314. the already mapped runlist fragment which causes
  315. ntfs_mapping_pairs_decompress() to fail and return error. Update
  316. ntfs_attr_find_vcn_nolock() accordingly.
  317. - Fix a nasty deadlock that appeared in recent kernels.
  318. The situation: VFS inode X on a mounted ntfs volume is dirty. For
  319. same inode X, the ntfs_inode is dirty and thus corresponding on-disk
  320. inode, i.e. mft record, which is in a dirty PAGE_CACHE_PAGE belonging
  321. to the table of inodes, i.e. $MFT, inode 0.
  322. What happens:
  323. Process 1: sys_sync()/umount()/whatever... calls
  324. __sync_single_inode() for $MFT -> do_writepages() -> write_page for
  325. the dirty page containing the on-disk inode X, the page is now locked
  326. -> ntfs_write_mst_block() which clears PageUptodate() on the page to
  327. prevent anyone else getting hold of it whilst it does the write out.
  328. This is necessary as the on-disk inode needs "fixups" applied before
  329. the write to disk which are removed again after the write and
  330. PageUptodate is then set again. It then analyses the page looking
  331. for dirty on-disk inodes and when it finds one it calls
  332. ntfs_may_write_mft_record() to see if it is safe to write this
  333. on-disk inode. This then calls ilookup5() to check if the
  334. corresponding VFS inode is in icache(). This in turn calls ifind()
  335. which waits on the inode lock via wait_on_inode whilst holding the
  336. global inode_lock.
  337. Process 2: pdflush results in a call to __sync_single_inode for the
  338. same VFS inode X on the ntfs volume. This locks the inode (I_LOCK)
  339. then calls write-inode -> ntfs_write_inode -> map_mft_record() ->
  340. read_cache_page() for the page (in page cache of table of inodes
  341. $MFT, inode 0) containing the on-disk inode. This page has
  342. PageUptodate() clear because of Process 1 (see above) so
  343. read_cache_page() blocks when it tries to take the page lock for the
  344. page so it can call ntfs_read_page().
  345. Thus Process 1 is holding the page lock on the page containing the
  346. on-disk inode X and it is waiting on the inode X to be unlocked in
  347. ifind() so it can write the page out and then unlock the page.
  348. And Process 2 is holding the inode lock on inode X and is waiting for
  349. the page to be unlocked so it can call ntfs_readpage() or discover
  350. that Process 1 set PageUptodate() again and use the page.
  351. Thus we have a deadlock due to ifind() waiting on the inode lock.
  352. The solution: The fix is to use the newly introduced
  353. ilookup5_nowait() which does not wait on the inode's lock and hence
  354. avoids the deadlock. This is safe as we do not care about the VFS
  355. inode and only use the fact that it is in the VFS inode cache and the
  356. fact that the vfs and ntfs inodes are one struct in memory to find
  357. the ntfs inode in memory if present. Also, the ntfs inode has its
  358. own locking so it does not matter if the vfs inode is locked.
  359. - Fix bug in mft record writing where we forgot to set the device in
  360. the buffers when mapping them after the VM had discarded them.
  361. Thanks to Martin MOKREJŠ for the bug report.
  362. 2.1.22 - Many bug and race fixes and error handling improvements.
  363. - Improve error handling in fs/ntfs/inode.c::ntfs_truncate().
  364. - Change fs/ntfs/inode.c::ntfs_truncate() to return an error code
  365. instead of void and provide a helper ntfs_truncate_vfs() for the
  366. vfs ->truncate method.
  367. - Add a new ntfs inode flag NInoTruncateFailed() and modify
  368. fs/ntfs/inode.c::ntfs_truncate() to set and clear it appropriately.
  369. - Fix min_size and max_size definitions in ATTR_DEF structure in
  370. fs/ntfs/layout.h to be signed.
  371. - Add attribute definition handling helpers to fs/ntfs/attrib.[hc]:
  372. ntfs_attr_size_bounds_check(), ntfs_attr_can_be_non_resident(), and
  373. ntfs_attr_can_be_resident(), which in turn use the new private helper
  374. ntfs_attr_find_in_attrdef().
  375. - In fs/ntfs/aops.c::mark_ntfs_record_dirty(), take the
  376. mapping->private_lock around the dirtying of the buffer heads
  377. analagous to the way it is done in __set_page_dirty_buffers().
  378. - Ensure the mft record size does not exceed the PAGE_CACHE_SIZE at
  379. mount time as this cannot work with the current implementation.
  380. - Check for location of attribute name and improve error handling in
  381. general in fs/ntfs/inode.c::ntfs_read_locked_inode() and friends.
  382. - In fs/ntfs/aops.c::ntfs_writepage(), if the page is fully outside
  383. i_size, i.e. race with truncate, invalidate the buffers on the page
  384. so that they become freeable and hence the page does not leak.
  385. - Remove unused function fs/ntfs/runlist.c::ntfs_rl_merge(). (Adrian
  386. Bunk)
  387. - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_find() that resulted in
  388. a NULL pointer dereference in the error code path when a corrupt
  389. attribute was found. (Thanks to Domen Puncer for the bug report.)
  390. - Add MODULE_VERSION() to fs/ntfs/super.c.
  391. - Make several functions and variables static. (Adrian Bunk)
  392. - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() so it allocates
  393. buffers for the page if they are not present and then marks the
  394. buffers belonging to the ntfs record dirty. This causes the buffers
  395. to become busy and hence they are safe from removal until the page
  396. has been written out.
  397. - Fix stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find() in the
  398. error handling code path that resulted in a BUG() due to trying to
  399. unmap an extent mft record when the mapping of it had failed and it
  400. thus was not mapped. (Thanks to Ken MacFerrin for the bug report.)
  401. - Drop the runlist lock after the vcn has been read in
  402. fs/ntfs/lcnalloc.c::__ntfs_cluster_free().
  403. - Rewrite handling of multi sector transfer errors. We now do not set
  404. PageError() when such errors are detected in the async i/o handler
  405. fs/ntfs/aops.c::ntfs_end_buffer_async_read(). All users of mst
  406. protected attributes now check the magic of each ntfs record as they
  407. use it and act appropriately. This has the effect of making errors
  408. granular per ntfs record rather than per page which solves the case
  409. where we cannot access any of the ntfs records in a page when a
  410. single one of them had an mst error. (Thanks to Ken MacFerrin for
  411. the bug report.)
  412. - Fix error handling in fs/ntfs/quota.c::ntfs_mark_quotas_out_of_date()
  413. where we failed to release i_mutex on the $Quota/$Q attribute inode.
  414. - Fix bug in handling of bad inodes in fs/ntfs/namei.c::ntfs_lookup().
  415. - Add mapping of unmapped buffers to all remaining code paths, i.e.
  416. fs/ntfs/aops.c::ntfs_write_mst_block(), mft.c::ntfs_sync_mft_mirror(),
  417. and write_mft_record_nolock(). From now on we require that the
  418. complete runlist for the mft mirror is always mapped into memory.
  419. - Add creation of buffers to fs/ntfs/mft.c::ntfs_sync_mft_mirror().
  420. - Improve error handling in fs/ntfs/aops.c::ntfs_{read,write}_block().
  421. - Cleanup fs/ntfs/aops.c::ntfs_{read,write}page() since we know that a
  422. resident attribute will be smaller than a page which makes the code
  423. simpler. Also make the code more tolerant to concurrent ->truncate.
  424. 2.1.21 - Fix some races and bugs, rewrite mft write code, add mft allocator.
  425. - Implement extent mft record deallocation
  426. fs/ntfs/mft.c::ntfs_extent_mft_record_free().
  427. - Splitt runlist related functions off from attrib.[hc] to runlist.[hc].
  428. - Add vol->mft_data_pos and initialize it at mount time.
  429. - Rename init_runlist() to ntfs_init_runlist(), ntfs_vcn_to_lcn() to
  430. ntfs_rl_vcn_to_lcn(), decompress_mapping_pairs() to
  431. ntfs_mapping_pairs_decompress(), ntfs_merge_runlists() to
  432. ntfs_runlists_merge() and adapt all callers.
  433. - Add fs/ntfs/runlist.[hc]::ntfs_get_nr_significant_bytes(),
  434. ntfs_get_size_for_mapping_pairs(), ntfs_write_significant_bytes(),
  435. and ntfs_mapping_pairs_build(), adapted from libntfs.
  436. - Make fs/ntfs/lcnalloc.c::ntfs_cluster_free_from_rl_nolock() not
  437. static and add a declaration for it to lcnalloc.h.
  438. - Add fs/ntfs/lcnalloc.h::ntfs_cluster_free_from_rl() which is a static
  439. inline wrapper for ntfs_cluster_free_from_rl_nolock() which takes the
  440. cluster bitmap lock for the duration of the call.
  441. - Add fs/ntfs/attrib.[hc]::ntfs_attr_record_resize().
  442. - Implement the equivalent of memset() for an ntfs attribute in
  443. fs/ntfs/attrib.[hc]::ntfs_attr_set() and switch
  444. fs/ntfs/logfile.c::ntfs_empty_logfile() to using it.
  445. - Remove unnecessary casts from LCN_* constants.
  446. - Implement fs/ntfs/runlist.c::ntfs_rl_truncate_nolock().
  447. - Add MFT_RECORD_OLD as a copy of MFT_RECORD in fs/ntfs/layout.h and
  448. change MFT_RECORD to contain the NTFS 3.1+ specific fields.
  449. - Add a helper function fs/ntfs/aops.c::mark_ntfs_record_dirty() which
  450. marks all buffers belonging to an ntfs record dirty, followed by
  451. marking the page the ntfs record is in dirty and also marking the vfs
  452. inode containing the ntfs record dirty (I_DIRTY_PAGES).
  453. - Switch fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to using the
  454. new helper fs/ntfs/aops.c::mark_ntfs_record_dirty() and remove the no
  455. longer needed fs/ntfs/index.[hc]::__ntfs_index_entry_mark_dirty().
  456. - Move ntfs_{un,}map_page() from ntfs.h to aops.h and fix resulting
  457. include errors.
  458. - Move the typedefs for runlist_element and runlist from types.h to
  459. runlist.h and fix resulting include errors.
  460. - Remove unused {__,}format_mft_record() from fs/ntfs/mft.c.
