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 - Change ntfs_map_runlist_nolock() and ntfs_attr_find_vcn_nolock() to
28 also take an optional attribute search context as argument. This
29 allows calling these functions with the mft record mapped. Update
32 2.1.24 - Lots of bug fixes and support more clean journal states.
34 - Support journals ($LogFile) which have been modified by chkdsk. This
35 means users can boot into Windows after we marked the volume dirty.
36 The Windows boot will run chkdsk and then reboot. The user can then
37 immediately boot into Linux rather than having to do a full Windows
38 boot first before rebooting into Linux and we will recognize such a
39 journal and empty it as it is clean by definition. Note, this only
40 works if chkdsk left the journal in an obviously clean state.
41 - Support journals ($LogFile) with only one restart page as well as
42 journals with two different restart pages. We sanity check both and
43 either use the only sane one or the more recent one of the two in the
44 case that both are valid.
45 - Add fs/ntfs/malloc.h::ntfs_malloc_nofs_nofail() which is analogous to
46 ntfs_malloc_nofs() but it performs allocations with __GFP_NOFAIL and
48 - Use ntfs_malloc_nofs_nofail() in the two critical regions in
49 fs/ntfs/runlist.c::ntfs_runlists_merge(). This means we no longer
50 need to panic() if the allocation fails as it now cannot fail.
51 - Fix two nasty runlist merging bugs that had gone unnoticed so far.
52 Thanks to Stefano Picerno for the bug report.
53 - Remove two bogus BUG_ON()s from fs/ntfs/mft.c.
54 - Fix handling of valid but empty mapping pairs array in
55 fs/ntfs/runlist.c::ntfs_mapping_pairs_decompress().
56 - Report unrepresentable inodes during ntfs_readdir() as KERN_WARNING
57 messages and include the inode number. Thanks to Yura Pakhuchiy for
59 - Change ntfs_rl_truncate_nolock() to throw away the runlist if the new
61 - Add runlist.[hc]::ntfs_rl_punch_nolock() which punches a caller
62 specified hole into a runlist.
63 - Fix a bug in fs/ntfs/index.c::ntfs_index_lookup(). When the returned
64 index entry is in the index root, we forgot to set the @ir pointer in
65 the index context. Thanks to Yura Pakhuchiy for finding this bug.
66 - Remove bogus setting of PageError in ntfs_read_compressed_block().
67 - Add fs/ntfs/attrib.[hc]::ntfs_resident_attr_value_resize().
68 - Fix a bug in ntfs_map_runlist_nolock() where we forgot to protect
69 access to the allocated size in the ntfs inode with the size lock.
70 - Fix ntfs_attr_vcn_to_lcn_nolock() and ntfs_attr_find_vcn_nolock() to
71 return LCN_ENOENT when there is no runlist and the allocated size is
73 - Fix load_attribute_list() to handle the case of a NULL runlist.
74 - Fix handling of sparse attributes in ntfs_attr_make_non_resident().
75 - Add BUG() checks to ntfs_attr_make_non_resident() and ntfs_attr_set()
76 to ensure that these functions are never called for compressed or
78 - Fix cluster (de)allocators to work when the runlist is NULL and more
79 importantly to take a locked runlist rather than them locking it
80 which leads to lock reversal.
81 - Truncate {a,c,m}time to the ntfs supported time granularity when
82 updating the times in the inode in ntfs_setattr().
83 - Fixup handling of sparse, compressed, and encrypted attributes in
84 fs/ntfs/inode.c::ntfs_read_locked_{,attr_,index_}inode(),
85 fs/ntfs/aops.c::ntfs_{read,write}page().
86 - Make ntfs_write_block() not instantiate sparse blocks if they contain
88 - Optimize fs/ntfs/aops.c::ntfs_write_block() by extending the page
89 lock protection over the buffer submission for i/o which allows the
90 removal of the get_bh()/put_bh() pairs for each buffer.
91 - Fix fs/ntfs/aops.c::ntfs_{read,write}_block() to handle the case
92 where a concurrent truncate has truncated the runlist under our feet.
93 - Fix page_has_buffers()/page_buffers() handling in fs/ntfs/aops.c.
94 - In fs/ntfs/aops.c::ntfs_end_buffer_async_read(), use a bit spin lock
95 in the first buffer head instead of a driver global spin lock to
97 - Minor fix to error handling and error message display in
98 fs/ntfs/aops.c::ntfs_prepare_nonresident_write().
99 - Change the mount options {u,f,d}mask to always parse the number as
100 an octal number to conform to how chmod(1) works, too. Thanks to
101 Giuseppe Bilotta and Horst von Brand for pointing out the errors of
103 - Fix various bugs in the runlist merging code. (Based on libntfs
104 changes by Richard Russon.)
105 - Fix sparse warnings that have crept in over time.
106 - Change ntfs_cluster_free() to require a write locked runlist on entry
107 since we otherwise get into a lock reversal deadlock if a read locked
108 runlist is passed in. In the process also change it to take an ntfs
109 inode instead of a vfs inode as parameter.
110 - Fix the definition of the CHKD ntfs record magic. It had an off by
111 two error causing it to be CHKB instead of CHKD.
112 - Fix a stupid bug in __ntfs_bitmap_set_bits_in_run() which caused the
113 count to become negative and hence we had a wild memset() scribbling
114 all over the system's ram.
116 2.1.23 - Implement extension of resident files and make writing safe as well as
117 many bug fixes, cleanups, and enhancements...
119 - Add printk rate limiting for ntfs_warning() and ntfs_error() when
120 compiled without debug. This avoids a possible denial of service
121 attack. Thanks to Carl-Daniel Hailfinger from SuSE for pointing this
123 - Fix compilation warnings on ia64. (Randy Dunlap)
124 - Use i_size_{read,write}() instead of reading i_size by hand and cache
125 the value where apropriate.
126 - Add size_lock to the ntfs_inode structure. This is an rw spinlock
127 and it locks against access to the inode sizes. Note, ->size_lock
128 is also accessed from irq context so you must use the _irqsave and
129 _irqrestore lock and unlock functions, respectively. Protect all
130 accesses to allocated_size, initialized_size, and compressed_size.
131 - Minor optimization to fs/ntfs/super.c::ntfs_statfs() and its helpers.
132 - Implement extension of resident files in the regular file write code
133 paths (fs/ntfs/aops.c::ntfs_{prepare,commit}_write()). At present
134 this only works until the data attribute becomes too big for the mft
135 record after which we abort the write returning -EOPNOTSUPP from
136 ntfs_prepare_write().
137 - Add disable_sparse mount option together with a per volume sparse
138 enable bit which is set appropriately and a per inode sparse disable
139 bit which is preset on some system file inodes as appropriate.
140 - Enforce that sparse support is disabled on NTFS volumes pre 3.0.
141 - Fix a bug in fs/ntfs/runlist.c::ntfs_mapping_pairs_decompress() in
142 the creation of the unmapped runlist element for the base attribute
144 - Split ntfs_map_runlist() into ntfs_map_runlist() and a non-locking
145 helper ntfs_map_runlist_nolock() which is used by ntfs_map_runlist().
146 This allows us to map runlist fragments with the runlist lock already
147 held without having to drop and reacquire it around the call. Adapt
149 - Change ntfs_find_vcn() to ntfs_find_vcn_nolock() which takes a locked
150 runlist. This allows us to find runlist elements with the runlist
151 lock already held without having to drop and reacquire it around the
152 call. Adapt all callers.
153 - Change time to u64 in time.h::ntfs2utc() as it otherwise generates a
154 warning in the do_div() call on sparc32. Thanks to Meelis Roos for
155 the report and analysis of the warning.
156 - Fix a nasty runlist merge bug when merging two holes.
157 - Set the ntfs_inode->allocated_size to the real allocated size in the
158 mft record for resident attributes (fs/ntfs/inode.c).
159 - Small readability cleanup to use "a" instead of "ctx->attr"
160 everywhere (fs/ntfs/inode.c).
161 - Make fs/ntfs/namei.c::ntfs_get_{parent,dentry} static and move the
162 definition of ntfs_export_ops from fs/ntfs/super.c to namei.c. Also,
163 declare ntfs_export_ops in fs/ntfs/ntfs.h.
164 - Correct sparse file handling. The compressed values need to be
165 checked and set in the ntfs inode as done for compressed files and
166 the compressed size needs to be used for vfs inode->i_blocks instead
167 of the allocated size, again, as done for compressed files.
168 - Add AT_EA in addition to AT_DATA to whitelist for being allowed to be
169 non-resident in fs/ntfs/attrib.c::ntfs_attr_can_be_non_resident().
170 - Add fs/ntfs/attrib.c::ntfs_attr_vcn_to_lcn_nolock() used by the new
172 - Fix bug in fs/ntfs/attrib.c::ntfs_find_vcn_nolock() where after
173 dropping the read lock and taking the write lock we were not checking
174 whether someone else did not already do the work we wanted to do.
175 - Rename fs/ntfs/attrib.c::ntfs_find_vcn_nolock() to
176 ntfs_attr_find_vcn_nolock() and update all callers.
177 - Add fs/ntfs/attrib.[hc]::ntfs_attr_make_non_resident().
178 - Fix sign of various error return values to be negative in
180 - Modify ->readpage and ->writepage (fs/ntfs/aops.c) so they detect and
181 handle the case where an attribute is converted from resident to
182 non-resident by a concurrent file write.
183 - Remove checks for NULL before calling kfree() since kfree() does the
184 checking itself. (Jesper Juhl)
185 - Some utilities modify the boot sector but do not update the checksum.
186 Thus, relax the checking in fs/ntfs/super.c::is_boot_sector_ntfs() to
187 only emit a warning when the checksum is incorrect rather than
188 refusing the mount. Thanks to Bernd Casimir for pointing this
190 - Update attribute definition handling.
191 - Add NTFS_MAX_CLUSTER_SIZE and NTFS_MAX_PAGES_PER_CLUSTER constants.
192 - Use NTFS_MAX_CLUSTER_SIZE in super.c instead of hard coding 0x10000.
193 - Use MAX_BUF_PER_PAGE instead of variable sized array allocation for
194 better code generation and one less sparse warning in fs/ntfs/aops.c.
195 - Remove spurious void pointer casts from fs/ntfs/. (Pekka Enberg)
196 - Use C99 style structure initialization after memory allocation where
197 possible (fs/ntfs/{attrib.c,index.c,super.c}). Thanks to Al Viro and
199 - Stamp the transaction log ($UsnJrnl), aka user space journal, if it
200 is active on the volume and we are mounting read-write or remounting
201 from read-only to read-write.
202 - Fix a bug in address space operations error recovery code paths where
203 if the runlist was not mapped at all and a mapping error occured we
204 would leave the runlist locked on exit to the function so that the
205 next access to the same file would try to take the lock and deadlock.
206 - Detect the case when Windows has been suspended to disk on the volume
207 to be mounted and if this is the case do not allow (re)mounting
208 read-write. This is done by parsing hiberfil.sys if present.
