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(), ntfs_attr_find_vcn_nolock() and
28 {__,}ntfs_cluster_free() to also take an optional attribute search
29 context as argument. This allows calling these functions with the
30 mft record mapped. Update all callers.
31 - Fix potential deadlock in ntfs_mft_data_extend_allocation_nolock()
32 error handling by passing in the active search context when calling
34 - Change ntfs_cluster_alloc() to take an extra boolean parameter
35 specifying whether the cluster are being allocated to extend an
36 attribute or to fill a hole.
37 - Change ntfs_attr_make_non_resident() to call ntfs_cluster_alloc()
38 with @is_extension set to TRUE and remove the runlist terminator
39 fixup code as this is now done by ntfs_cluster_alloc().
41 2.1.24 - Lots of bug fixes and support more clean journal states.
43 - Support journals ($LogFile) which have been modified by chkdsk. This
44 means users can boot into Windows after we marked the volume dirty.
45 The Windows boot will run chkdsk and then reboot. The user can then
46 immediately boot into Linux rather than having to do a full Windows
47 boot first before rebooting into Linux and we will recognize such a
48 journal and empty it as it is clean by definition. Note, this only
49 works if chkdsk left the journal in an obviously clean state.
50 - Support journals ($LogFile) with only one restart page as well as
51 journals with two different restart pages. We sanity check both and
52 either use the only sane one or the more recent one of the two in the
53 case that both are valid.
54 - Add fs/ntfs/malloc.h::ntfs_malloc_nofs_nofail() which is analogous to
55 ntfs_malloc_nofs() but it performs allocations with __GFP_NOFAIL and
57 - Use ntfs_malloc_nofs_nofail() in the two critical regions in
58 fs/ntfs/runlist.c::ntfs_runlists_merge(). This means we no longer
59 need to panic() if the allocation fails as it now cannot fail.
60 - Fix two nasty runlist merging bugs that had gone unnoticed so far.
61 Thanks to Stefano Picerno for the bug report.
62 - Remove two bogus BUG_ON()s from fs/ntfs/mft.c.
63 - Fix handling of valid but empty mapping pairs array in
64 fs/ntfs/runlist.c::ntfs_mapping_pairs_decompress().
65 - Report unrepresentable inodes during ntfs_readdir() as KERN_WARNING
66 messages and include the inode number. Thanks to Yura Pakhuchiy for
68 - Change ntfs_rl_truncate_nolock() to throw away the runlist if the new
70 - Add runlist.[hc]::ntfs_rl_punch_nolock() which punches a caller
71 specified hole into a runlist.
72 - Fix a bug in fs/ntfs/index.c::ntfs_index_lookup(). When the returned
73 index entry is in the index root, we forgot to set the @ir pointer in
74 the index context. Thanks to Yura Pakhuchiy for finding this bug.
75 - Remove bogus setting of PageError in ntfs_read_compressed_block().
76 - Add fs/ntfs/attrib.[hc]::ntfs_resident_attr_value_resize().
77 - Fix a bug in ntfs_map_runlist_nolock() where we forgot to protect
78 access to the allocated size in the ntfs inode with the size lock.
79 - Fix ntfs_attr_vcn_to_lcn_nolock() and ntfs_attr_find_vcn_nolock() to
80 return LCN_ENOENT when there is no runlist and the allocated size is
82 - Fix load_attribute_list() to handle the case of a NULL runlist.
83 - Fix handling of sparse attributes in ntfs_attr_make_non_resident().
84 - Add BUG() checks to ntfs_attr_make_non_resident() and ntfs_attr_set()
85 to ensure that these functions are never called for compressed or
87 - Fix cluster (de)allocators to work when the runlist is NULL and more
88 importantly to take a locked runlist rather than them locking it
89 which leads to lock reversal.
90 - Truncate {a,c,m}time to the ntfs supported time granularity when
91 updating the times in the inode in ntfs_setattr().
92 - Fixup handling of sparse, compressed, and encrypted attributes in
93 fs/ntfs/inode.c::ntfs_read_locked_{,attr_,index_}inode(),
94 fs/ntfs/aops.c::ntfs_{read,write}page().
95 - Make ntfs_write_block() not instantiate sparse blocks if they contain
97 - Optimize fs/ntfs/aops.c::ntfs_write_block() by extending the page
98 lock protection over the buffer submission for i/o which allows the
99 removal of the get_bh()/put_bh() pairs for each buffer.
100 - Fix fs/ntfs/aops.c::ntfs_{read,write}_block() to handle the case
101 where a concurrent truncate has truncated the runlist under our feet.
102 - Fix page_has_buffers()/page_buffers() handling in fs/ntfs/aops.c.
103 - In fs/ntfs/aops.c::ntfs_end_buffer_async_read(), use a bit spin lock
104 in the first buffer head instead of a driver global spin lock to
106 - Minor fix to error handling and error message display in
107 fs/ntfs/aops.c::ntfs_prepare_nonresident_write().
108 - Change the mount options {u,f,d}mask to always parse the number as
109 an octal number to conform to how chmod(1) works, too. Thanks to
110 Giuseppe Bilotta and Horst von Brand for pointing out the errors of
112 - Fix various bugs in the runlist merging code. (Based on libntfs
113 changes by Richard Russon.)