  461. - Modify fs/ntfs/mft.c::__mark_mft_record_dirty() to use the helper
  462. mark_ntfs_record_dirty() which also changes the behaviour in that we
  463. now set the buffers belonging to the mft record dirty as well as the
  464. page itself.
  465. - Update fs/ntfs/mft.c::write_mft_record_nolock() and sync_mft_mirror()
  466. to cope with the fact that there now are dirty buffers in mft pages.
  467. - Update fs/ntfs/inode.c::ntfs_write_inode() to also use the helper
  468. mark_ntfs_record_dirty() and thus to set the buffers belonging to the
  469. mft record dirty as well as the page itself.
  470. - Fix compiler warnings on x86-64 in fs/ntfs/dir.c. (Randy Dunlap,
  471. slightly modified by me)
  472. - Add fs/ntfs/mft.c::try_map_mft_record() which fails with -EALREADY if
  473. the mft record is already locked and otherwise behaves the same way
  474. as fs/ntfs/mft.c::map_mft_record().
  475. - Modify fs/ntfs/mft.c::write_mft_record_nolock() so that it only
  476. writes the mft record if the buffers belonging to it are dirty.
  477. Otherwise we assume that it was written out by other means already.
  478. - Attempting to write outside initialized size is _not_ a bug so remove
  479. the bug check from fs/ntfs/aops.c::ntfs_write_mst_block(). It is in
  480. fact required to write outside initialized size when preparing to
  481. extend the initialized size.
  482. - Map the page instead of using page_address() before writing to it in
  483. fs/ntfs/aops.c::ntfs_mft_writepage().
  484. - Provide exclusion between opening an inode / mapping an mft record
  485. and accessing the mft record in fs/ntfs/mft.c::ntfs_mft_writepage()
  486. by setting the page not uptodate throughout ntfs_mft_writepage().
  487. - Clear the page uptodate flag in fs/ntfs/aops.c::ntfs_write_mst_block()
  488. to ensure noone can see the page whilst the mst fixups are applied.
  489. - Add the helper fs/ntfs/mft.c::ntfs_may_write_mft_record() which
  490. checks if an mft record may be written out safely obtaining any
  491. necessary locks in the process. This is used by
  492. fs/ntfs/aops.c::ntfs_write_mst_block().
  493. - Modify fs/ntfs/aops.c::ntfs_write_mst_block() to also work for
  494. writing mft records and improve its error handling in the process.
  495. Now if any of the records in the page fail to be written out, all
  496. other records will be written out instead of aborting completely.
  497. - Remove ntfs_mft_aops and update all users to use ntfs_mst_aops.
  498. - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to set the
  499. ntfs_mst_aops for all inodes which are NInoMstProtected() and
  500. ntfs_aops for all other inodes.
  501. - Rename fs/ntfs/mft.c::sync_mft_mirror{,_umount}() to
  502. ntfs_sync_mft_mirror{,_umount}() and change their parameters so they
  503. no longer require an ntfs inode to be present. Update all callers.
  504. - Cleanup the error handling in fs/ntfs/mft.c::ntfs_sync_mft_mirror().
  505. - Clear the page uptodate flag in fs/ntfs/mft.c::ntfs_sync_mft_mirror()
  506. to ensure noone can see the page whilst the mst fixups are applied.
  507. - Remove the no longer needed fs/ntfs/mft.c::ntfs_mft_writepage() and
  508. fs/ntfs/mft.c::try_map_mft_record().
  509. - Fix callers of fs/ntfs/aops.c::mark_ntfs_record_dirty() to call it
  510. with the ntfs inode which contains the page rather than the ntfs
  511. inode the mft record of which is in the page.
  512. - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by moving the
  513. index inode bitmap inode release code from there to
  514. fs/ntfs/inode.c::ntfs_clear_big_inode(). (Thanks to Christoph
  515. Hellwig for spotting this.)
  516. - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by taking the
  517. inode semaphore around the code that sets ni->itype.index.bmp_ino to
  518. NULL and reorganize the code to optimize it a bit. (Thanks to
  519. Christoph Hellwig for spotting this.)
  520. - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() to no longer take the
  521. ntfs inode as a parameter as this is confusing and misleading and the
  522. needed ntfs inode is available via NTFS_I(page->mapping->host).
  523. Adapt all callers to this change.
  524. - Modify fs/ntfs/mft.c::write_mft_record_nolock() and
  525. fs/ntfs/aops.c::ntfs_write_mst_block() to only check the dirty state
  526. of the first buffer in a record and to take this as the ntfs record
  527. dirty state. We cannot look at the dirty state for subsequent
  528. buffers because we might be racing with
  529. fs/ntfs/aops.c::mark_ntfs_record_dirty().
  530. - Move the static inline ntfs_init_big_inode() from fs/ntfs/inode.c to
  531. inode.h and make fs/ntfs/inode.c::__ntfs_init_inode() non-static and
  532. add a declaration for it to inode.h. Fix some compilation issues
  533. that resulted due to #includes and header file interdependencies.
  534. - Simplify setup of i_mode in fs/ntfs/inode.c::ntfs_read_locked_inode().
  535. - Add helpers fs/ntfs/layout.h::MK_MREF() and MK_LE_MREF().
  536. - Modify fs/ntfs/mft.c::map_extent_mft_record() to only verify the mft
  537. record sequence number if it is specified (i.e. not zero).
  538. - Add fs/ntfs/mft.[hc]::ntfs_mft_record_alloc() and various helper
  539. functions used by it.
  540. - Update Documentation/filesystems/ntfs.txt with instructions on how to
  541. use the Device-Mapper driver with NTFS ftdisk/LDM raid. This removes
  542. the linear raid problem with the Software RAID / MD driver when one
  543. or more of the devices has an odd number of sectors.
  544. 2.1.20 - Fix two stupid bugs introduced in 2.1.18 release.
  545. - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_reinit_search_ctx()
  546. where we did not clear ctx->al_entry but it was still set due to
  547. changes in ntfs_attr_lookup() and ntfs_external_attr_find() in
  548. particular.
  549. - Fix another stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find()
  550. where we forgot to unmap the extent mft record when we had finished
  551. enumerating an attribute which caused a bug check to trigger when the
  552. VFS calls ->clear_inode.
  553. 2.1.19 - Many cleanups, improvements, and a minor bug fix.
  554. - Update ->setattr (fs/ntfs/inode.c::ntfs_setattr()) to refuse to
  555. change the uid, gid, and mode of an inode as we do not support NTFS
  556. ACLs yet.
  557. - Remove BKL use from ntfs_setattr() syncing up with the rest of the
  558. kernel.
  559. - Get rid of the ugly transparent union in fs/ntfs/dir.c::ntfs_readdir()
  560. and ntfs_filldir() as per suggestion from Al Viro.
  561. - Change '\0' and L'\0' to simply 0 as per advice from Linus Torvalds.
  562. - Update ->truncate (fs/ntfs/inode.c::ntfs_truncate()) to check if the
  563. inode size has changed and to only output an error if so.
  564. - Rename fs/ntfs/attrib.h::attribute_value_length() to ntfs_attr_size().
  565. - Add le{16,32,64} as well as sle{16,32,64} data types to
  566. fs/ntfs/types.h.
  567. - Change ntfschar to be le16 instead of u16 in fs/ntfs/types.h.
  568. - Add le versions of VCN, LCN, and LSN called leVCN, leLCN, and leLSN,
  569. respectively, to fs/ntfs/types.h.
  570. - Update endianness conversion macros in fs/ntfs/endian.h to use the
  571. new types as appropriate.
  572. - Do proper type casting when using sle64_to_cpup() in fs/ntfs/dir.c
  573. and index.c.
  574. - Add leMFT_REF data type to fs/ntfs/layout.h.
  575. - Update all NTFS header files with the new little endian data types.
  576. Affected files are fs/ntfs/layout.h, logfile.h, and time.h.
  577. - Do proper type casting when using ntfs_is_*_recordp() in
  578. fs/ntfs/logfile.c, mft.c, and super.c.
  579. - Fix all the sparse bitwise warnings. Had to change all the typedef
  580. enums storing little endian values to simple enums plus a typedef for
  581. the datatype to make sparse happy.
  582. - Fix a bug found by the new sparse bitwise warnings where the default
  583. upcase table was defined as a pointer to wchar_t rather than ntfschar
  584. in fs/ntfs/ntfs.h and super.c.
  585. - Change {const_,}cpu_to_le{16,32}(0) to just 0 as suggested by Al Viro.
  586. 2.1.18 - Fix scheduling latencies at mount time as well as an endianness bug.
  587. - Remove vol->nr_mft_records as it was pretty meaningless and optimize
  588. the calculation of total/free inodes as used by statfs().
  589. - Fix scheduling latencies in ntfs_fill_super() by dropping the BKL
  590. because the code itself is using the ntfs_lock semaphore which
  591. provides safe locking. (Ingo Molnar)
  592. - Fix a potential bug in fs/ntfs/mft.c::map_extent_mft_record() that
  593. could occur in the future for when we start closing/freeing extent
  594. inodes if we don't set base_ni->ext.extent_ntfs_inos to NULL after
  595. we free it.
  596. - Rename {find,lookup}_attr() to ntfs_attr_{find,lookup}() as well as
  597. find_external_attr() to ntfs_external_attr_find() to cleanup the
  598. namespace a bit and to be more consistent with libntfs.
  599. - Rename {{re,}init,get,put}_attr_search_ctx() to
  600. ntfs_attr_{{re,}init,get,put}_search_ctx() as well as the type
  601. attr_search_context to ntfs_attr_search_ctx.
  602. - Force use of ntfs_attr_find() in ntfs_attr_lookup() when searching
  603. for the attribute list attribute itself.
  604. - Fix endianness bug in ntfs_external_attr_find().
  605. - Change ntfs_{external_,}attr_find() to return 0 on success, -ENOENT
  606. if the attribute is not found, and -EIO on real error. In the case
  607. of -ENOENT, the search context is updated to describe the attribute
  608. before which the attribute being searched for would need to be
  609. inserted if such an action were to be desired and in the case of
  610. ntfs_external_attr_find() the search context is also updated to
  611. indicate the attribute list entry before which the attribute list
  612. entry of the attribute being searched for would need to be inserted
  613. if such an action were to be desired. Also make ntfs_find_attr()
  614. static and remove its prototype from attrib.h as it is not used
  615. anywhere other than attrib.c. Update ntfs_attr_lookup() and all
  616. callers of ntfs_{external,}attr_{find,lookup}() for the new return
  617. values.