209 - Fix several occurences of a bug where we would perform 'var & ~const'
210 with a 64-bit variable and a int, i.e. 32-bit, constant. This causes
211 the higher order 32-bits of the 64-bit variable to be zeroed. To fix
212 this cast the 'const' to the same 64-bit type as 'var'.
213 - Change the runlist terminator of the newly allocated cluster(s) to
214 LCN_ENOENT in ntfs_attr_make_non_resident(). Otherwise the runlist
216 - Add an extra parameter @last_vcn to ntfs_get_size_for_mapping_pairs()
217 and ntfs_mapping_pairs_build() to allow the runlist encoding to be
218 partial which is desirable when filling holes in sparse attributes.
220 - Change ntfs_map_runlist_nolock() to only decompress the mapping pairs
221 if the requested vcn is inside it. Otherwise we get into problems
222 when we try to map an out of bounds vcn because we then try to map
223 the already mapped runlist fragment which causes
224 ntfs_mapping_pairs_decompress() to fail and return error. Update
225 ntfs_attr_find_vcn_nolock() accordingly.
226 - Fix a nasty deadlock that appeared in recent kernels.
227 The situation: VFS inode X on a mounted ntfs volume is dirty. For
228 same inode X, the ntfs_inode is dirty and thus corresponding on-disk
229 inode, i.e. mft record, which is in a dirty PAGE_CACHE_PAGE belonging
230 to the table of inodes, i.e. $MFT, inode 0.
232 Process 1: sys_sync()/umount()/whatever... calls
233 __sync_single_inode() for $MFT -> do_writepages() -> write_page for
234 the dirty page containing the on-disk inode X, the page is now locked
235 -> ntfs_write_mst_block() which clears PageUptodate() on the page to
236 prevent anyone else getting hold of it whilst it does the write out.
237 This is necessary as the on-disk inode needs "fixups" applied before
238 the write to disk which are removed again after the write and
239 PageUptodate is then set again. It then analyses the page looking
240 for dirty on-disk inodes and when it finds one it calls
241 ntfs_may_write_mft_record() to see if it is safe to write this
242 on-disk inode. This then calls ilookup5() to check if the
243 corresponding VFS inode is in icache(). This in turn calls ifind()
244 which waits on the inode lock via wait_on_inode whilst holding the
246 Process 2: pdflush results in a call to __sync_single_inode for the
247 same VFS inode X on the ntfs volume. This locks the inode (I_LOCK)
248 then calls write-inode -> ntfs_write_inode -> map_mft_record() ->
249 read_cache_page() for the page (in page cache of table of inodes
250 $MFT, inode 0) containing the on-disk inode. This page has
251 PageUptodate() clear because of Process 1 (see above) so
252 read_cache_page() blocks when it tries to take the page lock for the
253 page so it can call ntfs_read_page().
254 Thus Process 1 is holding the page lock on the page containing the
255 on-disk inode X and it is waiting on the inode X to be unlocked in
256 ifind() so it can write the page out and then unlock the page.
257 And Process 2 is holding the inode lock on inode X and is waiting for
258 the page to be unlocked so it can call ntfs_readpage() or discover
259 that Process 1 set PageUptodate() again and use the page.
260 Thus we have a deadlock due to ifind() waiting on the inode lock.
261 The solution: The fix is to use the newly introduced
262 ilookup5_nowait() which does not wait on the inode's lock and hence
263 avoids the deadlock. This is safe as we do not care about the VFS
264 inode and only use the fact that it is in the VFS inode cache and the
265 fact that the vfs and ntfs inodes are one struct in memory to find
266 the ntfs inode in memory if present. Also, the ntfs inode has its
267 own locking so it does not matter if the vfs inode is locked.
268 - Fix bug in mft record writing where we forgot to set the device in
269 the buffers when mapping them after the VM had discarded them.
270 Thanks to Martin MOKREJÅ for the bug report.
272 2.1.22 - Many bug and race fixes and error handling improvements.
274 - Improve error handling in fs/ntfs/inode.c::ntfs_truncate().
275 - Change fs/ntfs/inode.c::ntfs_truncate() to return an error code
276 instead of void and provide a helper ntfs_truncate_vfs() for the
277 vfs ->truncate method.
278 - Add a new ntfs inode flag NInoTruncateFailed() and modify
279 fs/ntfs/inode.c::ntfs_truncate() to set and clear it appropriately.
280 - Fix min_size and max_size definitions in ATTR_DEF structure in
281 fs/ntfs/layout.h to be signed.
282 - Add attribute definition handling helpers to fs/ntfs/attrib.[hc]:
283 ntfs_attr_size_bounds_check(), ntfs_attr_can_be_non_resident(), and
284 ntfs_attr_can_be_resident(), which in turn use the new private helper
285 ntfs_attr_find_in_attrdef().
286 - In fs/ntfs/aops.c::mark_ntfs_record_dirty(), take the
287 mapping->private_lock around the dirtying of the buffer heads
288 analagous to the way it is done in __set_page_dirty_buffers().
289 - Ensure the mft record size does not exceed the PAGE_CACHE_SIZE at
290 mount time as this cannot work with the current implementation.
291 - Check for location of attribute name and improve error handling in
292 general in fs/ntfs/inode.c::ntfs_read_locked_inode() and friends.
293 - In fs/ntfs/aops.c::ntfs_writepage(), if the page is fully outside
294 i_size, i.e. race with truncate, invalidate the buffers on the page
295 so that they become freeable and hence the page does not leak.
296 - Remove unused function fs/ntfs/runlist.c::ntfs_rl_merge(). (Adrian
298 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_find() that resulted in
299 a NULL pointer dereference in the error code path when a corrupt
300 attribute was found. (Thanks to Domen Puncer for the bug report.)
301 - Add MODULE_VERSION() to fs/ntfs/super.c.
302 - Make several functions and variables static. (Adrian Bunk)
303 - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() so it allocates
304 buffers for the page if they are not present and then marks the
305 buffers belonging to the ntfs record dirty. This causes the buffers
306 to become busy and hence they are safe from removal until the page
307 has been written out.
308 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find() in the
309 error handling code path that resulted in a BUG() due to trying to
310 unmap an extent mft record when the mapping of it had failed and it
311 thus was not mapped. (Thanks to Ken MacFerrin for the bug report.)
312 - Drop the runlist lock after the vcn has been read in
313 fs/ntfs/lcnalloc.c::__ntfs_cluster_free().
314 - Rewrite handling of multi sector transfer errors. We now do not set
315 PageError() when such errors are detected in the async i/o handler
316 fs/ntfs/aops.c::ntfs_end_buffer_async_read(). All users of mst
317 protected attributes now check the magic of each ntfs record as they
318 use it and act appropriately. This has the effect of making errors
319 granular per ntfs record rather than per page which solves the case
320 where we cannot access any of the ntfs records in a page when a
321 single one of them had an mst error. (Thanks to Ken MacFerrin for
323 - Fix error handling in fs/ntfs/quota.c::ntfs_mark_quotas_out_of_date()
324 where we failed to release i_sem on the $Quota/$Q attribute inode.
325 - Fix bug in handling of bad inodes in fs/ntfs/namei.c::ntfs_lookup().
326 - Add mapping of unmapped buffers to all remaining code paths, i.e.
327 fs/ntfs/aops.c::ntfs_write_mst_block(), mft.c::ntfs_sync_mft_mirror(),
328 and write_mft_record_nolock(). From now on we require that the
329 complete runlist for the mft mirror is always mapped into memory.
330 - Add creation of buffers to fs/ntfs/mft.c::ntfs_sync_mft_mirror().
331 - Improve error handling in fs/ntfs/aops.c::ntfs_{read,write}_block().
332 - Cleanup fs/ntfs/aops.c::ntfs_{read,write}page() since we know that a
333 resident attribute will be smaller than a page which makes the code
334 simpler. Also make the code more tolerant to concurrent ->truncate.
336 2.1.21 - Fix some races and bugs, rewrite mft write code, add mft allocator.
338 - Implement extent mft record deallocation
339 fs/ntfs/mft.c::ntfs_extent_mft_record_free().
340 - Splitt runlist related functions off from attrib.[hc] to runlist.[hc].
341 - Add vol->mft_data_pos and initialize it at mount time.
342 - Rename init_runlist() to ntfs_init_runlist(), ntfs_vcn_to_lcn() to
343 ntfs_rl_vcn_to_lcn(), decompress_mapping_pairs() to
344 ntfs_mapping_pairs_decompress(), ntfs_merge_runlists() to
345 ntfs_runlists_merge() and adapt all callers.
346 - Add fs/ntfs/runlist.[hc]::ntfs_get_nr_significant_bytes(),
347 ntfs_get_size_for_mapping_pairs(), ntfs_write_significant_bytes(),
348 and ntfs_mapping_pairs_build(), adapted from libntfs.
349 - Make fs/ntfs/lcnalloc.c::ntfs_cluster_free_from_rl_nolock() not
350 static and add a declaration for it to lcnalloc.h.
351 - Add fs/ntfs/lcnalloc.h::ntfs_cluster_free_from_rl() which is a static
352 inline wrapper for ntfs_cluster_free_from_rl_nolock() which takes the
353 cluster bitmap lock for the duration of the call.
354 - Add fs/ntfs/attrib.[hc]::ntfs_attr_record_resize().
355 - Implement the equivalent of memset() for an ntfs attribute in
356 fs/ntfs/attrib.[hc]::ntfs_attr_set() and switch
357 fs/ntfs/logfile.c::ntfs_empty_logfile() to using it.
358 - Remove unnecessary casts from LCN_* constants.
359 - Implement fs/ntfs/runlist.c::ntfs_rl_truncate_nolock().
360 - Add MFT_RECORD_OLD as a copy of MFT_RECORD in fs/ntfs/layout.h and
361 change MFT_RECORD to contain the NTFS 3.1+ specific fields.
362 - Add a helper function fs/ntfs/aops.c::mark_ntfs_record_dirty() which
363 marks all buffers belonging to an ntfs record dirty, followed by
364 marking the page the ntfs record is in dirty and also marking the vfs
365 inode containing the ntfs record dirty (I_DIRTY_PAGES).
366 - Switch fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to using the
367 new helper fs/ntfs/aops.c::mark_ntfs_record_dirty() and remove the no
368 longer needed fs/ntfs/index.[hc]::__ntfs_index_entry_mark_dirty().
369 - Move ntfs_{un,}map_page() from ntfs.h to aops.h and fix resulting
371 - Move the typedefs for runlist_element and runlist from types.h to
372 runlist.h and fix resulting include errors.
373 - Remove unused {__,}format_mft_record() from fs/ntfs/mft.c.
374 - Modify fs/ntfs/mft.c::__mark_mft_record_dirty() to use the helper
375 mark_ntfs_record_dirty() which also changes the behaviour in that we
376 now set the buffers belonging to the mft record dirty as well as the
378 - Update fs/ntfs/mft.c::write_mft_record_nolock() and sync_mft_mirror()
379 to cope with the fact that there now are dirty buffers in mft pages.
380 - Update fs/ntfs/inode.c::ntfs_write_inode() to also use the helper
381 mark_ntfs_record_dirty() and thus to set the buffers belonging to the
382 mft record dirty as well as the page itself.
