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