  618. - Minor cleanup of fs/ntfs/inode.c::ntfs_init_locked_inode().
  619. 2.1.17 - Fix bugs in mount time error code paths and other updates.
  620. - Implement bitmap modification code (fs/ntfs/bitmap.[hc]). This
  621. includes functions to set/clear a single bit or a run of bits.
  622. - Add fs/ntfs/attrib.[hc]::ntfs_find_vcn() which returns the locked
  623. runlist element containing a particular vcn. It also takes care of
  624. mapping any needed runlist fragments.
  625. - Implement cluster (de-)allocation code (fs/ntfs/lcnalloc.[hc]).
  626. - Load attribute definition table from $AttrDef at mount time.
  627. - Fix bugs in mount time error code paths involving (de)allocation of
  628. the default and volume upcase tables.
  629. - Remove ntfs_nr_mounts as it is no longer used.
  630. 2.1.16 - Implement access time updates, file sync, async io, and read/writev.
  631. - Add support for readv/writev and aio_read/aio_write (fs/ntfs/file.c).
  632. This is done by setting the appropriate file operations pointers to
  633. the generic helper functions provided by mm/filemap.c.
  634. - Implement fsync, fdatasync, and msync both for files (fs/ntfs/file.c)
  635. and directories (fs/ntfs/dir.c).
  636. - Add support for {a,m,c}time updates to inode.c::ntfs_write_inode().
  637. Note, except for the root directory and any other system files opened
  638. by the user, the system files will not have their access times
  639. updated as they are only accessed at the inode level an hence the
  640. file level functions which cause the times to be updated are never
  641. invoked.
  642. 2.1.15 - Invalidate quotas when (re)mounting read-write.
  643. - Add new element itype.index.collation_rule to the ntfs inode
  644. structure and set it appropriately in ntfs_read_locked_inode().
  645. - Implement a new inode type "index" to allow efficient access to the
  646. indices found in various system files and adapt inode handling
  647. accordingly (fs/ntfs/inode.[hc]). An index inode is essentially an
  648. attribute inode (NInoAttr() is true) with an attribute type of
  649. AT_INDEX_ALLOCATION. As such, it is no longer allowed to call
  650. ntfs_attr_iget() with an attribute type of AT_INDEX_ALLOCATION as
  651. there would be no way to distinguish between normal attribute inodes
  652. and index inodes. The function to obtain an index inode is
  653. ntfs_index_iget() and it uses the helper function
  654. ntfs_read_locked_index_inode(). Note, we do not overload
  655. ntfs_attr_iget() as indices consist of multiple attributes so using
  656. ntfs_attr_iget() to obtain an index inode would be confusing.
  657. - Ensure that there is no overflow when doing page->index <<
  658. PAGE_CACHE_SHIFT by casting page->index to s64 in fs/ntfs/aops.c.
  659. - Use atomic kmap instead of kmap() in fs/ntfs/aops.c::ntfs_read_page()
  660. and ntfs_read_block().
  661. - Use case sensitive attribute lookups instead of case insensitive ones.
  662. - Lock all page cache pages belonging to mst protected attributes while
  663. accessing them to ensure we never see corrupt data while the page is
  664. under writeout.
  665. - Add framework for generic ntfs collation (fs/ntfs/collation.[hc]).
  666. We have ntfs_is_collation_rule_supported() to check if the collation
  667. rule you want to use is supported and ntfs_collation() which actually
  668. collates two data items. We currently only support COLLATION_BINARY
  669. and COLLATION_NTOFS_ULONG but support for other collation rules will
  670. be added as the need arises.
  671. - Add a new type, ntfs_index_context, to allow retrieval of an index
  672. entry using the corresponding index key. To get an index context,
  673. use ntfs_index_ctx_get() and to release it, use ntfs_index_ctx_put().
  674. This also adds a new slab cache for the index contexts. To lookup a
  675. key in an index inode, use ntfs_index_lookup(). After modifying an
  676. index entry, call ntfs_index_entry_flush_dcache_page() followed by
  677. ntfs_index_entry_mark_dirty() to ensure the changes are written out
  678. to disk. For details see fs/ntfs/index.[hc]. Note, at present, if
  679. an index entry is in the index allocation attribute rather than the
  680. index root attribute it will not be written out (you will get a
  681. warning message about discarded changes instead).
  682. - Load the quota file ($Quota) and check if quota tracking is enabled
  683. and if so, mark the quotas out of date. This causes windows to
  684. rescan the volume on boot and update all quota entries.
  685. - Add a set_page_dirty address space operation for ntfs_m[fs]t_aops.
  686. It is simply set to __set_page_dirty_nobuffers() to make sure that
  687. running set_page_dirty() on a page containing mft/ntfs records will
  688. not affect the dirty state of the page buffers.
  689. - Add fs/ntfs/index.c::__ntfs_index_entry_mark_dirty() which sets all
  690. buffers that are inside the ntfs record in the page dirty after which
  691. it sets the page dirty. This allows ->writepage to only write the
  692. dirty index records rather than having to write all the records in
  693. the page. Modify fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to
  694. use this rather than __set_page_dirty_nobuffers().
  695. - Implement fs/ntfs/aops.c::ntfs_write_mst_block() which enables the
  696. writing of page cache pages belonging to mst protected attributes
  697. like the index allocation attribute in directory indices and other
  698. indices like $Quota/$Q, etc. This means that the quota is now marked
  699. out of date on all volumes rather than only on ones where the quota
  700. defaults entry is in the index root attribute of the $Quota/$Q index.
  701. 2.1.14 - Fix an NFSd caused deadlock reported by several users.
  702. - Modify fs/ntfs/ntfs_readdir() to copy the index root attribute value
  703. to a buffer so that we can put the search context and unmap the mft
  704. record before calling the filldir() callback. We need to do this
  705. because of NFSd which calls ->lookup() from its filldir callback()
  706. and this causes NTFS to deadlock as ntfs_lookup() maps the mft record
  707. of the directory and since ntfs_readdir() has got it mapped already
  708. ntfs_lookup() deadlocks.
  709. 2.1.13 - Enable overwriting of resident files and housekeeping of system files.
  710. - Implement writing of mft records (fs/ntfs/mft.[hc]), which includes
  711. keeping the mft mirror in sync with the mft when mirrored mft records
  712. are written. The functions are write_mft_record{,_nolock}(). The
  713. implementation is quite rudimentary for now with lots of things not
  714. implemented yet but I am not sure any of them can actually occur so
  715. I will wait for people to hit each one and only then implement it.
  716. - Commit open system inodes at umount time. This should make it
  717. virtually impossible for sync_mft_mirror_umount() to ever be needed.
  718. - Implement ->write_inode (fs/ntfs/inode.c::ntfs_write_inode()) for the
  719. ntfs super operations. This gives us inode writing via the VFS inode
  720. dirty code paths. Note: Access time updates are not implemented yet.
  721. - Implement fs/ntfs/mft.[hc]::{,__}mark_mft_record_dirty() and make
  722. fs/ntfs/aops.c::ntfs_writepage() and ntfs_commit_write() use it, thus
  723. finally enabling resident file overwrite! (-8 This also includes a
  724. placeholder for ->writepage (ntfs_mft_writepage()), which for now
  725. just redirties the page and returns. Also, at umount time, we for
  726. now throw away all mft data page cache pages after the last call to
  727. ntfs_commit_inode() in the hope that all inodes will have been
  728. written out by then and hence no dirty (meta)data will be lost. We
  729. also check for this case and emit an error message telling the user
  730. to run chkdsk.
  731. - Use set_page_writeback() and end_page_writeback() in the resident
  732. attribute code path of fs/ntfs/aops.c::ntfs_writepage() otherwise
  733. the radix-tree tag PAGECACHE_TAG_DIRTY remains set even though the
  734. page is clean.
  735. - Implement ntfs_mft_writepage() so it now checks if any of the mft
  736. records in the page are dirty and if so redirties the page and
  737. returns. Otherwise it just returns (after doing set_page_writeback(),
  738. unlock_page(), end_page_writeback() or the radix-tree tag
  739. PAGECACHE_TAG_DIRTY remains set even though the page is clean), thus
  740. alowing the VM to do with the page as it pleases. Also, at umount
  741. time, now only throw away dirty mft (meta)data pages if dirty inodes
  742. are present and ask the user to email us if they see this happening.
  743. - Add functions ntfs_{clear,set}_volume_flags(), to modify the volume
  744. information flags (fs/ntfs/super.c).
  745. - Mark the volume dirty when (re)mounting read-write and mark it clean
  746. when unmounting or remounting read-only. If any volume errors are
  747. found, the volume is left marked dirty to force chkdsk to run.
  748. - Add code to set the NT4 compatibility flag when (re)mounting
  749. read-write for newer NTFS versions but leave it commented out for now
  750. since we do not make any modifications that are NTFS 1.2 specific yet
  751. and since setting this flag breaks Captive-NTFS which is not nice.
  752. This code must be enabled once we start writing NTFS 1.2 specific
  753. changes otherwise Windows NTFS driver might crash / cause corruption.
  754. 2.1.12 - Fix the second fix to the decompression engine and some cleanups.
  755. - Add a new address space operations struct, ntfs_mst_aops, for mst
  756. protected attributes. This is because the default ntfs_aops do not
  757. make sense with mst protected data and were they to write anything to
  758. such an attribute they would cause data corruption so we provide
  759. ntfs_mst_aops which does not have any write related operations set.
  760. - Cleanup dirty ntfs inode handling (fs/ntfs/inode.[hc]) which also
  761. includes an adapted ntfs_commit_inode() and an implementation of
  762. ntfs_write_inode() which for now just cleans dirty inodes without
  763. writing them (it does emit a warning that this is happening).
  764. - Undo the second decompression engine fix (see 2.1.9 release ChangeLog
  765. entry) as it was only fixing a theoretical bug but at the same time
  766. it badly broke the handling of sparse and uncompressed compression
  767. blocks.
  768. 2.1.11 - Driver internal cleanups.
  769. - Only build logfile.o if building the driver with read-write support.