383 - Fix compiler warnings on x86-64 in fs/ntfs/dir.c. (Randy Dunlap,
384 slightly modified by me)
385 - Add fs/ntfs/mft.c::try_map_mft_record() which fails with -EALREADY if
386 the mft record is already locked and otherwise behaves the same way
387 as fs/ntfs/mft.c::map_mft_record().
388 - Modify fs/ntfs/mft.c::write_mft_record_nolock() so that it only
389 writes the mft record if the buffers belonging to it are dirty.
390 Otherwise we assume that it was written out by other means already.
391 - Attempting to write outside initialized size is _not_ a bug so remove
392 the bug check from fs/ntfs/aops.c::ntfs_write_mst_block(). It is in
393 fact required to write outside initialized size when preparing to
394 extend the initialized size.
395 - Map the page instead of using page_address() before writing to it in
396 fs/ntfs/aops.c::ntfs_mft_writepage().
397 - Provide exclusion between opening an inode / mapping an mft record
398 and accessing the mft record in fs/ntfs/mft.c::ntfs_mft_writepage()
399 by setting the page not uptodate throughout ntfs_mft_writepage().
400 - Clear the page uptodate flag in fs/ntfs/aops.c::ntfs_write_mst_block()
401 to ensure noone can see the page whilst the mst fixups are applied.
402 - Add the helper fs/ntfs/mft.c::ntfs_may_write_mft_record() which
403 checks if an mft record may be written out safely obtaining any
404 necessary locks in the process. This is used by
405 fs/ntfs/aops.c::ntfs_write_mst_block().
406 - Modify fs/ntfs/aops.c::ntfs_write_mst_block() to also work for
407 writing mft records and improve its error handling in the process.
408 Now if any of the records in the page fail to be written out, all
409 other records will be written out instead of aborting completely.
410 - Remove ntfs_mft_aops and update all users to use ntfs_mst_aops.
411 - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to set the
412 ntfs_mst_aops for all inodes which are NInoMstProtected() and
413 ntfs_aops for all other inodes.
414 - Rename fs/ntfs/mft.c::sync_mft_mirror{,_umount}() to
415 ntfs_sync_mft_mirror{,_umount}() and change their parameters so they
416 no longer require an ntfs inode to be present. Update all callers.
417 - Cleanup the error handling in fs/ntfs/mft.c::ntfs_sync_mft_mirror().
418 - Clear the page uptodate flag in fs/ntfs/mft.c::ntfs_sync_mft_mirror()
419 to ensure noone can see the page whilst the mst fixups are applied.
420 - Remove the no longer needed fs/ntfs/mft.c::ntfs_mft_writepage() and
421 fs/ntfs/mft.c::try_map_mft_record().
422 - Fix callers of fs/ntfs/aops.c::mark_ntfs_record_dirty() to call it
423 with the ntfs inode which contains the page rather than the ntfs
424 inode the mft record of which is in the page.
425 - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by moving the
426 index inode bitmap inode release code from there to
427 fs/ntfs/inode.c::ntfs_clear_big_inode(). (Thanks to Christoph
428 Hellwig for spotting this.)
429 - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by taking the
430 inode semaphore around the code that sets ni->itype.index.bmp_ino to
431 NULL and reorganize the code to optimize it a bit. (Thanks to
432 Christoph Hellwig for spotting this.)
433 - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() to no longer take the
434 ntfs inode as a parameter as this is confusing and misleading and the
435 needed ntfs inode is available via NTFS_I(page->mapping->host).
436 Adapt all callers to this change.
437 - Modify fs/ntfs/mft.c::write_mft_record_nolock() and
438 fs/ntfs/aops.c::ntfs_write_mst_block() to only check the dirty state
439 of the first buffer in a record and to take this as the ntfs record
440 dirty state. We cannot look at the dirty state for subsequent
441 buffers because we might be racing with
442 fs/ntfs/aops.c::mark_ntfs_record_dirty().
443 - Move the static inline ntfs_init_big_inode() from fs/ntfs/inode.c to
444 inode.h and make fs/ntfs/inode.c::__ntfs_init_inode() non-static and
445 add a declaration for it to inode.h. Fix some compilation issues
446 that resulted due to #includes and header file interdependencies.
447 - Simplify setup of i_mode in fs/ntfs/inode.c::ntfs_read_locked_inode().
448 - Add helpers fs/ntfs/layout.h::MK_MREF() and MK_LE_MREF().
449 - Modify fs/ntfs/mft.c::map_extent_mft_record() to only verify the mft
450 record sequence number if it is specified (i.e. not zero).
451 - Add fs/ntfs/mft.[hc]::ntfs_mft_record_alloc() and various helper
452 functions used by it.
453 - Update Documentation/filesystems/ntfs.txt with instructions on how to
454 use the Device-Mapper driver with NTFS ftdisk/LDM raid. This removes
455 the linear raid problem with the Software RAID / MD driver when one
456 or more of the devices has an odd number of sectors.
458 2.1.20 - Fix two stupid bugs introduced in 2.1.18 release.
460 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_reinit_search_ctx()
461 where we did not clear ctx->al_entry but it was still set due to
462 changes in ntfs_attr_lookup() and ntfs_external_attr_find() in
464 - Fix another stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find()
465 where we forgot to unmap the extent mft record when we had finished
466 enumerating an attribute which caused a bug check to trigger when the
467 VFS calls ->clear_inode.
469 2.1.19 - Many cleanups, improvements, and a minor bug fix.
471 - Update ->setattr (fs/ntfs/inode.c::ntfs_setattr()) to refuse to
472 change the uid, gid, and mode of an inode as we do not support NTFS
474 - Remove BKL use from ntfs_setattr() syncing up with the rest of the
476 - Get rid of the ugly transparent union in fs/ntfs/dir.c::ntfs_readdir()
477 and ntfs_filldir() as per suggestion from Al Viro.
478 - Change '\0' and L'\0' to simply 0 as per advice from Linus Torvalds.
479 - Update ->truncate (fs/ntfs/inode.c::ntfs_truncate()) to check if the
480 inode size has changed and to only output an error if so.
481 - Rename fs/ntfs/attrib.h::attribute_value_length() to ntfs_attr_size().
482 - Add le{16,32,64} as well as sle{16,32,64} data types to
484 - Change ntfschar to be le16 instead of u16 in fs/ntfs/types.h.
485 - Add le versions of VCN, LCN, and LSN called leVCN, leLCN, and leLSN,
486 respectively, to fs/ntfs/types.h.
487 - Update endianness conversion macros in fs/ntfs/endian.h to use the
488 new types as appropriate.
489 - Do proper type casting when using sle64_to_cpup() in fs/ntfs/dir.c
491 - Add leMFT_REF data type to fs/ntfs/layout.h.
492 - Update all NTFS header files with the new little endian data types.
493 Affected files are fs/ntfs/layout.h, logfile.h, and time.h.
494 - Do proper type casting when using ntfs_is_*_recordp() in
495 fs/ntfs/logfile.c, mft.c, and super.c.
496 - Fix all the sparse bitwise warnings. Had to change all the typedef
497 enums storing little endian values to simple enums plus a typedef for
498 the datatype to make sparse happy.
499 - Fix a bug found by the new sparse bitwise warnings where the default
500 upcase table was defined as a pointer to wchar_t rather than ntfschar
501 in fs/ntfs/ntfs.h and super.c.
502 - Change {const_,}cpu_to_le{16,32}(0) to just 0 as suggested by Al Viro.
504 2.1.18 - Fix scheduling latencies at mount time as well as an endianness bug.
506 - Remove vol->nr_mft_records as it was pretty meaningless and optimize
507 the calculation of total/free inodes as used by statfs().
508 - Fix scheduling latencies in ntfs_fill_super() by dropping the BKL
509 because the code itself is using the ntfs_lock semaphore which
510 provides safe locking. (Ingo Molnar)
511 - Fix a potential bug in fs/ntfs/mft.c::map_extent_mft_record() that
512 could occur in the future for when we start closing/freeing extent
513 inodes if we don't set base_ni->ext.extent_ntfs_inos to NULL after
515 - Rename {find,lookup}_attr() to ntfs_attr_{find,lookup}() as well as
516 find_external_attr() to ntfs_external_attr_find() to cleanup the
517 namespace a bit and to be more consistent with libntfs.
518 - Rename {{re,}init,get,put}_attr_search_ctx() to
519 ntfs_attr_{{re,}init,get,put}_search_ctx() as well as the type
520 attr_search_context to ntfs_attr_search_ctx.
521 - Force use of ntfs_attr_find() in ntfs_attr_lookup() when searching
522 for the attribute list attribute itself.
523 - Fix endianness bug in ntfs_external_attr_find().
524 - Change ntfs_{external_,}attr_find() to return 0 on success, -ENOENT
525 if the attribute is not found, and -EIO on real error. In the case
526 of -ENOENT, the search context is updated to describe the attribute
527 before which the attribute being searched for would need to be
528 inserted if such an action were to be desired and in the case of
529 ntfs_external_attr_find() the search context is also updated to
530 indicate the attribute list entry before which the attribute list
531 entry of the attribute being searched for would need to be inserted
532 if such an action were to be desired. Also make ntfs_find_attr()
533 static and remove its prototype from attrib.h as it is not used
534 anywhere other than attrib.c. Update ntfs_attr_lookup() and all
535 callers of ntfs_{external,}attr_{find,lookup}() for the new return
537 - Minor cleanup of fs/ntfs/inode.c::ntfs_init_locked_inode().
539 2.1.17 - Fix bugs in mount time error code paths and other updates.
541 - Implement bitmap modification code (fs/ntfs/bitmap.[hc]). This
542 includes functions to set/clear a single bit or a run of bits.
543 - Add fs/ntfs/attrib.[hc]::ntfs_find_vcn() which returns the locked
544 runlist element containing a particular vcn. It also takes care of
545 mapping any needed runlist fragments.
546 - Implement cluster (de-)allocation code (fs/ntfs/lcnalloc.[hc]).
547 - Load attribute definition table from $AttrDef at mount time.
548 - Fix bugs in mount time error code paths involving (de)allocation of
549 the default and volume upcase tables.
550 - Remove ntfs_nr_mounts as it is no longer used.
552 2.1.16 - Implement access time updates, file sync, async io, and read/writev.
554 - Add support for readv/writev and aio_read/aio_write (fs/ntfs/file.c).
555 This is done by setting the appropriate file operations pointers to
556 the generic helper functions provided by mm/filemap.c.
557 - Implement fsync, fdatasync, and msync both for files (fs/ntfs/file.c)
558 and directories (fs/ntfs/dir.c).
559 - Add support for {a,m,c}time updates to inode.c::ntfs_write_inode().
560 Note, except for the root directory and any other system files opened
561 by the user, the system files will not have their access times
562 updated as they are only accessed at the inode level an hence the
563 file level functions which cause the times to be updated are never
566 2.1.15 - Invalidate quotas when (re)mounting read-write.