  770. - Really final white space cleanups.
  771. - Use generic_ffs() instead of ffs() in logfile.c which allows the
  772. log_page_size variable to be optimized by gcc into a constant.
  773. - Rename uchar_t to ntfschar everywhere as uchar_t is unsigned 1-byte
  774. char as defined by POSIX and as found on some systems.
  775. 2.1.10 - Force read-only (re)mounting of volumes with unsupported volume flags.
  776. - Finish off the white space cleanups (remove trailing spaces, etc).
  777. - Clean up ntfs_fill_super() and ntfs_read_inode_mount() by removing
  778. the kludges around the first iget(). Instead of (re)setting ->s_op
  779. we have the $MFT inode set up by explicit new_inode() / set ->i_ino /
  780. insert_inode_hash() / call ntfs_read_inode_mount() directly. This
  781. kills the need for second super_operations and allows to return error
  782. from ntfs_read_inode_mount() without resorting to ugly "poisoning"
  783. tricks. (Al Viro)
  784. - Force read-only (re)mounting if any of the following bits are set in
  785. the volume information flags:
  786. VOLUME_IS_DIRTY, VOLUME_RESIZE_LOG_FILE,
  787. VOLUME_UPGRADE_ON_MOUNT, VOLUME_DELETE_USN_UNDERWAY,
  788. VOLUME_REPAIR_OBJECT_ID, VOLUME_MODIFIED_BY_CHKDSK
  789. To make this easier we define VOLUME_MUST_MOUNT_RO_MASK with all the
  790. above bits set so the test is made easy.
  791. 2.1.9 - Fix two bugs in decompression engine.
  792. - Fix a bug where we would not always detect that we have reached the
  793. end of a compression block because we were ending at minus one byte
  794. which is effectively the same as being at the end. The fix is to
  795. check whether the uncompressed buffer has been fully filled and if so
  796. we assume we have reached the end of the compression block. A big
  797. thank you to Marcin Gibuła for the bug report, the assistance in
  798. tracking down the bug and testing the fix.
  799. - Fix a possible bug where when a compressed read is truncated to the
  800. end of the file, the offset inside the last page was not truncated.
  801. 2.1.8 - Handle $MFT mirror and $LogFile, improve time handling, and cleanups.
  802. - Use get_bh() instead of manual atomic_inc() in fs/ntfs/compress.c.
  803. - Modify fs/ntfs/time.c::ntfs2utc(), get_current_ntfs_time(), and
  804. utc2ntfs() to work with struct timespec instead of time_t on the
  805. Linux UTC time side thus preserving the full precision of the NTFS
  806. time and only loosing up to 99 nano-seconds in the Linux UTC time.
  807. - Move fs/ntfs/time.c to fs/ntfs/time.h and make the time functions
  808. static inline.
  809. - Remove unused ntfs_dirty_inode().
  810. - Cleanup super operations declaration in fs/ntfs/super.c.
  811. - Wrap flush_dcache_mft_record_page() in #ifdef NTFS_RW.
  812. - Add NInoTestSetFoo() and NInoTestClearFoo() macro magic to
  813. fs/ntfs/inode.h and use it to declare NInoTest{Set,Clear}Dirty.
  814. - Move typedefs for ntfs_attr and test_t from fs/ntfs/inode.c to
  815. fs/ntfs/inode.h so they can be used elsewhere.
  816. - Determine the mft mirror size as the number of mirrored mft records
  817. and store it in ntfs_volume->mftmirr_size (fs/ntfs/super.c).
  818. - Load the mft mirror at mount time and compare the mft records stored
  819. in it to the ones in the mft. Force a read-only mount if the two do
  820. not match (fs/ntfs/super.c).
  821. - Fix type casting related warnings on 64-bit architectures. Thanks
  822. to Meelis Roos for reporting them.
  823. - Move %L to %ll as %L is floating point and %ll is integer which is
  824. what we want.
  825. - Read the journal ($LogFile) and determine if the volume has been
  826. shutdown cleanly and force a read-only mount if not (fs/ntfs/super.c
  827. and fs/ntfs/logfile.c). This is a little bit of a crude check in
  828. that we only look at the restart areas and not at the actual log
  829. records so that there will be a very small number of cases where we
  830. think that a volume is dirty when in fact it is clean. This should
  831. only affect volumes that have not been shutdown cleanly and did not
  832. have any pending, non-check-pointed i/o.
  833. - If the $LogFile indicates a clean shutdown and a read-write (re)mount
  834. is requested, empty $LogFile by overwriting it with 0xff bytes to
  835. ensure that Windows cannot cause data corruption by replaying a stale
  836. journal after Linux has written to the volume.
  837. 2.1.7 - Enable NFS exporting of mounted NTFS volumes.
  838. - Set i_generation in the VFS inode from the seq_no of the NTFS inode.
  839. - Make ntfs_lookup() NFS export safe, i.e. use d_splice_alias(), etc.
  840. - Implement ->get_dentry() in fs/ntfs/namei.c::ntfs_get_dentry() as the
  841. default doesn't allow inode number 0 which is a valid inode on NTFS
  842. and even if it did allow that it uses iget() instead of ntfs_iget()
  843. which makes it useless for us.
  844. - Implement ->get_parent() in fs/ntfs/namei.c::ntfs_get_parent() as the
  845. default just returns -EACCES which is not very useful.
  846. - Define export operations (->s_export_op) for NTFS (ntfs_export_ops)
  847. and set them up in the super block at mount time (super.c) this
  848. allows mounted NTFS volumes to be exported via NFS.
  849. - Add missing return -EOPNOTSUPP; in
  850. fs/ntfs/aops.c::ntfs_commit_nonresident_write().
  851. - Enforce no atime and no dir atime updates at mount/remount time as
  852. they are not implemented yet anyway.
  853. - Move a few assignments in fs/ntfs/attrib.c::load_attribute_list() to
  854. after a NULL check. Thanks to Dave Jones for pointing this out.
  855. 2.1.6 - Fix minor bug in handling of compressed directories.
  856. - Fix bug in handling of compressed directories. A compressed
  857. directory is not really compressed so when we set the ->i_blocks
  858. field of a compressed directory inode we were setting it from the
  859. non-existing field ni->itype.compressed.size which gave random
  860. results... For directories we now always use ni->allocated_size.
  861. 2.1.5 - Fix minor bug in attribute list attribute handling.
  862. - Fix bug in attribute list handling. Actually it is not as much a bug
  863. as too much protection in that we were not allowing attribute lists
  864. which waste space on disk while Windows XP clearly allows it and in
  865. fact creates such attribute lists so our driver was failing.
  866. - Update NTFS documentation ready for 2.6 kernel release.
  867. 2.1.4 - Reduce compiler requirements.
  868. - Remove all uses of unnamed structs and unions in the driver to make
  869. old and newer gcc versions happy. Makes it a bit uglier IMO but at
  870. least people will stop hassling me about it.
  871. 2.1.3 - Important bug fixes in corner cases.
  872. - super.c::parse_ntfs_boot_sector(): Correct the check for 64-bit
  873. clusters. (Philipp Thomas)
  874. - attrib.c::load_attribute_list(): Fix bug when initialized_size is a
  875. multiple of the block_size but not the cluster size. (Szabolcs
  876. Szakacsits)
  877. 2.1.2 - Important bug fixes aleviating the hangs in statfs.
  878. - Fix buggy free cluster and free inode determination logic.
  879. 2.1.1 - Minor updates.
  880. - Add handling for initialized_size != data_size in compressed files.
  881. - Reduce function local stack usage from 0x3d4 bytes to just noise in
  882. fs/ntfs/upcase.c. (Randy Dunlap)
  883. - Remove compiler warnings for newer gcc.
  884. - Pages are no longer kmapped by mm/filemap.c::generic_file_write()
  885. around calls to ->{prepare,commit}_write. Adapt NTFS appropriately
  886. in fs/ntfs/aops.c::ntfs_prepare_nonresident_write() by using
  887. kmap_atomic(KM_USER0).
  888. 2.1.0 - First steps towards write support: implement file overwrite.
  889. - Add configuration option for developmental write support with an
  890. appropriately scary configuration help text.
  891. - Initial implementation of fs/ntfs/aops.c::ntfs_writepage() and its
  892. helper fs/ntfs/aops.c::ntfs_write_block(). This enables mmap(2) based
  893. overwriting of existing files on ntfs. Note: Resident files are
  894. only written into memory, and not written out to disk at present, so
  895. avoid writing to files smaller than about 1kiB.
  896. - Initial implementation of fs/ntfs/aops.c::ntfs_prepare_write(), its
  897. helper fs/ntfs/aops.c::ntfs_prepare_nonresident_write() and their
  898. counterparts, fs/ntfs/aops.c::ntfs_commit_write(), and
  899. fs/ntfs/aops.c::ntfs_commit_nonresident_write(), respectively. Also,
  900. add generic_file_write() to the ntfs file operations (fs/ntfs/file.c).
  901. This enables write(2) based overwriting of existing files on ntfs.
  902. Note: As with mmap(2) based overwriting, resident files are only
  903. written into memory, and not written out to disk at present, so avoid
  904. writing to files smaller than about 1kiB.
  905. - Implement ->truncate (fs/ntfs/inode.c::ntfs_truncate()) and
  906. ->setattr() (fs/ntfs/inode.c::ntfs_setattr()) inode operations for
  907. files with the purpose of intercepting and aborting all i_size
  908. changes which we do not support yet. ntfs_truncate() actually only
  909. emits a warning message but AFAICS our interception of i_size changes
  910. elsewhere means ntfs_truncate() never gets called for i_size changes.
  911. It is only called from generic_file_write() when we fail in
  912. ntfs_prepare_{,nonresident_}write() in order to discard any
  913. instantiated buffers beyond i_size. Thus i_size is not actually
  914. changed so our warning message is enough. Unfortunately it is not
  915. possible to easily determine if i_size is being changed or not hence
  916. we just emit an appropriately worded error message.
  917. 2.0.25 - Small bug fixes and cleanups.
  918. - Unlock the page in an out of memory error code path in
  919. fs/ntfs/aops.c::ntfs_read_block().
  920. - If fs/ntfs/aops.c::ntfs_read_page() is called on an uptodate page,
  921. just unlock the page and return. (This can happen due to ->writepage
  922. clearing PageUptodate() during write out of MstProtected()
  923. attributes.
  924. - Remove leaked write code again.
  925. 2.0.24 - Cleanups.
  926. - Treat BUG_ON() as ASSERT() not VERIFY(), i.e. do not use side effects
  927. inside BUG_ON(). (Adam J. Richter)
  928. - Split logical OR expressions inside BUG_ON() into individual BUG_ON()
  929. calls for improved debugging. (Adam J. Richter)
  930. - Add errors flag to the ntfs volume state, accessed via
  931. NVol{,Set,Clear}Errors(vol).
  932. - Do not allow read-write remounts of read-only volumes with errors.
  933. - Clarify comment for ntfs file operation sendfile which was added by
  934. Christoph Hellwig a while ago (just using generic_file_sendfile())
  935. to say that ntfs ->sendfile is only used for the case where the
  936. source data is on the ntfs partition and the destination is
  937. somewhere else, i.e. nothing we need to concern ourselves with.
  938. - Add generic_file_write() as our ntfs file write operation.
  939. 2.0.23 - Major bug fixes (races, deadlocks, non-i386 architectures).
  940. - Massive internal locking changes to mft record locking. Fixes lock
  941. recursion and replaces the mrec_lock read/write semaphore with a
  942. mutex. Also removes the now superfluous mft_count. This fixes several
  943. race conditions and deadlocks, especially in the future write code.
  944. - Fix ntfs over loopback for compressed files by adding an
  945. optimization barrier. (gcc was screwing up otherwise ?)
  946. - Miscellaneous cleanups all over the code and a fix or two in error
  947. handling code paths.
  948. Thanks go to Christoph Hellwig for pointing out the following two:
  949. - Remove now unused function fs/ntfs/malloc.h::vmalloc_nofs().
  950. - Fix ntfs_free() for ia64 and parisc by checking for VMALLOC_END, too.
  951. 2.0.22 - Cleanups, mainly to ntfs_readdir(), and use C99 initializers.
  952. - Change fs/ntfs/dir.c::ntfs_reddir() to only read/write ->f_pos once
  953. at entry/exit respectively.
  954. - Use C99 initializers for structures.
  955. - Remove unused variable blocks from fs/ntfs/aops.c::ntfs_read_block().
  956. 2.0.21 - Check for, and refuse to work with too large files/directories/volumes.
  957. - Limit volume size at mount time to 2TiB on architectures where
  958. unsigned long is 32-bits (fs/ntfs/super.c::parse_ntfs_boot_sector()).
  959. This is the most we can do without overflowing the 32-bit limit of
  960. the block device size imposed on us by sb_bread() and sb_getblk()
  961. for the time being.
  962. - Limit file/directory size at open() time to 16TiB on architectures
  963. where unsigned long is 32-bits (fs/ntfs/file.c::ntfs_file_open() and
  964. fs/ntfs/dir.c::ntfs_dir_open()). This is the most we can do without
  965. overflowing the page cache page index.
  966. 2.0.20 - Support non-resident directory index bitmaps, fix page leak in readdir.
  967. - Move the directory index bitmap to use an attribute inode instead of
  968. having special fields for it inside the ntfs inode structure. This
  969. means that the index bitmaps now use the page cache for i/o, too,
  970. and also as a side effect we get support for non-resident index
  971. bitmaps for free.
  972. - Simplify/cleanup error handling in fs/ntfs/dir.c::ntfs_readdir() and
  973. fix a page leak that manifested itself in some cases.
  974. - Add fs/ntfs/inode.c::ntfs_put_inode(), which we need to release the
  975. index bitmap inode on the final iput().
  976. 2.0.19 - Fix race condition, improvements, and optimizations in i/o interface.
  977. - Apply block optimization added to fs/ntfs/aops.c::ntfs_read_block()
  978. to fs/ntfs/compress.c::ntfs_file_read_compressed_block() as well.
  979. - Drop the "file" from ntfs_file_read_compressed_block().
  980. - Rename fs/ntfs/aops.c::ntfs_enb_buffer_read_async() to
  981. ntfs_end_buffer_async_read() (more like the fs/buffer.c counterpart).
  982. - Update ntfs_end_buffer_async_read() with the improved logic from
  983. its updated counterpart fs/buffer.c::end_buffer_async_read(). Apply
  984. further logic improvements to better determine when we set PageError.
  985. - Update submission of buffers in fs/ntfs/aops.c::ntfs_read_block() to
  986. check for the buffers being uptodate first in line with the updated
  987. fs/buffer.c::block_read_full_page(). This plugs a small race
  988. condition.
  989. 2.0.18 - Fix race condition in reading of compressed files.
  990. - There was a narrow window between checking a buffer head for being
  991. uptodate and locking it in ntfs_file_read_compressed_block(). We now
  992. lock the buffer and then check whether it is uptodate or not.
  993. 2.0.17 - Cleanups and optimizations - shrinking the ToDo list.
  994. - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to return an error
  995. code and update callers, i.e. ntfs_iget(), to pass that error code
  996. up instead of just using -EIO.
  997. - Modifications to super.c to ensure that both mount and remount
  998. cannot set any write related options when the driver is compiled
  999. read-only.
  1000. - Optimize block resolution in fs/ntfs/aops.c::ntfs_read_block() to
  1001. cache the current runlist element. This should improve performance
  1002. when reading very large and/or very fragmented data.
  1003. 2.0.16 - Convert access to $MFT/$BITMAP to attribute inode API.
  1004. - Fix a stupid bug introduced in 2.0.15 where we were unmapping the
  1005. wrong inode in fs/ntfs/inode.c::ntfs_attr_iget().
  1006. - Fix debugging check in fs/ntfs/aops.c::ntfs_read_block().
  1007. - Convert $MFT/$BITMAP access to attribute inode API and remove all
  1008. remnants of the ugly mftbmp address space and operations hack. This
  1009. means we finally have only one readpage function as well as only one
  1010. async io completion handler. Yey! The mft bitmap is now just an
  1011. attribute inode and is accessed from vol->mftbmp_ino just as if it
  1012. were a normal file. Fake inodes rule. (-:
  1013. 2.0.15 - Fake inodes based attribute i/o via the pagecache, fixes and cleanups.
  1014. - Fix silly bug in fs/ntfs/super.c::parse_options() which was causing
  1015. remounts to fail when the partition had an entry in /etc/fstab and
  1016. the entry specified the nls= option.
  1017. - Apply same macro magic used in fs/ntfs/inode.h to fs/ntfs/volume.h to
  1018. expand all the helper functions NVolFoo(), NVolSetFoo(), and
  1019. NVolClearFoo().
  1020. - Move copyright statement from driver initialisation message to
  1021. module description (fs/super.c). This makes the initialisation
  1022. message fit on one line and fits in better with rest of kernel.
  1023. - Update fs/ntfs/attrib.c::map_run_list() to work on both real and
  1024. attribute inodes, and both for files and directories.
  1025. - Implement fake attribute inodes allowing all attribute i/o to go via
  1026. the page cache and to use all the normal vfs/mm functionality:
  1027. - Add ntfs_attr_iget() and its helper ntfs_read_locked_attr_inode()
  1028. to fs/ntfs/inode.c.
  1029. - Add needed cleanup code to ntfs_clear_big_inode().
  1030. - Merge address space operations for files and directories (aops.c),
  1031. now just have ntfs_aops:
  1032. - Rename:
  1033. end_buffer_read_attr_async() -> ntfs_end_buffer_read_async(),
  1034. ntfs_attr_read_block() -> ntfs_read_block(),
  1035. ntfs_file_read_page() -> ntfs_readpage().
  1036. - Rewrite fs/ntfs/aops.c::ntfs_readpage() to work on both real and
  1037. attribute inodes, and both for files and directories.
  1038. - Remove obsolete fs/ntfs/aops.c::ntfs_mst_readpage().
  1039. 2.0.14 - Run list merging code cleanup, minor locking changes, typo fixes.
  1040. - Change fs/ntfs/super.c::ntfs_statfs() to not rely on BKL by moving
  1041. the locking out of super.c::get_nr_free_mft_records() and taking and
  1042. dropping the mftbmp_lock rw_semaphore in ntfs_statfs() itself.
  1043. - Bring attribute runlist merging code (fs/ntfs/attrib.c) in sync with
  1044. current userspace ntfs library code. This means that if a merge
  1045. fails the original runlists are always left unmodified instead of
  1046. being silently corrupted.
  1047. - Misc typo fixes.
  1048. 2.0.13 - Use iget5_locked() in preparation for fake inodes and small cleanups.
  1049. - Remove nr_mft_bits and the now superfluous union with nr_mft_records
  1050. from ntfs_volume structure.
  1051. - Remove nr_lcn_bits and the now superfluous union with nr_clusters
  1052. from ntfs_volume structure.
  1053. - Use iget5_locked() and friends instead of conventional iget(). Wrap
  1054. the call in fs/ntfs/inode.c::ntfs_iget() and update callers of iget()
  1055. to use ntfs_iget(). Leave only one iget() call at mount time so we
  1056. don't need an ntfs_iget_mount().
  1057. - Change fs/ntfs/inode.c::ntfs_new_extent_inode() to take mft_no as an
  1058. additional argument.
  1059. 2.0.12 - Initial cleanup of address space operations following 2.0.11 changes.
  1060. - Merge fs/ntfs/aops.c::end_buffer_read_mst_async() and
  1061. fs/ntfs/aops.c::end_buffer_read_file_async() into one function
  1062. fs/ntfs/aops.c::end_buffer_read_attr_async() using NInoMstProtected()
  1063. to determine whether to apply mst fixups or not.
  1064. - Above change allows merging fs/ntfs/aops.c::ntfs_file_read_block()
  1065. and fs/ntfs/aops.c::ntfs_mst_readpage() into one function
  1066. fs/ntfs/aops.c::ntfs_attr_read_block(). Also, create a tiny wrapper
  1067. fs/ntfs/aops.c::ntfs_mst_readpage() to transform the parameters from
  1068. the VFS readpage function prototype to the ntfs_attr_read_block()
  1069. function prototype.
  1070. 2.0.11 - Initial preparations for fake inode based attribute i/o.