568 - Add new element itype.index.collation_rule to the ntfs inode
569 structure and set it appropriately in ntfs_read_locked_inode().
570 - Implement a new inode type "index" to allow efficient access to the
571 indices found in various system files and adapt inode handling
572 accordingly (fs/ntfs/inode.[hc]). An index inode is essentially an
573 attribute inode (NInoAttr() is true) with an attribute type of
574 AT_INDEX_ALLOCATION. As such, it is no longer allowed to call
575 ntfs_attr_iget() with an attribute type of AT_INDEX_ALLOCATION as
576 there would be no way to distinguish between normal attribute inodes
577 and index inodes. The function to obtain an index inode is
578 ntfs_index_iget() and it uses the helper function
579 ntfs_read_locked_index_inode(). Note, we do not overload
580 ntfs_attr_iget() as indices consist of multiple attributes so using
581 ntfs_attr_iget() to obtain an index inode would be confusing.
582 - Ensure that there is no overflow when doing page->index <<
583 PAGE_CACHE_SHIFT by casting page->index to s64 in fs/ntfs/aops.c.
584 - Use atomic kmap instead of kmap() in fs/ntfs/aops.c::ntfs_read_page()
585 and ntfs_read_block().
586 - Use case sensitive attribute lookups instead of case insensitive ones.
587 - Lock all page cache pages belonging to mst protected attributes while
588 accessing them to ensure we never see corrupt data while the page is
590 - Add framework for generic ntfs collation (fs/ntfs/collation.[hc]).
591 We have ntfs_is_collation_rule_supported() to check if the collation
592 rule you want to use is supported and ntfs_collation() which actually
593 collates two data items. We currently only support COLLATION_BINARY
594 and COLLATION_NTOFS_ULONG but support for other collation rules will
595 be added as the need arises.
596 - Add a new type, ntfs_index_context, to allow retrieval of an index
597 entry using the corresponding index key. To get an index context,
598 use ntfs_index_ctx_get() and to release it, use ntfs_index_ctx_put().
599 This also adds a new slab cache for the index contexts. To lookup a
600 key in an index inode, use ntfs_index_lookup(). After modifying an
601 index entry, call ntfs_index_entry_flush_dcache_page() followed by
602 ntfs_index_entry_mark_dirty() to ensure the changes are written out
603 to disk. For details see fs/ntfs/index.[hc]. Note, at present, if
604 an index entry is in the index allocation attribute rather than the
605 index root attribute it will not be written out (you will get a
606 warning message about discarded changes instead).
607 - Load the quota file ($Quota) and check if quota tracking is enabled
608 and if so, mark the quotas out of date. This causes windows to
609 rescan the volume on boot and update all quota entries.
610 - Add a set_page_dirty address space operation for ntfs_m[fs]t_aops.
611 It is simply set to __set_page_dirty_nobuffers() to make sure that
612 running set_page_dirty() on a page containing mft/ntfs records will
613 not affect the dirty state of the page buffers.
614 - Add fs/ntfs/index.c::__ntfs_index_entry_mark_dirty() which sets all
615 buffers that are inside the ntfs record in the page dirty after which
616 it sets the page dirty. This allows ->writepage to only write the
617 dirty index records rather than having to write all the records in
618 the page. Modify fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to
619 use this rather than __set_page_dirty_nobuffers().
620 - Implement fs/ntfs/aops.c::ntfs_write_mst_block() which enables the
621 writing of page cache pages belonging to mst protected attributes
622 like the index allocation attribute in directory indices and other
623 indices like $Quota/$Q, etc. This means that the quota is now marked
624 out of date on all volumes rather than only on ones where the quota
625 defaults entry is in the index root attribute of the $Quota/$Q index.
627 2.1.14 - Fix an NFSd caused deadlock reported by several users.
629 - Modify fs/ntfs/ntfs_readdir() to copy the index root attribute value
630 to a buffer so that we can put the search context and unmap the mft
631 record before calling the filldir() callback. We need to do this
632 because of NFSd which calls ->lookup() from its filldir callback()
633 and this causes NTFS to deadlock as ntfs_lookup() maps the mft record
634 of the directory and since ntfs_readdir() has got it mapped already
635 ntfs_lookup() deadlocks.
637 2.1.13 - Enable overwriting of resident files and housekeeping of system files.
639 - Implement writing of mft records (fs/ntfs/mft.[hc]), which includes
640 keeping the mft mirror in sync with the mft when mirrored mft records
641 are written. The functions are write_mft_record{,_nolock}(). The
642 implementation is quite rudimentary for now with lots of things not
643 implemented yet but I am not sure any of them can actually occur so
644 I will wait for people to hit each one and only then implement it.
645 - Commit open system inodes at umount time. This should make it
646 virtually impossible for sync_mft_mirror_umount() to ever be needed.
647 - Implement ->write_inode (fs/ntfs/inode.c::ntfs_write_inode()) for the
648 ntfs super operations. This gives us inode writing via the VFS inode
649 dirty code paths. Note: Access time updates are not implemented yet.
650 - Implement fs/ntfs/mft.[hc]::{,__}mark_mft_record_dirty() and make
651 fs/ntfs/aops.c::ntfs_writepage() and ntfs_commit_write() use it, thus
652 finally enabling resident file overwrite! (-8 This also includes a
653 placeholder for ->writepage (ntfs_mft_writepage()), which for now
654 just redirties the page and returns. Also, at umount time, we for
655 now throw away all mft data page cache pages after the last call to
656 ntfs_commit_inode() in the hope that all inodes will have been
657 written out by then and hence no dirty (meta)data will be lost. We
658 also check for this case and emit an error message telling the user
660 - Use set_page_writeback() and end_page_writeback() in the resident
661 attribute code path of fs/ntfs/aops.c::ntfs_writepage() otherwise
662 the radix-tree tag PAGECACHE_TAG_DIRTY remains set even though the
664 - Implement ntfs_mft_writepage() so it now checks if any of the mft
665 records in the page are dirty and if so redirties the page and
666 returns. Otherwise it just returns (after doing set_page_writeback(),
667 unlock_page(), end_page_writeback() or the radix-tree tag
668 PAGECACHE_TAG_DIRTY remains set even though the page is clean), thus
669 alowing the VM to do with the page as it pleases. Also, at umount
670 time, now only throw away dirty mft (meta)data pages if dirty inodes
671 are present and ask the user to email us if they see this happening.
672 - Add functions ntfs_{clear,set}_volume_flags(), to modify the volume
673 information flags (fs/ntfs/super.c).
674 - Mark the volume dirty when (re)mounting read-write and mark it clean
675 when unmounting or remounting read-only. If any volume errors are
676 found, the volume is left marked dirty to force chkdsk to run.
677 - Add code to set the NT4 compatibility flag when (re)mounting
678 read-write for newer NTFS versions but leave it commented out for now
679 since we do not make any modifications that are NTFS 1.2 specific yet
680 and since setting this flag breaks Captive-NTFS which is not nice.
681 This code must be enabled once we start writing NTFS 1.2 specific
682 changes otherwise Windows NTFS driver might crash / cause corruption.
684 2.1.12 - Fix the second fix to the decompression engine and some cleanups.
686 - Add a new address space operations struct, ntfs_mst_aops, for mst
687 protected attributes. This is because the default ntfs_aops do not
688 make sense with mst protected data and were they to write anything to
689 such an attribute they would cause data corruption so we provide
690 ntfs_mst_aops which does not have any write related operations set.
691 - Cleanup dirty ntfs inode handling (fs/ntfs/inode.[hc]) which also
692 includes an adapted ntfs_commit_inode() and an implementation of
693 ntfs_write_inode() which for now just cleans dirty inodes without
694 writing them (it does emit a warning that this is happening).
695 - Undo the second decompression engine fix (see 2.1.9 release ChangeLog
696 entry) as it was only fixing a theoretical bug but at the same time
697 it badly broke the handling of sparse and uncompressed compression
700 2.1.11 - Driver internal cleanups.
702 - Only build logfile.o if building the driver with read-write support.
703 - Really final white space cleanups.
704 - Use generic_ffs() instead of ffs() in logfile.c which allows the
705 log_page_size variable to be optimized by gcc into a constant.
706 - Rename uchar_t to ntfschar everywhere as uchar_t is unsigned 1-byte
707 char as defined by POSIX and as found on some systems.
709 2.1.10 - Force read-only (re)mounting of volumes with unsupported volume flags.
711 - Finish off the white space cleanups (remove trailing spaces, etc).
712 - Clean up ntfs_fill_super() and ntfs_read_inode_mount() by removing
713 the kludges around the first iget(). Instead of (re)setting ->s_op
714 we have the $MFT inode set up by explicit new_inode() / set ->i_ino /
715 insert_inode_hash() / call ntfs_read_inode_mount() directly. This
716 kills the need for second super_operations and allows to return error
717 from ntfs_read_inode_mount() without resorting to ugly "poisoning"
719 - Force read-only (re)mounting if any of the following bits are set in
720 the volume information flags:
721 VOLUME_IS_DIRTY, VOLUME_RESIZE_LOG_FILE,
722 VOLUME_UPGRADE_ON_MOUNT, VOLUME_DELETE_USN_UNDERWAY,
723 VOLUME_REPAIR_OBJECT_ID, VOLUME_MODIFIED_BY_CHKDSK
724 To make this easier we define VOLUME_MUST_MOUNT_RO_MASK with all the
725 above bits set so the test is made easy.
727 2.1.9 - Fix two bugs in decompression engine.
729 - Fix a bug where we would not always detect that we have reached the
730 end of a compression block because we were ending at minus one byte
731 which is effectively the same as being at the end. The fix is to
732 check whether the uncompressed buffer has been fully filled and if so
733 we assume we have reached the end of the compression block. A big
734 thank you to Marcin Gibuła for the bug report, the assistance in
735 tracking down the bug and testing the fix.
736 - Fix a possible bug where when a compressed read is truncated to the
737 end of the file, the offset inside the last page was not truncated.
739 2.1.8 - Handle $MFT mirror and $LogFile, improve time handling, and cleanups.
741 - Use get_bh() instead of manual atomic_inc() in fs/ntfs/compress.c.
742 - Modify fs/ntfs/time.c::ntfs2utc(), get_current_ntfs_time(), and
743 utc2ntfs() to work with struct timespec instead of time_t on the
744 Linux UTC time side thus preserving the full precision of the NTFS
745 time and only loosing up to 99 nano-seconds in the Linux UTC time.
746 - Move fs/ntfs/time.c to fs/ntfs/time.h and make the time functions
748 - Remove unused ntfs_dirty_inode().
749 - Cleanup super operations declaration in fs/ntfs/super.c.
750 - Wrap flush_dcache_mft_record_page() in #ifdef NTFS_RW.
751 - Add NInoTestSetFoo() and NInoTestClearFoo() macro magic to
752 fs/ntfs/inode.h and use it to declare NInoTest{Set,Clear}Dirty.
753 - Move typedefs for ntfs_attr and test_t from fs/ntfs/inode.c to
754 fs/ntfs/inode.h so they can be used elsewhere.