  1071. - Move definition of ntfs_inode_state_bits to fs/ntfs/inode.h and
  1072. do some macro magic (adapted from include/linux/buffer_head.h) to
  1073. expand all the helper functions NInoFoo(), NInoSetFoo(), and
  1074. NInoClearFoo().
  1075. - Add new flag to ntfs_inode_state_bits: NI_Sparse.
  1076. - Add new fields to ntfs_inode structure to allow use of fake inodes
  1077. for attribute i/o: type, name, name_len. Also add new state bits:
  1078. NI_Attr, which, if set, indicates the inode is a fake inode, and
  1079. NI_MstProtected, which, if set, indicates the attribute uses multi
  1080. sector transfer protection, i.e. fixups need to be applied after
  1081. reads and before/after writes.
  1082. - Rename fs/ntfs/inode.c::ntfs_{new,clear,destroy}_inode() to
  1083. ntfs_{new,clear,destroy}_extent_inode() and update callers.
  1084. - Use ntfs_clear_extent_inode() in fs/ntfs/inode.c::__ntfs_clear_inode()
  1085. instead of ntfs_destroy_extent_inode().
  1086. - Cleanup memory deallocations in {__,}ntfs_clear_{,big_}inode().
  1087. - Make all operations on ntfs inode state bits use the NIno* functions.
  1088. - Set up the new ntfs inode fields and state bits in
  1089. fs/ntfs/inode.c::ntfs_read_inode() and add appropriate cleanup of
  1090. allocated memory to __ntfs_clear_inode().
  1091. - Cleanup ntfs_inode structure a bit for better ordering of elements
  1092. w.r.t. their size to allow better packing of the structure in memory.
  1093. 2.0.10 - There can only be 2^32 - 1 inodes on an NTFS volume.
  1094. - Add check at mount time to verify that the number of inodes on the
  1095. volume does not exceed 2^32 - 1, which is the maximum allowed for
  1096. NTFS according to Microsoft.
  1097. - Change mft_no member of ntfs_inode structure to be unsigned long.
  1098. Update all users. This makes ntfs_inode->mft_no just a copy of struct
  1099. inode->i_ino. But we can't just always use struct inode->i_ino and
  1100. remove mft_no because extent inodes do not have an attached struct
  1101. inode.
  1102. 2.0.9 - Decompression engine now uses a single buffer and other cleanups.
  1103. - Change decompression engine to use a single buffer protected by a
  1104. spin lock instead of per-CPU buffers. (Rusty Russell)
  1105. - Do not update cb_pos when handling a partial final page during
  1106. decompression of a sparse compression block, as the value is later
  1107. reset without being read/used. (Rusty Russell)
  1108. - Switch to using the new KM_BIO_SRC_IRQ for atomic kmap()s. (Andrew
  1109. Morton)
  1110. - Change buffer size in ntfs_readdir()/ntfs_filldir() to use
  1111. NLS_MAX_CHARSET_SIZE which makes the buffers almost 1kiB each but
  1112. it also makes everything safer so it is a good thing.
  1113. - Miscellaneous minor cleanups to comments.
  1114. 2.0.8 - Major updates for handling of case sensitivity and dcache aliasing.
  1115. Big thanks go to Al Viro and other inhabitants of #kernel for investing
  1116. their time to discuss the case sensitivity and dcache aliasing issues.
  1117. - Remove unused source file fs/ntfs/attraops.c.
  1118. - Remove show_inodes mount option(s), thus dropping support for
  1119. displaying of short file names.
  1120. - Remove deprecated mount option posix.
  1121. - Restore show_sys_files mount option.
  1122. - Add new mount option case_sensitive, to determine if the driver
  1123. treats file names as case sensitive or not. If case sensitive, create
  1124. file names in the POSIX namespace. Otherwise create file names in the
  1125. LONG/WIN32 namespace. Note, files remain accessible via their short
  1126. file name, if it exists.
  1127. - Remove really dumb logic bug in boot sector recovery code.
  1128. - Fix dcache aliasing issues wrt short/long file names via changes
  1129. to fs/ntfs/dir.c::ntfs_lookup_inode_by_name() and
  1130. fs/ntfs/namei.c::ntfs_lookup():
  1131. - Add additional argument to ntfs_lookup_inode_by_name() in which we
  1132. return information about the matching file name if the case is not
  1133. matching or the match is a short file name. See comments above the
  1134. function definition for details.
  1135. - Change ntfs_lookup() to only create dcache entries for the correctly
  1136. cased file name and only for the WIN32 namespace counterpart of DOS
  1137. namespace file names. This ensures we have only one dentry per
  1138. directory and also removes all dcache aliasing issues between short
  1139. and long file names once we add write support. See comments above
  1140. function for details.
  1141. - Fix potential 1 byte overflow in fs/ntfs/unistr.c::ntfs_ucstonls().
  1142. 2.0.7 - Minor cleanups and updates for changes in core kernel code.
  1143. - Remove much of the NULL struct element initializers.
  1144. - Various updates to make compatible with recent kernels.
  1145. - Remove defines of MAX_BUF_PER_PAGE and include linux/buffer_head.h
  1146. in fs/ntfs/ntfs.h instead.
  1147. - Remove no longer needed KERNEL_VERSION checks. We are now in the
  1148. kernel proper so they are no longer needed.
  1149. 2.0.6 - Major bugfix to make compatible with other kernel changes.
  1150. - Initialize the mftbmp address space properly now that there are more
  1151. fields in the struct address_space. This was leading to hangs and
  1152. oopses on umount since 2.5.12 because of changes to other parts of
  1153. the kernel. We probably want a kernel generic init_address_space()
  1154. function...
  1155. - Drop BKL from ntfs_readdir() after consultation with Al Viro. The
  1156. only caller of ->readdir() is vfs_readdir() which holds i_mutex
  1157. during the call, and i_mutex is sufficient protection against changes
  1158. in the directory inode (including ->i_size).
  1159. - Use generic_file_llseek() for directories (as opposed to
  1160. default_llseek()) as this downs i_mutex instead of the BKL which is
  1161. what we now need for exclusion against ->f_pos changes considering we
  1162. no longer take the BKL in ntfs_readdir().
  1163. 2.0.5 - Major bugfix. Buffer overflow in extent inode handling.
  1164. - No need to set old blocksize in super.c::ntfs_fill_super() as the
  1165. VFS does so via invocation of deactivate_super() calling
  1166. fs->fill_super() calling block_kill_super() which does it.
  1167. - BKL moved from VFS into dir.c::ntfs_readdir(). (Linus Torvalds)
  1168. -> Do we really need it? I don't think so as we have exclusion on
  1169. the directory ntfs_inode rw_semaphore mrec_lock. We mmight have to
  1170. move the ->f_pos accesses under the mrec_lock though. Check this...
  1171. - Fix really, really, really stupid buffer overflow in extent inode
  1172. handling in mft.c::map_extent_mft_record().
  1173. 2.0.4 - Cleanups and updates for kernel 2.5.11.
  1174. - Add documentation on how to use the MD driver to be able to use NTFS
  1175. stripe and volume sets in Linux and generally cleanup documentation
  1176. a bit.
  1177. Remove all uses of kdev_t in favour of struct block_device *:
  1178. - Change compress.c::ntfs_file_read_compressed_block() to use
  1179. sb_getblk() instead of getblk().
  1180. - Change super.c::ntfs_fill_super() to use bdev_hardsect_size() instead
  1181. of get_hardsect_size().
  1182. - No need to get old blocksize in super.c::ntfs_fill_super() as
  1183. fs/super.c::get_sb_bdev() already does this.
  1184. - Set bh->b_bdev instead of bh->b_dev throughout aops.c.
  1185. 2.0.3 - Small bug fixes, cleanups, and performance improvements.
  1186. - Remove some dead code from mft.c.
  1187. - Optimize readpage and read_block functions throughout aops.c so that
  1188. only initialized blocks are read. Non-initialized ones have their
  1189. buffer head mapped, zeroed, and set up to date, without scheduling
  1190. any i/o. Thanks to Al Viro for advice on how to avoid the device i/o.
  1191. Thanks go to Andrew Morton for spotting the below:
  1192. - Fix buglet in allocate_compression_buffers() error code path.
  1193. - Call flush_dcache_page() after modifying page cache page contents in
  1194. ntfs_file_readpage().
  1195. - Check for existence of page buffers throughout aops.c before calling
  1196. create_empty_buffers(). This happens when an I/O error occurs and the
  1197. read is retried. (It also happens once writing is implemented so that
  1198. needed doing anyway but I had left it for later...)
  1199. - Don't BUG_ON() uptodate and/or mapped buffers throughout aops.c in
  1200. readpage and read_block functions. Reasoning same as above (i.e. I/O
  1201. error retries and future write code paths.)
  1202. 2.0.2 - Minor updates and cleanups.
  1203. - Cleanup: rename mst.c::__post_read_mst_fixup to post_write_mst_fixup
  1204. and cleanup the code a bit, removing the unused size parameter.
  1205. - Change default fmask to 0177 and update documentation.
  1206. - Change attrib.c::get_attr_search_ctx() to return the search context
  1207. directly instead of taking the address of a pointer. A return value
  1208. of NULL means the allocation failed. Updated all callers
  1209. appropriately.
  1210. - Update to 2.5.9 kernel (preserving backwards compatibility) by
  1211. replacing all occurences of page->buffers with page_buffers(page).
  1212. - Fix minor bugs in runlist merging, also minor cleanup.
  1213. - Updates to bootsector layout and mft mirror contents descriptions.
  1214. - Small bug fix in error detection in unistr.c and some cleanups.
  1215. - Grow name buffer allocations in unistr.c in aligned mutlipled of 64
  1216. bytes.
  1217. 2.0.1 - Minor updates.
  1218. - Make default umask correspond to documentation.
  1219. - Improve documentation.
  1220. - Set default mode to include execute bit. The {u,f,d}mask can be used
  1221. to take it away if desired. This allows binaries to be executed from
  1222. a mounted ntfs partition.
  1223. 2.0.0 - New version number. Remove TNG from the name. Now in the kernel.
  1224. - Add kill_super, just keeping up with the vfs changes in the kernel.
  1225. - Repeat some changes from tng-0.0.8 that somehow got lost on the way
  1226. from the CVS import into BitKeeper.