755 - Determine the mft mirror size as the number of mirrored mft records
756 and store it in ntfs_volume->mftmirr_size (fs/ntfs/super.c).
757 - Load the mft mirror at mount time and compare the mft records stored
758 in it to the ones in the mft. Force a read-only mount if the two do
759 not match (fs/ntfs/super.c).
760 - Fix type casting related warnings on 64-bit architectures. Thanks
761 to Meelis Roos for reporting them.
762 - Move %L to %ll as %L is floating point and %ll is integer which is
764 - Read the journal ($LogFile) and determine if the volume has been
765 shutdown cleanly and force a read-only mount if not (fs/ntfs/super.c
766 and fs/ntfs/logfile.c). This is a little bit of a crude check in
767 that we only look at the restart areas and not at the actual log
768 records so that there will be a very small number of cases where we
769 think that a volume is dirty when in fact it is clean. This should
770 only affect volumes that have not been shutdown cleanly and did not
771 have any pending, non-check-pointed i/o.
772 - If the $LogFile indicates a clean shutdown and a read-write (re)mount
773 is requested, empty $LogFile by overwriting it with 0xff bytes to
774 ensure that Windows cannot cause data corruption by replaying a stale
775 journal after Linux has written to the volume.
777 2.1.7 - Enable NFS exporting of mounted NTFS volumes.
779 - Set i_generation in the VFS inode from the seq_no of the NTFS inode.
780 - Make ntfs_lookup() NFS export safe, i.e. use d_splice_alias(), etc.
781 - Implement ->get_dentry() in fs/ntfs/namei.c::ntfs_get_dentry() as the
782 default doesn't allow inode number 0 which is a valid inode on NTFS
783 and even if it did allow that it uses iget() instead of ntfs_iget()
784 which makes it useless for us.
785 - Implement ->get_parent() in fs/ntfs/namei.c::ntfs_get_parent() as the
786 default just returns -EACCES which is not very useful.
787 - Define export operations (->s_export_op) for NTFS (ntfs_export_ops)
788 and set them up in the super block at mount time (super.c) this
789 allows mounted NTFS volumes to be exported via NFS.
790 - Add missing return -EOPNOTSUPP; in
791 fs/ntfs/aops.c::ntfs_commit_nonresident_write().
792 - Enforce no atime and no dir atime updates at mount/remount time as
793 they are not implemented yet anyway.
794 - Move a few assignments in fs/ntfs/attrib.c::load_attribute_list() to
795 after a NULL check. Thanks to Dave Jones for pointing this out.
797 2.1.6 - Fix minor bug in handling of compressed directories.
799 - Fix bug in handling of compressed directories. A compressed
800 directory is not really compressed so when we set the ->i_blocks
801 field of a compressed directory inode we were setting it from the
802 non-existing field ni->itype.compressed.size which gave random
803 results... For directories we now always use ni->allocated_size.
805 2.1.5 - Fix minor bug in attribute list attribute handling.
807 - Fix bug in attribute list handling. Actually it is not as much a bug
808 as too much protection in that we were not allowing attribute lists
809 which waste space on disk while Windows XP clearly allows it and in
810 fact creates such attribute lists so our driver was failing.
811 - Update NTFS documentation ready for 2.6 kernel release.
813 2.1.4 - Reduce compiler requirements.
815 - Remove all uses of unnamed structs and unions in the driver to make
816 old and newer gcc versions happy. Makes it a bit uglier IMO but at
817 least people will stop hassling me about it.
819 2.1.3 - Important bug fixes in corner cases.
821 - super.c::parse_ntfs_boot_sector(): Correct the check for 64-bit
822 clusters. (Philipp Thomas)
823 - attrib.c::load_attribute_list(): Fix bug when initialized_size is a
824 multiple of the block_size but not the cluster size. (Szabolcs
825 Szakacsits <szaka@sienet.hu>)
827 2.1.2 - Important bug fixes aleviating the hangs in statfs.
829 - Fix buggy free cluster and free inode determination logic.
831 2.1.1 - Minor updates.
833 - Add handling for initialized_size != data_size in compressed files.
834 - Reduce function local stack usage from 0x3d4 bytes to just noise in
835 fs/ntfs/upcase.c. (Randy Dunlap <rddunlap@osdl.ord>)
836 - Remove compiler warnings for newer gcc.
837 - Pages are no longer kmapped by mm/filemap.c::generic_file_write()
838 around calls to ->{prepare,commit}_write. Adapt NTFS appropriately
839 in fs/ntfs/aops.c::ntfs_prepare_nonresident_write() by using
840 kmap_atomic(KM_USER0).
842 2.1.0 - First steps towards write support: implement file overwrite.
844 - Add configuration option for developmental write support with an
845 appropriately scary configuration help text.
846 - Initial implementation of fs/ntfs/aops.c::ntfs_writepage() and its
847 helper fs/ntfs/aops.c::ntfs_write_block(). This enables mmap(2) based
848 overwriting of existing files on ntfs. Note: Resident files are
849 only written into memory, and not written out to disk at present, so
850 avoid writing to files smaller than about 1kiB.
851 - Initial implementation of fs/ntfs/aops.c::ntfs_prepare_write(), its
852 helper fs/ntfs/aops.c::ntfs_prepare_nonresident_write() and their
853 counterparts, fs/ntfs/aops.c::ntfs_commit_write(), and
854 fs/ntfs/aops.c::ntfs_commit_nonresident_write(), respectively. Also,
855 add generic_file_write() to the ntfs file operations (fs/ntfs/file.c).
856 This enables write(2) based overwriting of existing files on ntfs.
857 Note: As with mmap(2) based overwriting, resident files are only
858 written into memory, and not written out to disk at present, so avoid
859 writing to files smaller than about 1kiB.
860 - Implement ->truncate (fs/ntfs/inode.c::ntfs_truncate()) and
861 ->setattr() (fs/ntfs/inode.c::ntfs_setattr()) inode operations for
862 files with the purpose of intercepting and aborting all i_size
863 changes which we do not support yet. ntfs_truncate() actually only
864 emits a warning message but AFAICS our interception of i_size changes
865 elsewhere means ntfs_truncate() never gets called for i_size changes.
866 It is only called from generic_file_write() when we fail in
867 ntfs_prepare_{,nonresident_}write() in order to discard any
868 instantiated buffers beyond i_size. Thus i_size is not actually
869 changed so our warning message is enough. Unfortunately it is not
870 possible to easily determine if i_size is being changed or not hence
871 we just emit an appropriately worded error message.
873 2.0.25 - Small bug fixes and cleanups.
875 - Unlock the page in an out of memory error code path in
876 fs/ntfs/aops.c::ntfs_read_block().
877 - If fs/ntfs/aops.c::ntfs_read_page() is called on an uptodate page,
878 just unlock the page and return. (This can happen due to ->writepage
879 clearing PageUptodate() during write out of MstProtected()
881 - Remove leaked write code again.
885 - Treat BUG_ON() as ASSERT() not VERIFY(), i.e. do not use side effects
886 inside BUG_ON(). (Adam J. Richter)
887 - Split logical OR expressions inside BUG_ON() into individual BUG_ON()
888 calls for improved debugging. (Adam J. Richter)
889 - Add errors flag to the ntfs volume state, accessed via
890 NVol{,Set,Clear}Errors(vol).
891 - Do not allow read-write remounts of read-only volumes with errors.
892 - Clarify comment for ntfs file operation sendfile which was added by
893 Christoph Hellwig a while ago (just using generic_file_sendfile())
894 to say that ntfs ->sendfile is only used for the case where the
895 source data is on the ntfs partition and the destination is
896 somewhere else, i.e. nothing we need to concern ourselves with.
897 - Add generic_file_write() as our ntfs file write operation.
899 2.0.23 - Major bug fixes (races, deadlocks, non-i386 architectures).
901 - Massive internal locking changes to mft record locking. Fixes lock
902 recursion and replaces the mrec_lock read/write semaphore with a
903 mutex. Also removes the now superfluous mft_count. This fixes several
904 race conditions and deadlocks, especially in the future write code.
905 - Fix ntfs over loopback for compressed files by adding an
906 optimization barrier. (gcc was screwing up otherwise ?)
907 - Miscellaneous cleanups all over the code and a fix or two in error
909 Thanks go to Christoph Hellwig for pointing out the following two:
910 - Remove now unused function fs/ntfs/malloc.h::vmalloc_nofs().
911 - Fix ntfs_free() for ia64 and parisc by checking for VMALLOC_END, too.
913 2.0.22 - Cleanups, mainly to ntfs_readdir(), and use C99 initializers.
915 - Change fs/ntfs/dir.c::ntfs_reddir() to only read/write ->f_pos once
916 at entry/exit respectively.
917 - Use C99 initializers for structures.
918 - Remove unused variable blocks from fs/ntfs/aops.c::ntfs_read_block().
920 2.0.21 - Check for, and refuse to work with too large files/directories/volumes.
922 - Limit volume size at mount time to 2TiB on architectures where
923 unsigned long is 32-bits (fs/ntfs/super.c::parse_ntfs_boot_sector()).
924 This is the most we can do without overflowing the 32-bit limit of
925 the block device size imposed on us by sb_bread() and sb_getblk()
927 - Limit file/directory size at open() time to 16TiB on architectures
928 where unsigned long is 32-bits (fs/ntfs/file.c::ntfs_file_open() and
929 fs/ntfs/dir.c::ntfs_dir_open()). This is the most we can do without
930 overflowing the page cache page index.
932 2.0.20 - Support non-resident directory index bitmaps, fix page leak in readdir.
934 - Move the directory index bitmap to use an attribute inode instead of
935 having special fields for it inside the ntfs inode structure. This
936 means that the index bitmaps now use the page cache for i/o, too,
937 and also as a side effect we get support for non-resident index
939 - Simplify/cleanup error handling in fs/ntfs/dir.c::ntfs_readdir() and
940 fix a page leak that manifested itself in some cases.
941 - Add fs/ntfs/inode.c::ntfs_put_inode(), which we need to release the
942 index bitmap inode on the final iput().
944 2.0.19 - Fix race condition, improvements, and optimizations in i/o interface.
946 - Apply block optimization added to fs/ntfs/aops.c::ntfs_read_block()
947 to fs/ntfs/compress.c::ntfs_file_read_compressed_block() as well.
948 - Drop the "file" from ntfs_file_read_compressed_block().
949 - Rename fs/ntfs/aops.c::ntfs_enb_buffer_read_async() to
950 ntfs_end_buffer_async_read() (more like the fs/buffer.c counterpart).
951 - Update ntfs_end_buffer_async_read() with the improved logic from
952 its updated counterpart fs/buffer.c::end_buffer_async_read(). Apply
953 further logic improvements to better determine when we set PageError.
954 - Update submission of buffers in fs/ntfs/aops.c::ntfs_read_block() to
955 check for the buffers being uptodate first in line with the updated
956 fs/buffer.c::block_read_full_page(). This plugs a small race
959 2.0.18 - Fix race condition in reading of compressed files.
961 - There was a narrow window between checking a buffer head for being
962 uptodate and locking it in ntfs_file_read_compressed_block(). We now
963 lock the buffer and then check whether it is uptodate or not.
965 2.0.17 - Cleanups and optimizations - shrinking the ToDo list.
967 - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to return an error
968 code and update callers, i.e. ntfs_iget(), to pass that error code
969 up instead of just using -EIO.
970 - Modifications to super.c to ensure that both mount and remount
971 cannot set any write related options when the driver is compiled
973 - Optimize block resolution in fs/ntfs/aops.c::ntfs_read_block() to
974 cache the current runlist element. This should improve performance
975 when reading very large and/or very fragmented data.
977 2.0.16 - Convert access to $MFT/$BITMAP to attribute inode API.
979 - Fix a stupid bug introduced in 2.0.15 where we were unmapping the
980 wrong inode in fs/ntfs/inode.c::ntfs_attr_iget().
981 - Fix debugging check in fs/ntfs/aops.c::ntfs_read_block().
982 - Convert $MFT/$BITMAP access to attribute inode API and remove all
983 remnants of the ugly mftbmp address space and operations hack. This
984 means we finally have only one readpage function as well as only one
985 async io completion handler. Yey! The mft bitmap is now just an
986 attribute inode and is accessed from vol->mftbmp_ino just as if it
987 were a normal file. Fake inodes rule. (-:
989 2.0.15 - Fake inodes based attribute i/o via the pagecache, fixes and cleanups.
991 - Fix silly bug in fs/ntfs/super.c::parse_options() which was causing
992 remounts to fail when the partition had an entry in /etc/fstab and
993 the entry specified the nls= option.
994 - Apply same macro magic used in fs/ntfs/inode.h to fs/ntfs/volume.h to
995 expand all the helper functions NVolFoo(), NVolSetFoo(), and
997 - Move copyright statement from driver initialisation message to
998 module description (fs/super.c). This makes the initialisation
999 message fit on one line and fits in better with rest of kernel.
1000 - Update fs/ntfs/attrib.c::map_run_list() to work on both real and
1001 attribute inodes, and both for files and directories.
1002 - Implement fake attribute inodes allowing all attribute i/o to go via
1003 the page cache and to use all the normal vfs/mm functionality:
1004 - Add ntfs_attr_iget() and its helper ntfs_read_locked_attr_inode()
1006 - Add needed cleanup code to ntfs_clear_big_inode().
1007 - Merge address space operations for files and directories (aops.c),
1008 now just have ntfs_aops:
1010 end_buffer_read_attr_async() -> ntfs_end_buffer_read_async(),
1011 ntfs_attr_read_block() -> ntfs_read_block(),
1012 ntfs_file_read_page() -> ntfs_readpage().
1013 - Rewrite fs/ntfs/aops.c::ntfs_readpage() to work on both real and
1014 attribute inodes, and both for files and directories.
1015 - Remove obsolete fs/ntfs/aops.c::ntfs_mst_readpage().
1017 2.0.14 - Run list merging code cleanup, minor locking changes, typo fixes.
1019 - Change fs/ntfs/super.c::ntfs_statfs() to not rely on BKL by moving
1020 the locking out of super.c::get_nr_free_mft_records() and taking and
1021 dropping the mftbmp_lock rw_semaphore in ntfs_statfs() itself.
1022 - Bring attribute runlist merging code (fs/ntfs/attrib.c) in sync with
1023 current userspace ntfs library code. This means that if a merge
1024 fails the original runlists are always left unmodified instead of
1025 being silently corrupted.
1028 2.0.13 - Use iget5_locked() in preparation for fake inodes and small cleanups.
1030 - Remove nr_mft_bits and the now superfluous union with nr_mft_records
1031 from ntfs_volume structure.
1032 - Remove nr_lcn_bits and the now superfluous union with nr_clusters
1033 from ntfs_volume structure.
1034 - Use iget5_locked() and friends instead of conventional iget(). Wrap
1035 the call in fs/ntfs/inode.c::ntfs_iget() and update callers of iget()
1036 to use ntfs_iget(). Leave only one iget() call at mount time so we
1037 don't need an ntfs_iget_mount().
1038 - Change fs/ntfs/inode.c::ntfs_new_extent_inode() to take mft_no as an
1039 additional argument.
1041 2.0.12 - Initial cleanup of address space operations following 2.0.11 changes.
1043 - Merge fs/ntfs/aops.c::end_buffer_read_mst_async() and
1044 fs/ntfs/aops.c::end_buffer_read_file_async() into one function
1045 fs/ntfs/aops.c::end_buffer_read_attr_async() using NInoMstProtected()
1046 to determine whether to apply mst fixups or not.
1047 - Above change allows merging fs/ntfs/aops.c::ntfs_file_read_block()
1048 and fs/ntfs/aops.c::ntfs_mst_readpage() into one function
1049 fs/ntfs/aops.c::ntfs_attr_read_block(). Also, create a tiny wrapper
1050 fs/ntfs/aops.c::ntfs_mst_readpage() to transform the parameters from
1051 the VFS readpage function prototype to the ntfs_attr_read_block()
1054 2.0.11 - Initial preparations for fake inode based attribute i/o.
1056 - Move definition of ntfs_inode_state_bits to fs/ntfs/inode.h and
1057 do some macro magic (adapted from include/linux/buffer_head.h) to
1058 expand all the helper functions NInoFoo(), NInoSetFoo(), and
1060 - Add new flag to ntfs_inode_state_bits: NI_Sparse.
1061 - Add new fields to ntfs_inode structure to allow use of fake inodes
1062 for attribute i/o: type, name, name_len. Also add new state bits:
1063 NI_Attr, which, if set, indicates the inode is a fake inode, and
1064 NI_MstProtected, which, if set, indicates the attribute uses multi
1065 sector transfer protection, i.e. fixups need to be applied after
1066 reads and before/after writes.
1067 - Rename fs/ntfs/inode.c::ntfs_{new,clear,destroy}_inode() to
1068 ntfs_{new,clear,destroy}_extent_inode() and update callers.
1069 - Use ntfs_clear_extent_inode() in fs/ntfs/inode.c::__ntfs_clear_inode()
1070 instead of ntfs_destroy_extent_inode().
1071 - Cleanup memory deallocations in {__,}ntfs_clear_{,big_}inode().
1072 - Make all operations on ntfs inode state bits use the NIno* functions.
1073 - Set up the new ntfs inode fields and state bits in
1074 fs/ntfs/inode.c::ntfs_read_inode() and add appropriate cleanup of
1075 allocated memory to __ntfs_clear_inode().
1076 - Cleanup ntfs_inode structure a bit for better ordering of elements
1077 w.r.t. their size to allow better packing of the structure in memory.
1079 2.0.10 - There can only be 2^32 - 1 inodes on an NTFS volume.
1081 - Add check at mount time to verify that the number of inodes on the
1082 volume does not exceed 2^32 - 1, which is the maximum allowed for
1083 NTFS according to Microsoft.
1084 - Change mft_no member of ntfs_inode structure to be unsigned long.
1085 Update all users. This makes ntfs_inode->mft_no just a copy of struct
1086 inode->i_ino. But we can't just always use struct inode->i_ino and
1087 remove mft_no because extent inodes do not have an attached struct
1090 2.0.9 - Decompression engine now uses a single buffer and other cleanups.
1092 - Change decompression engine to use a single buffer protected by a
1093 spin lock instead of per-CPU buffers. (Rusty Russell)
1094 - Do not update cb_pos when handling a partial final page during
1095 decompression of a sparse compression block, as the value is later
1096 reset without being read/used. (Rusty Russell)
1097 - Switch to using the new KM_BIO_SRC_IRQ for atomic kmap()s. (Andrew
1099 - Change buffer size in ntfs_readdir()/ntfs_filldir() to use
1100 NLS_MAX_CHARSET_SIZE which makes the buffers almost 1kiB each but
1101 it also makes everything safer so it is a good thing.
1102 - Miscellaneous minor cleanups to comments.
1104 2.0.8 - Major updates for handling of case sensitivity and dcache aliasing.
1106 Big thanks go to Al Viro and other inhabitants of #kernel for investing
1107 their time to discuss the case sensitivity and dcache aliasing issues.
1109 - Remove unused source file fs/ntfs/attraops.c.
1110 - Remove show_inodes mount option(s), thus dropping support for
1111 displaying of short file names.
1112 - Remove deprecated mount option posix.
1113 - Restore show_sys_files mount option.
1114 - Add new mount option case_sensitive, to determine if the driver
1115 treats file names as case sensitive or not. If case sensitive, create
1116 file names in the POSIX namespace. Otherwise create file names in the
1117 LONG/WIN32 namespace. Note, files remain accessible via their short
1118 file name, if it exists.
1119 - Remove really dumb logic bug in boot sector recovery code.
1120 - Fix dcache aliasing issues wrt short/long file names via changes
1121 to fs/ntfs/dir.c::ntfs_lookup_inode_by_name() and
1122 fs/ntfs/namei.c::ntfs_lookup():
1123 - Add additional argument to ntfs_lookup_inode_by_name() in which we
1124 return information about the matching file name if the case is not
1125 matching or the match is a short file name. See comments above the
1126 function definition for details.
1127 - Change ntfs_lookup() to only create dcache entries for the correctly
1128 cased file name and only for the WIN32 namespace counterpart of DOS
1129 namespace file names. This ensures we have only one dentry per
1130 directory and also removes all dcache aliasing issues between short
1131 and long file names once we add write support. See comments above
1132 function for details.
1133 - Fix potential 1 byte overflow in fs/ntfs/unistr.c::ntfs_ucstonls().
1135 2.0.7 - Minor cleanups and updates for changes in core kernel code.
1137 - Remove much of the NULL struct element initializers.
1138 - Various updates to make compatible with recent kernels.
1139 - Remove defines of MAX_BUF_PER_PAGE and include linux/buffer_head.h
1140 in fs/ntfs/ntfs.h instead.
1141 - Remove no longer needed KERNEL_VERSION checks. We are now in the
1142 kernel proper so they are no longer needed.
1144 2.0.6 - Major bugfix to make compatible with other kernel changes.
1146 - Initialize the mftbmp address space properly now that there are more
1147 fields in the struct address_space. This was leading to hangs and
1148 oopses on umount since 2.5.12 because of changes to other parts of
1149 the kernel. We probably want a kernel generic init_address_space()
1151 - Drop BKL from ntfs_readdir() after consultation with Al Viro. The
1152 only caller of ->readdir() is vfs_readdir() which holds i_sem during
1153 the call, and i_sem is sufficient protection against changes in the
1154 directory inode (including ->i_size).
1155 - Use generic_file_llseek() for directories (as opposed to
1156 default_llseek()) as this downs i_sem instead of the BKL which is
1157 what we now need for exclusion against ->f_pos changes considering we
1158 no longer take the BKL in ntfs_readdir().
1160 2.0.5 - Major bugfix. Buffer overflow in extent inode handling.
1162 - No need to set old blocksize in super.c::ntfs_fill_super() as the
1163 VFS does so via invocation of deactivate_super() calling
1164 fs->fill_super() calling block_kill_super() which does it.