  1227. - Begin to implement proper handling of allocated_size vs
  1228. initialized_size vs data_size (i.e. i_size). Done are
  1229. mft.c::ntfs_mft_readpage(), aops.c::end_buffer_read_index_async(),
  1230. and attrib.c::load_attribute_list().
  1231. - Lock the runlist in attrib.c::load_attribute_list() while using it.
  1232. - Fix memory leak in ntfs_file_read_compressed_block() and generally
  1233. clean up compress.c a little, removing some uncommented/unused debug
  1234. code.
  1235. - Tidy up dir.c a little bit.
  1236. - Don't bother getting the runlist in inode.c::ntfs_read_inode().
  1237. - Merge mft.c::ntfs_mft_readpage() and aops.c::ntfs_index_readpage()
  1238. creating aops.c::ntfs_mst_readpage(), improving the handling of
  1239. holes and overflow in the process and implementing the correct
  1240. equivalent of ntfs_file_get_block() in ntfs_mst_readpage() itself.
  1241. I am aiming for correctness at the moment. Modularisation can come
  1242. later.
  1243. - Rename aops.c::end_buffer_read_index_async() to
  1244. end_buffer_read_mst_async() and optimize the overflow checking and
  1245. handling.
  1246. - Use the host of the mftbmp address space mapping to hold the ntfs
  1247. volume. This is needed so the async i/o completion handler can
  1248. retrieve a pointer to the volume. Hopefully this will not cause
  1249. problems elsewhere in the kernel... Otherwise will need to use a
  1250. fake inode.
  1251. - Complete implementation of proper handling of allocated_size vs
  1252. initialized_size vs data_size (i.e. i_size) in whole driver.
  1253. Basically aops.c is now completely rewritten.
  1254. - Change NTFS driver name to just NTFS and set version number to 2.0.0
  1255. to make a clear distinction from the old driver which is still on
  1256. version 1.1.22.
  1257. tng-0.0.8 - 08/03/2002 - Now using BitKeeper, http://linux-ntfs.bkbits.net/
  1258. - Replace bdevname(sb->s_dev) with sb->s_id.
  1259. - Remove now superfluous new-line characters in all callers of
  1260. ntfs_debug().
  1261. - Apply kludge in ntfs_read_inode(), setting i_nlink to 1 for
  1262. directories. Without this the "find" utility gets very upset which is
  1263. fair enough as Linux/Unix do not support directory hard links.
  1264. - Further runlist merging work. (Richard Russon)
  1265. - Backwards compatibility for gcc-2.95. (Richard Russon)
  1266. - Update to kernel 2.5.5-pre1 and rediff the now tiny patch.
  1267. - Convert to new filesystem declaration using ->ntfs_get_sb() and
  1268. replacing ntfs_read_super() with ntfs_fill_super().
  1269. - Set s_maxbytes to MAX_LFS_FILESIZE to avoid page cache page index
  1270. overflow on 32-bit architectures.
  1271. - Cleanup upcase loading code to use ntfs_(un)map_page().
  1272. - Disable/reenable preemtion in critical sections of compession engine.
  1273. - Replace device size determination in ntfs_fill_super() with
  1274. sb->s_bdev->bd_inode->i_size (in bytes) and remove now superfluous
  1275. function super.c::get_nr_blocks().
  1276. - Implement a mount time option (show_inodes) allowing choice of which
  1277. types of inode names readdir() returns and modify ntfs_filldir()
  1278. accordingly. There are several parameters to show_inodes:
  1279. system: system files
  1280. win32: long file names (including POSIX file names) [DEFAULT]
  1281. long: same as win32
  1282. dos: short file names only (excluding POSIX file names)
  1283. short: same as dos
  1284. posix: same as both win32 and dos
  1285. all: all file names
  1286. Note that the options are additive, i.e. specifying:
  1287. -o show_inodes=system,show_inodes=win32,show_inodes=dos
  1288. is the same as specifying:
  1289. -o show_inodes=all
  1290. Note that the "posix" and "all" options will show all directory
  1291. names, BUT the link count on each directory inode entry is set to 1,
  1292. due to Linux not supporting directory hard links. This may well
  1293. confuse some userspace applications, since the directory names will
  1294. have the same inode numbers. Thus it is NOT advisable to use the
  1295. "posix" or "all" options. We provide them only for completeness sake.
  1296. - Add copies of allocated_size, initialized_size, and compressed_size to
  1297. the ntfs inode structure and set them up in
  1298. inode.c::ntfs_read_inode(). These reflect the unnamed data attribute
  1299. for files and the index allocation attribute for directories.
  1300. - Add copies of allocated_size and initialized_size to ntfs inode for
  1301. $BITMAP attribute of large directories and set them up in
  1302. inode.c::ntfs_read_inode().
  1303. - Add copies of allocated_size and initialized_size to ntfs volume for
  1304. $BITMAP attribute of $MFT and set them up in
  1305. super.c::load_system_files().
  1306. - Parse deprecated ntfs driver options (iocharset, show_sys_files,
  1307. posix, and utf8) and tell user what the new options to use are. Note
  1308. we still do support them but they will be removed with kernel 2.7.x.
  1309. - Change all occurences of integer long long printf formatting to hex
  1310. as printk() will not support long long integer format if/when the
  1311. div64 patch goes into the kernel.
  1312. - Make slab caches have stable names and change the names to what they
  1313. were intended to be. These changes are required/made possible by the
  1314. new slab cache name handling which removes the length limitation by
  1315. requiring the caller of kmem_cache_create() to supply a stable name
  1316. which is then referenced but not copied.
  1317. - Rename run_list structure to run_list_element and create a new
  1318. run_list structure containing a pointer to a run_list_element
  1319. structure and a read/write semaphore. Adapt all users of runlists
  1320. to new scheme and take and release the lock as needed. This fixes a
  1321. nasty race as the run_list changes even when inodes are locked for
  1322. reading and even when the inode isn't locked at all, so we really
  1323. needed the serialization. We use a semaphore rather than a spinlock
  1324. as memory allocations can sleep and doing everything GFP_ATOMIC
  1325. would be silly.
  1326. - Cleanup read_inode() removing all code checking for lowest_vcn != 0.
  1327. This can never happen due to the nature of lookup_attr() and how we
  1328. support attribute lists. If it did happen it would imply the inode
  1329. being corrupt.
  1330. - Check for lowest_vcn != 0 in ntfs_read_inode() and mark the inode as
  1331. bad if found.
  1332. - Update to 2.5.6-pre2 changes in struct address_space.
  1333. - Use parent_ino() when accessing d_parent inode number in dir.c.
  1334. - Import Sourceforge CVS repository into BitKeeper repository:
  1335. http://linux-ntfs.bkbits.net/ntfs-tng-2.5
  1336. - Update fs/Makefile, fs/Config.help, fs/Config.in, and
  1337. Documentation/filesystems/ntfs.txt for NTFS TNG.
  1338. - Create kernel configuration option controlling whether debugging
  1339. is enabled or not.
  1340. - Add the required export of end_buffer_io_sync() from the patches
  1341. directory to the kernel code.
  1342. - Update inode.c::ntfs_show_options() with show_inodes mount option.
  1343. - Update errors mount option.
  1344. tng-0.0.7 - 13/02/2002 - The driver is now feature complete for read-only!
  1345. - Cleanup mft.c and it's debug/error output in particular. Fix a minor
  1346. bug in mapping of extent inodes. Update all the comments to fit all
  1347. the recent code changes.
  1348. - Modify vcn_to_lcn() to cope with entirely unmapped runlists.
  1349. - Cleanups in compress.c, mostly comments and folding help.
  1350. - Implement attrib.c::map_run_list() as a generic helper.
  1351. - Make compress.c::ntfs_file_read_compressed_block() use map_run_list()
  1352. thus making code shorter and enabling attribute list support.
  1353. - Cleanup incorrect use of [su]64 with %L printf format specifier in
  1354. all source files. Type casts to [unsigned] long long added to correct
  1355. the mismatches (important for architectures which have long long not
  1356. being 64 bits).
  1357. - Merge async io completion handlers for directory indexes and $MFT
  1358. data into one by setting the index_block_size{_bits} of the ntfs
  1359. inode for $MFT to the mft_record_size{_bits} of the ntfs_volume.
  1360. - Cleanup aops.c, update comments.
  1361. - Make ntfs_file_get_block() use map_run_list() so all files now
  1362. support attribute lists.
  1363. - Make ntfs_dir_readpage() almost verbatim copy of
  1364. block_read_full_page() by using ntfs_file_get_block() with only real
  1365. difference being the use of our own async io completion handler
  1366. rather than the default one, thus reducing the amount of code and
  1367. automatically enabling attribute list support for directory indices.
  1368. - Fix bug in load_attribute_list() - forgot to call brelse in error
  1369. code path.
  1370. - Change parameters to find_attr() and lookup_attr(). We no longer
  1371. pass in the upcase table and its length. These can be gotten from
  1372. ctx->ntfs_ino->vol->upcase{_len}. Update all callers.
  1373. - Cleanups in attrib.c.
  1374. - Implement merging of runlists, attrib.c::merge_run_lists() and its
  1375. helpers. (Richard Russon)
  1376. - Attribute lists part 2, attribute extents and multi part runlists:
  1377. enable proper support for LCN_RL_NOT_MAPPED and automatic mapping of
  1378. further runlist parts via attrib.c::map_run_list().
  1379. - Tiny endianness bug fix in decompress_mapping_pairs().
  1380. tng-0.0.6 - Encrypted directories, bug fixes, cleanups, debugging enhancements.
  1381. - Enable encrypted directories. (Their index root is marked encrypted
  1382. to indicate that new files in that directory should be created
  1383. encrypted.)
  1384. - Fix bug in NInoBmpNonResident() macro. (Cut and paste error.)
  1385. - Enable $Extend system directory. Most (if not all) extended system
  1386. files do not have unnamed data attributes so ntfs_read_inode() had to
  1387. special case them but that is ok, as the special casing recovery
  1388. happens inside an error code path so there is zero slow down in the
  1389. normal fast path. The special casing is done by introducing a new
  1390. function inode.c::ntfs_is_extended_system_file() which checks if any
  1391. of the hard links in the inode point to $Extend as being their parent
  1392. directory and if they do we assume this is an extended system file.
  1393. - Create a sysctl/proc interface to allow {dis,en}abling of debug output
  1394. when compiled with -DDEBUG. Default is debug messages to be disabled.
  1395. To enable them, one writes a non-zero value to /proc/sys/fs/ntfs-debug
  1396. (if /proc is enabled) or uses sysctl(2) to effect the same (if sysctl
  1397. interface is enabled). Inspired by old ntfs driver.
  1398. - Add debug_msgs insmod/kernel boot parameter to set whether debug
  1399. messages are {dis,en}abled. This is useful to enable debug messages
  1400. during ntfs initialization and is the only way to activate debugging
  1401. when the sysctl interface is not enabled.
  1402. - Cleanup debug output in various places.
  1403. - Remove all dollar signs ($) from the source (except comments) to
  1404. enable compilation on architectures whose gcc compiler does not
  1405. support dollar signs in the names of variables/constants. Attribute
  1406. types now start with AT_ instead of $ and $I30 is now just I30.
  1407. - Cleanup ntfs_lookup() and add consistency check of sequence numbers.
  1408. - Load complete runlist for $MFT/$BITMAP during mount and cleanup
  1409. access functions. This means we now cope with $MFT/$BITMAP being
  1410. spread accross several mft records.
  1411. - Disable modification of mft_zone_multiplier on remount. We can always
  1412. reenable this later on if we really want to, but we will need to make
  1413. sure we readjust the mft_zone size / layout accordingly.
  1414. tng-0.0.5 - Modernize for 2.5.x and further in line-ing with Al Viro's comments.
  1415. - Use sb_set_blocksize() instead of set_blocksize() and verify the
  1416. return value.
  1417. - Use sb_bread() instead of bread() throughout.
  1418. - Add index_vcn_size{_bits} to ntfs_inode structure to store the size
  1419. of a directory index block vcn. Apply resulting simplifications in
  1420. dir.c everywhere.
  1421. - Fix a small bug somewhere (but forgot what it was).
  1422. - Change ntfs_{debug,error,warning} to enable gcc to do type checking
  1423. on the printf-format parameter list and fix bugs reported by gcc
  1424. as a result. (Richard Russon)
  1425. - Move inode allocation strategy to Al's new stuff but maintain the
  1426. divorce of ntfs_inode from struct inode. To achieve this we have two
  1427. separate slab caches, one for big ntfs inodes containing a struct
  1428. inode and pure ntfs inodes and at the same time fix some faulty
  1429. error code paths in ntfs_read_inode().
  1430. - Show mount options in proc (inode.c::ntfs_show_options()).
  1431. tng-0.0.4 - Big changes, getting in line with Al Viro's comments.
  1432. - Modified (un)map_mft_record functions to be common for read and write
  1433. case. To specify which is which, added extra parameter at front of
  1434. parameter list. Pass either READ or WRITE to this, each has the
  1435. obvious meaning.
  1436. - General cleanups to allow for easier folding in vi.
  1437. - attrib.c::decompress_mapping_pairs() now accepts the old runlist
  1438. argument, and invokes attrib.c::merge_run_lists() to merge the old
  1439. and the new runlists.
  1440. - Removed attrib.c::find_first_attr().
  1441. - Implemented loading of attribute list and complete runlist for $MFT.
  1442. This means we now cope with $MFT being spread across several mft
  1443. records.
  1444. - Adapt to 2.5.2-pre9 and the changed create_empty_buffers() syntax.
  1445. - Adapt major/minor/kdev_t/[bk]devname stuff to new 2.5.x kernels.
  1446. - Make ntfs_volume be allocated via kmalloc() instead of using a slab
  1447. cache. There are too little ntfs_volume structures at any one time
  1448. to justify a private slab cache.
  1449. - Fix bogus kmap() use in async io completion. Now use kmap_atomic().
  1450. Use KM_BIO_IRQ on advice from IRC/kernel...
  1451. - Use ntfs_map_page() in map_mft_record() and create ->readpage method
  1452. for reading $MFT (ntfs_mft_readpage). In the process create dedicated
  1453. address space operations (ntfs_mft_aops) for $MFT inode mapping. Also
  1454. removed the now superfluous exports from the kernel core patch.
  1455. - Fix a bug where kfree() was used insted of ntfs_free().
  1456. - Change map_mft_record() to take ntfs_inode as argument instead of
  1457. vfs inode. Dito for unmap_mft_record(). Adapt all callers.
  1458. - Add pointer to ntfs_volume to ntfs_inode.
  1459. - Add mft record number and sequence number to ntfs_inode. Stop using
  1460. i_ino and i_generation for in-driver purposes.
  1461. - Implement attrib.c::merge_run_lists(). (Richard Russon)
  1462. - Remove use of proper inodes by extent inodes. Move i_ino and
  1463. i_generation to ntfs_inode to do this. Apply simplifications that
  1464. result and remove iget_no_wait(), etc.
  1465. - Pass ntfs_inode everywhere in the driver (used to be struct inode).
  1466. - Add reference counting in ntfs_inode for the ntfs inode itself and
  1467. for the mapped mft record.
  1468. - Extend mft record mapping so we can (un)map extent mft records (new
  1469. functions (un)map_extent_mft_record), and so mappings are reference
  1470. counted and don't have to happen twice if already mapped - just ref
  1471. count increases.
  1472. - Add -o iocharset as alias to -o nls for backwards compatibility.
  1473. - The latest core patch is now tiny. In fact just a single additional
  1474. export is necessary over the base kernel.
  1475. tng-0.0.3 - Cleanups, enhancements, bug fixes.
  1476. - Work on attrib.c::decompress_mapping_pairs() to detect base extents
  1477. and setup the runlist appropriately using knowledge provided by the
  1478. sizes in the base attribute record.
  1479. - Balance the get_/put_attr_search_ctx() calls so we don't leak memory
  1480. any more.
  1481. - Introduce ntfs_malloc_nofs() and ntfs_free() to allocate/free a single
  1482. page or use vmalloc depending on the amount of memory requested.
  1483. - Cleanup error output. The __FUNCTION__ "(): " is now added
  1484. automatically. Introduced a new header file debug.h to support this
  1485. and also moved ntfs_debug() function into it.
  1486. - Make reading of compressed files more intelligent and especially get
  1487. rid of the vmalloc_nofs() from readpage(). This now uses per CPU
  1488. buffers (allocated at first mount with cluster size <= 4kiB and
  1489. deallocated on last umount with cluster size <= 4kiB), and
  1490. asynchronous io for the compressed data using a list of buffer heads.
  1491. Er, we use synchronous io as async io only works on whole pages
  1492. covered by buffers and not on individual buffer heads...
  1493. - Bug fix for reading compressed files with sparse compression blocks.
  1494. tng-0.0.2 - Now handles larger/fragmented/compressed volumes/files/dirs.
  1495. - Fixed handling of directories when cluster size exceeds index block
  1496. size.
  1497. - Hide DOS only name space directory entries from readdir() but allow
  1498. them in lookup(). This should fix the problem that Linux doesn't
  1499. support directory hard links, while still allowing access to entries
  1500. via their short file name. This also has the benefit of mimicking
  1501. what Windows users are used to, so it is the ideal solution.
  1502. - Implemented sync_page everywhere so no more hangs in D state when
  1503. waiting for a page.
  1504. - Stop using bforget() in favour of brelse().
  1505. - Stop locking buffers unnecessarily.
  1506. - Implemented compressed files (inode->mapping contains uncompressed
  1507. data, raw compressed data is currently bread() into a vmalloc()ed
  1508. memory buffer).
  1509. - Enable compressed directories. (Their index root is marked compressed
  1510. to indicate that new files in that directory should be created
  1511. compressed.)
  1512. - Use vsnprintf rather than vsprintf in the ntfs_error and ntfs_warning
  1513. functions. (Thanks to Will Dyson for pointing this out.)
  1514. - Moved the ntfs_inode and ntfs_volume (the former ntfs_inode_info and
  1515. ntfs_sb_info) out of the common inode and super_block structures and
  1516. started using the generic_ip and generic_sbp pointers instead. This
  1517. makes ntfs entirely private with respect to the kernel tree.
  1518. - Detect compiler version and abort with error message if gcc less than
  1519. 2.96 is used.
  1520. - Fix bug in name comparison function in unistr.c.
  1521. - Implement attribute lists part 1, the infrastructure: search contexts
  1522. and operations, find_external_attr(), lookup_attr()) and make the
  1523. code use the infrastructure.
  1524. - Fix stupid buffer overflow bug that became apparent on larger run
  1525. list containing attributes.
  1526. - Fix bugs in readdir() that became apparent on larger directories.
  1527. The driver is now really useful and survives the test
  1528. find . -type f -exec md5sum "{}" \;
  1529. without any error messages on a over 1GiB sized partition with >16k
  1530. files on it, including compressed files and directories and many files
  1531. and directories with attribute lists.
  1532. tng-0.0.1 - The first useful version.
  1533. - Added ntfs_lookup().
  1534. - Added default upcase generation and handling.
  1535. - Added compile options to be shown on module init.
  1536. - Many bug fixes that were "hidden" before.
  1537. - Update to latest kernel.
  1538. - Added ntfs_readdir().
  1539. - Added file operations for mmap(), read(), open() and llseek(). We just
  1540. use the generic ones. The whole point of going through implementing
  1541. readpage() methods and where possible get_block() call backs is that
  1542. this allows us to make use of the generic high level methods provided
  1543. by the kernel.
  1544. The driver is now actually useful! Yey. (-: It undoubtedly has got bugs
  1545. though and it doesn't implement accesssing compressed files yet. Also,
  1546. accessing files with attribute list attributes is not implemented yet
  1547. either. But for small or simple filesystems it should work and allow
  1548. you to list directories, use stat on directory entries and the file
  1549. system, open, read, mmap and llseek around in files. A big mile stone
  1550. has been reached!
  1551. tng-0.0.0 - Initial version tag.
  1552. Initial driver implementation. The driver can mount and umount simple
  1553. NTFS filesystems (i.e. ones without attribute lists in the system
  1554. files). If the mount fails there might be problems in the error handling
  1555. code paths, so be warned. Otherwise it seems to be loading the system
  1556. files nicely and the mft record read mapping/unmapping seems to be
  1557. working nicely, too. Proof of inode metadata in the page cache and non-
  1558. resident file unnamed stream data in the page cache concepts is thus
  1559. complete.