1165 - BKL moved from VFS into dir.c::ntfs_readdir(). (Linus Torvalds)
1166 -> Do we really need it? I don't think so as we have exclusion on
1167 the directory ntfs_inode rw_semaphore mrec_lock. We mmight have to
1168 move the ->f_pos accesses under the mrec_lock though. Check this...
1169 - Fix really, really, really stupid buffer overflow in extent inode
1170 handling in mft.c::map_extent_mft_record().
1172 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
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.
1186 2.0.3 - Small bug fixes, cleanups, and performance improvements.
1188 - Remove some dead code from mft.c.
1189 - Optimize readpage and read_block functions throughout aops.c so that
1190 only initialized blocks are read. Non-initialized ones have their
1191 buffer head mapped, zeroed, and set up to date, without scheduling
1192 any i/o. Thanks to Al Viro for advice on how to avoid the device i/o.
1193 Thanks go to Andrew Morton for spotting the below:
1194 - Fix buglet in allocate_compression_buffers() error code path.
1195 - Call flush_dcache_page() after modifying page cache page contents in
1196 ntfs_file_readpage().
1197 - Check for existence of page buffers throughout aops.c before calling
1198 create_empty_buffers(). This happens when an I/O error occurs and the
1199 read is retried. (It also happens once writing is implemented so that
1200 needed doing anyway but I had left it for later...)
1201 - Don't BUG_ON() uptodate and/or mapped buffers throughout aops.c in
1202 readpage and read_block functions. Reasoning same as above (i.e. I/O
1203 error retries and future write code paths.)
1205 2.0.2 - Minor updates and cleanups.
1207 - Cleanup: rename mst.c::__post_read_mst_fixup to post_write_mst_fixup
1208 and cleanup the code a bit, removing the unused size parameter.
1209 - Change default fmask to 0177 and update documentation.
1210 - Change attrib.c::get_attr_search_ctx() to return the search context
1211 directly instead of taking the address of a pointer. A return value
1212 of NULL means the allocation failed. Updated all callers
1214 - Update to 2.5.9 kernel (preserving backwards compatibility) by
1215 replacing all occurences of page->buffers with page_buffers(page).
1216 - Fix minor bugs in runlist merging, also minor cleanup.
1217 - Updates to bootsector layout and mft mirror contents descriptions.
1218 - Small bug fix in error detection in unistr.c and some cleanups.
1219 - Grow name buffer allocations in unistr.c in aligned mutlipled of 64
1222 2.0.1 - Minor updates.
1224 - Make default umask correspond to documentation.
1225 - Improve documentation.
1226 - Set default mode to include execute bit. The {u,f,d}mask can be used
1227 to take it away if desired. This allows binaries to be executed from
1228 a mounted ntfs partition.
1230 2.0.0 - New version number. Remove TNG from the name. Now in the kernel.
1232 - Add kill_super, just keeping up with the vfs changes in the kernel.
1233 - Repeat some changes from tng-0.0.8 that somehow got lost on the way
1234 from the CVS import into BitKeeper.
1235 - Begin to implement proper handling of allocated_size vs
1236 initialized_size vs data_size (i.e. i_size). Done are
1237 mft.c::ntfs_mft_readpage(), aops.c::end_buffer_read_index_async(),
1238 and attrib.c::load_attribute_list().
1239 - Lock the runlist in attrib.c::load_attribute_list() while using it.
1240 - Fix memory leak in ntfs_file_read_compressed_block() and generally
1241 clean up compress.c a little, removing some uncommented/unused debug
1243 - Tidy up dir.c a little bit.
1244 - Don't bother getting the runlist in inode.c::ntfs_read_inode().
1245 - Merge mft.c::ntfs_mft_readpage() and aops.c::ntfs_index_readpage()
1246 creating aops.c::ntfs_mst_readpage(), improving the handling of
1247 holes and overflow in the process and implementing the correct
1248 equivalent of ntfs_file_get_block() in ntfs_mst_readpage() itself.
1249 I am aiming for correctness at the moment. Modularisation can come
1251 - Rename aops.c::end_buffer_read_index_async() to
1252 end_buffer_read_mst_async() and optimize the overflow checking and
1254 - Use the host of the mftbmp address space mapping to hold the ntfs
1255 volume. This is needed so the async i/o completion handler can
1256 retrieve a pointer to the volume. Hopefully this will not cause
1257 problems elsewhere in the kernel... Otherwise will need to use a
1259 - Complete implementation of proper handling of allocated_size vs
1260 initialized_size vs data_size (i.e. i_size) in whole driver.
1261 Basically aops.c is now completely rewritten.
1262 - Change NTFS driver name to just NTFS and set version number to 2.0.0
1263 to make a clear distinction from the old driver which is still on
1266 tng-0.0.8 - 08/03/2002 - Now using BitKeeper, http://linux-ntfs.bkbits.net/
1268 - Replace bdevname(sb->s_dev) with sb->s_id.
1269 - Remove now superfluous new-line characters in all callers of
1271 - Apply kludge in ntfs_read_inode(), setting i_nlink to 1 for
1272 directories. Without this the "find" utility gets very upset which is
1273 fair enough as Linux/Unix do not support directory hard links.
1274 - Further runlist merging work. (Richard Russon)
1275 - Backwards compatibility for gcc-2.95. (Richard Russon)
1276 - Update to kernel 2.5.5-pre1 and rediff the now tiny patch.
1277 - Convert to new filesystem declaration using ->ntfs_get_sb() and
1278 replacing ntfs_read_super() with ntfs_fill_super().
1279 - Set s_maxbytes to MAX_LFS_FILESIZE to avoid page cache page index
1280 overflow on 32-bit architectures.
1281 - Cleanup upcase loading code to use ntfs_(un)map_page().
1282 - Disable/reenable preemtion in critical sections of compession engine.
1283 - Replace device size determination in ntfs_fill_super() with
1284 sb->s_bdev->bd_inode->i_size (in bytes) and remove now superfluous
1285 function super.c::get_nr_blocks().
1286 - Implement a mount time option (show_inodes) allowing choice of which
1287 types of inode names readdir() returns and modify ntfs_filldir()
1288 accordingly. There are several parameters to show_inodes:
1289 system: system files
1290 win32: long file names (including POSIX file names) [DEFAULT]
1292 dos: short file names only (excluding POSIX file names)
1294 posix: same as both win32 and dos
1296 Note that the options are additive, i.e. specifying:
1297 -o show_inodes=system,show_inodes=win32,show_inodes=dos
1298 is the same as specifying:
1300 Note that the "posix" and "all" options will show all directory
1301 names, BUT the link count on each directory inode entry is set to 1,
1302 due to Linux not supporting directory hard links. This may well
1303 confuse some userspace applications, since the directory names will
1304 have the same inode numbers. Thus it is NOT advisable to use the
1305 "posix" or "all" options. We provide them only for completeness sake.
1306 - Add copies of allocated_size, initialized_size, and compressed_size to
1307 the ntfs inode structure and set them up in
1308 inode.c::ntfs_read_inode(). These reflect the unnamed data attribute
1309 for files and the index allocation attribute for directories.
1310 - Add copies of allocated_size and initialized_size to ntfs inode for
1311 $BITMAP attribute of large directories and set them up in
1312 inode.c::ntfs_read_inode().
1313 - Add copies of allocated_size and initialized_size to ntfs volume for
1314 $BITMAP attribute of $MFT and set them up in
1315 super.c::load_system_files().
1316 - Parse deprecated ntfs driver options (iocharset, show_sys_files,
1317 posix, and utf8) and tell user what the new options to use are. Note
1318 we still do support them but they will be removed with kernel 2.7.x.
1319 - Change all occurences of integer long long printf formatting to hex
1320 as printk() will not support long long integer format if/when the
1321 div64 patch goes into the kernel.
1322 - Make slab caches have stable names and change the names to what they
1323 were intended to be. These changes are required/made possible by the
1324 new slab cache name handling which removes the length limitation by
1325 requiring the caller of kmem_cache_create() to supply a stable name
1326 which is then referenced but not copied.
1327 - Rename run_list structure to run_list_element and create a new
1328 run_list structure containing a pointer to a run_list_element
1329 structure and a read/write semaphore. Adapt all users of runlists
1330 to new scheme and take and release the lock as needed. This fixes a
1331 nasty race as the run_list changes even when inodes are locked for
1332 reading and even when the inode isn't locked at all, so we really
1333 needed the serialization. We use a semaphore rather than a spinlock
1334 as memory allocations can sleep and doing everything GFP_ATOMIC
1336 - Cleanup read_inode() removing all code checking for lowest_vcn != 0.
1337 This can never happen due to the nature of lookup_attr() and how we
1338 support attribute lists. If it did happen it would imply the inode
1340 - Check for lowest_vcn != 0 in ntfs_read_inode() and mark the inode as
1342 - Update to 2.5.6-pre2 changes in struct address_space.
1343 - Use parent_ino() when accessing d_parent inode number in dir.c.
1344 - Import Sourceforge CVS repository into BitKeeper repository:
1345 http://linux-ntfs.bkbits.net/ntfs-tng-2.5
1346 - Update fs/Makefile, fs/Config.help, fs/Config.in, and
1347 Documentation/filesystems/ntfs.txt for NTFS TNG.
1348 - Create kernel configuration option controlling whether debugging
1350 - Add the required export of end_buffer_io_sync() from the patches
1351 directory to the kernel code.
1352 - Update inode.c::ntfs_show_options() with show_inodes mount option.
1353 - Update errors mount option.
1355 tng-0.0.7 - 13/02/2002 - The driver is now feature complete for read-only!
1357 - Cleanup mft.c and it's debug/error output in particular. Fix a minor
1358 bug in mapping of extent inodes. Update all the comments to fit all
1359 the recent code changes.
1360 - Modify vcn_to_lcn() to cope with entirely unmapped runlists.
1361 - Cleanups in compress.c, mostly comments and folding help.
1362 - Implement attrib.c::map_run_list() as a generic helper.
1363 - Make compress.c::ntfs_file_read_compressed_block() use map_run_list()
1364 thus making code shorter and enabling attribute list support.
1365 - Cleanup incorrect use of [su]64 with %L printf format specifier in
1366 all source files. Type casts to [unsigned] long long added to correct
1367 the mismatches (important for architectures which have long long not
1369 - Merge async io completion handlers for directory indexes and $MFT
1370 data into one by setting the index_block_size{_bits} of the ntfs
1371 inode for $MFT to the mft_record_size{_bits} of the ntfs_volume.
1372 - Cleanup aops.c, update comments.
1373 - Make ntfs_file_get_block() use map_run_list() so all files now
1374 support attribute lists.
1375 - Make ntfs_dir_readpage() almost verbatim copy of
1376 block_read_full_page() by using ntfs_file_get_block() with only real
1377 difference being the use of our own async io completion handler
1378 rather than the default one, thus reducing the amount of code and
1379 automatically enabling attribute list support for directory indices.
1380 - Fix bug in load_attribute_list() - forgot to call brelse in error
1382 - Change parameters to find_attr() and lookup_attr(). We no longer
1383 pass in the upcase table and its length. These can be gotten from
1384 ctx->ntfs_ino->vol->upcase{_len}. Update all callers.
1385 - Cleanups in attrib.c.
1386 - Implement merging of runlists, attrib.c::merge_run_lists() and its
1387 helpers. (Richard Russon)
1388 - Attribute lists part 2, attribute extents and multi part runlists:
1389 enable proper support for LCN_RL_NOT_MAPPED and automatic mapping of
1390 further runlist parts via attrib.c::map_run_list().
1391 - Tiny endianness bug fix in decompress_mapping_pairs().
1393 tng-0.0.6 - Encrypted directories, bug fixes, cleanups, debugging enhancements.
1395 - Enable encrypted directories. (Their index root is marked encrypted
1396 to indicate that new files in that directory should be created
1398 - Fix bug in NInoBmpNonResident() macro. (Cut and paste error.)
1399 - Enable $Extend system directory. Most (if not all) extended system
1400 files do not have unnamed data attributes so ntfs_read_inode() had to
1401 special case them but that is ok, as the special casing recovery
1402 happens inside an error code path so there is zero slow down in the
1403 normal fast path. The special casing is done by introducing a new
1404 function inode.c::ntfs_is_extended_system_file() which checks if any
1405 of the hard links in the inode point to $Extend as being their parent
1406 directory and if they do we assume this is an extended system file.
1407 - Create a sysctl/proc interface to allow {dis,en}abling of debug output
1408 when compiled with -DDEBUG. Default is debug messages to be disabled.
1409 To enable them, one writes a non-zero value to /proc/sys/fs/ntfs-debug
1410 (if /proc is enabled) or uses sysctl(2) to effect the same (if sysctl
1411 interface is enabled). Inspired by old ntfs driver.
1412 - Add debug_msgs insmod/kernel boot parameter to set whether debug
1413 messages are {dis,en}abled. This is useful to enable debug messages
1414 during ntfs initialization and is the only way to activate debugging
1415 when the sysctl interface is not enabled.
1416 - Cleanup debug output in various places.
1417 - Remove all dollar signs ($) from the source (except comments) to
1418 enable compilation on architectures whose gcc compiler does not
1419 support dollar signs in the names of variables/constants. Attribute
1420 types now start with AT_ instead of $ and $I30 is now just I30.
1421 - Cleanup ntfs_lookup() and add consistency check of sequence numbers.
1422 - Load complete runlist for $MFT/$BITMAP during mount and cleanup
1423 access functions. This means we now cope with $MFT/$BITMAP being
1424 spread accross several mft records.
1425 - Disable modification of mft_zone_multiplier on remount. We can always
1426 reenable this later on if we really want to, but we will need to make
1427 sure we readjust the mft_zone size / layout accordingly.
1429 tng-0.0.5 - Modernize for 2.5.x and further in line-ing with Al Viro's comments.
1431 - Use sb_set_blocksize() instead of set_blocksize() and verify the
1433 - Use sb_bread() instead of bread() throughout.
1434 - Add index_vcn_size{_bits} to ntfs_inode structure to store the size
1435 of a directory index block vcn. Apply resulting simplifications in
1437 - Fix a small bug somewhere (but forgot what it was).
1438 - Change ntfs_{debug,error,warning} to enable gcc to do type checking
1439 on the printf-format parameter list and fix bugs reported by gcc
1440 as a result. (Richard Russon)
1441 - Move inode allocation strategy to Al's new stuff but maintain the
1442 divorce of ntfs_inode from struct inode. To achieve this we have two
1443 separate slab caches, one for big ntfs inodes containing a struct
1444 inode and pure ntfs inodes and at the same time fix some faulty
1445 error code paths in ntfs_read_inode().
1446 - Show mount options in proc (inode.c::ntfs_show_options()).
1448 tng-0.0.4 - Big changes, getting in line with Al Viro's comments.
1450 - Modified (un)map_mft_record functions to be common for read and write
1451 case. To specify which is which, added extra parameter at front of
1452 parameter list. Pass either READ or WRITE to this, each has the
1454 - General cleanups to allow for easier folding in vi.
1455 - attrib.c::decompress_mapping_pairs() now accepts the old runlist
1456 argument, and invokes attrib.c::merge_run_lists() to merge the old
1457 and the new runlists.
1458 - Removed attrib.c::find_first_attr().
1459 - Implemented loading of attribute list and complete runlist for $MFT.
1460 This means we now cope with $MFT being spread across several mft
1462 - Adapt to 2.5.2-pre9 and the changed create_empty_buffers() syntax.
1463 - Adapt major/minor/kdev_t/[bk]devname stuff to new 2.5.x kernels.
1464 - Make ntfs_volume be allocated via kmalloc() instead of using a slab
1465 cache. There are too little ntfs_volume structures at any one time
1466 to justify a private slab cache.
1467 - Fix bogus kmap() use in async io completion. Now use kmap_atomic().
1468 Use KM_BIO_IRQ on advice from IRC/kernel...
1469 - Use ntfs_map_page() in map_mft_record() and create ->readpage method
1470 for reading $MFT (ntfs_mft_readpage). In the process create dedicated
1471 address space operations (ntfs_mft_aops) for $MFT inode mapping. Also
1472 removed the now superfluous exports from the kernel core patch.
1473 - Fix a bug where kfree() was used insted of ntfs_free().
1474 - Change map_mft_record() to take ntfs_inode as argument instead of
1475 vfs inode. Dito for unmap_mft_record(). Adapt all callers.
1476 - Add pointer to ntfs_volume to ntfs_inode.
1477 - Add mft record number and sequence number to ntfs_inode. Stop using
1478 i_ino and i_generation for in-driver purposes.
1479 - Implement attrib.c::merge_run_lists(). (Richard Russon)
1480 - Remove use of proper inodes by extent inodes. Move i_ino and
1481 i_generation to ntfs_inode to do this. Apply simplifications that
1482 result and remove iget_no_wait(), etc.
1483 - Pass ntfs_inode everywhere in the driver (used to be struct inode).
1484 - Add reference counting in ntfs_inode for the ntfs inode itself and
1485 for the mapped mft record.
1486 - Extend mft record mapping so we can (un)map extent mft records (new
1487 functions (un)map_extent_mft_record), and so mappings are reference
1488 counted and don't have to happen twice if already mapped - just ref
1490 - Add -o iocharset as alias to -o nls for backwards compatibility.
1491 - The latest core patch is now tiny. In fact just a single additional
1492 export is necessary over the base kernel.
1494 tng-0.0.3 - Cleanups, enhancements, bug fixes.
1496 - Work on attrib.c::decompress_mapping_pairs() to detect base extents
1497 and setup the runlist appropriately using knowledge provided by the
1498 sizes in the base attribute record.
1499 - Balance the get_/put_attr_search_ctx() calls so we don't leak memory
1501 - Introduce ntfs_malloc_nofs() and ntfs_free() to allocate/free a single
1502 page or use vmalloc depending on the amount of memory requested.
1503 - Cleanup error output. The __FUNCTION__ "(): " is now added
1504 automatically. Introduced a new header file debug.h to support this
1505 and also moved ntfs_debug() function into it.
1506 - Make reading of compressed files more intelligent and especially get
1507 rid of the vmalloc_nofs() from readpage(). This now uses per CPU
1508 buffers (allocated at first mount with cluster size <= 4kiB and
1509 deallocated on last umount with cluster size <= 4kiB), and
1510 asynchronous io for the compressed data using a list of buffer heads.
1511 Er, we use synchronous io as async io only works on whole pages
1512 covered by buffers and not on individual buffer heads...
1513 - Bug fix for reading compressed files with sparse compression blocks.
1515 tng-0.0.2 - Now handles larger/fragmented/compressed volumes/files/dirs.
1517 - Fixed handling of directories when cluster size exceeds index block
1519 - Hide DOS only name space directory entries from readdir() but allow
1520 them in lookup(). This should fix the problem that Linux doesn't
1521 support directory hard links, while still allowing access to entries
1522 via their short file name. This also has the benefit of mimicking
1523 what Windows users are used to, so it is the ideal solution.
1524 - Implemented sync_page everywhere so no more hangs in D state when
1526 - Stop using bforget() in favour of brelse().
1527 - Stop locking buffers unnecessarily.
1528 - Implemented compressed files (inode->mapping contains uncompressed
1529 data, raw compressed data is currently bread() into a vmalloc()ed
1531 - Enable compressed directories. (Their index root is marked compressed
1532 to indicate that new files in that directory should be created
1534 - Use vsnprintf rather than vsprintf in the ntfs_error and ntfs_warning
1535 functions. (Thanks to Will Dyson for pointing this out.)
1536 - Moved the ntfs_inode and ntfs_volume (the former ntfs_inode_info and
1537 ntfs_sb_info) out of the common inode and super_block structures and
1538 started using the generic_ip and generic_sbp pointers instead. This
1539 makes ntfs entirely private with respect to the kernel tree.
1540 - Detect compiler version and abort with error message if gcc less than
1542 - Fix bug in name comparison function in unistr.c.
1543 - Implement attribute lists part 1, the infrastructure: search contexts
1544 and operations, find_external_attr(), lookup_attr()) and make the
1545 code use the infrastructure.
1546 - Fix stupid buffer overflow bug that became apparent on larger run
1547 list containing attributes.
1548 - Fix bugs in readdir() that became apparent on larger directories.
1550 The driver is now really useful and survives the test
1551 find . -type f -exec md5sum "{}" \;
1552 without any error messages on a over 1GiB sized partition with >16k
1553 files on it, including compressed files and directories and many files
1554 and directories with attribute lists.
1556 tng-0.0.1 - The first useful version.
1558 - Added ntfs_lookup().
1559 - Added default upcase generation and handling.
1560 - Added compile options to be shown on module init.
1561 - Many bug fixes that were "hidden" before.
1562 - Update to latest kernel.
1563 - Added ntfs_readdir().
1564 - Added file operations for mmap(), read(), open() and llseek(). We just
1565 use the generic ones. The whole point of going through implementing
1566 readpage() methods and where possible get_block() call backs is that
1567 this allows us to make use of the generic high level methods provided
1570 The driver is now actually useful! Yey. (-: It undoubtedly has got bugs
1571 though and it doesn't implement accesssing compressed files yet. Also,
1572 accessing files with attribute list attributes is not implemented yet
1573 either. But for small or simple filesystems it should work and allow
1574 you to list directories, use stat on directory entries and the file
1575 system, open, read, mmap and llseek around in files. A big mile stone
1578 tng-0.0.0 - Initial version tag.
1580 Initial driver implementation. The driver can mount and umount simple
1581 NTFS filesystems (i.e. ones without attribute lists in the system
1582 files). If the mount fails there might be problems in the error handling
1583 code paths, so be warned. Otherwise it seems to be loading the system
1584 files nicely and the mft record read mapping/unmapping seems to be
1585 working nicely, too. Proof of inode metadata in the page cache and non-
1586 resident file unnamed stream data in the page cache concepts is thus