114 - Fix sparse warnings that have crept in over time.
115 - Change ntfs_cluster_free() to require a write locked runlist on entry
116 since we otherwise get into a lock reversal deadlock if a read locked
117 runlist is passed in. In the process also change it to take an ntfs
118 inode instead of a vfs inode as parameter.
119 - Fix the definition of the CHKD ntfs record magic. It had an off by
120 two error causing it to be CHKB instead of CHKD.
121 - Fix a stupid bug in __ntfs_bitmap_set_bits_in_run() which caused the
122 count to become negative and hence we had a wild memset() scribbling
123 all over the system's ram.
125 2.1.23 - Implement extension of resident files and make writing safe as well as
126 many bug fixes, cleanups, and enhancements...
128 - Add printk rate limiting for ntfs_warning() and ntfs_error() when
129 compiled without debug. This avoids a possible denial of service
130 attack. Thanks to Carl-Daniel Hailfinger from SuSE for pointing this
132 - Fix compilation warnings on ia64. (Randy Dunlap)
133 - Use i_size_{read,write}() instead of reading i_size by hand and cache
134 the value where apropriate.
135 - Add size_lock to the ntfs_inode structure. This is an rw spinlock
136 and it locks against access to the inode sizes. Note, ->size_lock
137 is also accessed from irq context so you must use the _irqsave and
138 _irqrestore lock and unlock functions, respectively. Protect all
139 accesses to allocated_size, initialized_size, and compressed_size.
140 - Minor optimization to fs/ntfs/super.c::ntfs_statfs() and its helpers.
141 - Implement extension of resident files in the regular file write code
142 paths (fs/ntfs/aops.c::ntfs_{prepare,commit}_write()). At present
143 this only works until the data attribute becomes too big for the mft
144 record after which we abort the write returning -EOPNOTSUPP from
145 ntfs_prepare_write().
146 - Add disable_sparse mount option together with a per volume sparse
147 enable bit which is set appropriately and a per inode sparse disable
148 bit which is preset on some system file inodes as appropriate.
149 - Enforce that sparse support is disabled on NTFS volumes pre 3.0.
150 - Fix a bug in fs/ntfs/runlist.c::ntfs_mapping_pairs_decompress() in
151 the creation of the unmapped runlist element for the base attribute
153 - Split ntfs_map_runlist() into ntfs_map_runlist() and a non-locking
154 helper ntfs_map_runlist_nolock() which is used by ntfs_map_runlist().
155 This allows us to map runlist fragments with the runlist lock already
156 held without having to drop and reacquire it around the call. Adapt
158 - Change ntfs_find_vcn() to ntfs_find_vcn_nolock() which takes a locked
159 runlist. This allows us to find runlist elements with the runlist
160 lock already held without having to drop and reacquire it around the
161 call. Adapt all callers.
162 - Change time to u64 in time.h::ntfs2utc() as it otherwise generates a
163 warning in the do_div() call on sparc32. Thanks to Meelis Roos for
164 the report and analysis of the warning.
165 - Fix a nasty runlist merge bug when merging two holes.
166 - Set the ntfs_inode->allocated_size to the real allocated size in the
167 mft record for resident attributes (fs/ntfs/inode.c).
168 - Small readability cleanup to use "a" instead of "ctx->attr"
169 everywhere (fs/ntfs/inode.c).
170 - Make fs/ntfs/namei.c::ntfs_get_{parent,dentry} static and move the
171 definition of ntfs_export_ops from fs/ntfs/super.c to namei.c. Also,
172 declare ntfs_export_ops in fs/ntfs/ntfs.h.
173 - Correct sparse file handling. The compressed values need to be
174 checked and set in the ntfs inode as done for compressed files and
175 the compressed size needs to be used for vfs inode->i_blocks instead
176 of the allocated size, again, as done for compressed files.
177 - Add AT_EA in addition to AT_DATA to whitelist for being allowed to be
178 non-resident in fs/ntfs/attrib.c::ntfs_attr_can_be_non_resident().
179 - Add fs/ntfs/attrib.c::ntfs_attr_vcn_to_lcn_nolock() used by the new
181 - Fix bug in fs/ntfs/attrib.c::ntfs_find_vcn_nolock() where after
182 dropping the read lock and taking the write lock we were not checking
183 whether someone else did not already do the work we wanted to do.
184 - Rename fs/ntfs/attrib.c::ntfs_find_vcn_nolock() to
185 ntfs_attr_find_vcn_nolock() and update all callers.
186 - Add fs/ntfs/attrib.[hc]::ntfs_attr_make_non_resident().
187 - Fix sign of various error return values to be negative in
189 - Modify ->readpage and ->writepage (fs/ntfs/aops.c) so they detect and
190 handle the case where an attribute is converted from resident to
191 non-resident by a concurrent file write.
192 - Remove checks for NULL before calling kfree() since kfree() does the
193 checking itself. (Jesper Juhl)
194 - Some utilities modify the boot sector but do not update the checksum.
195 Thus, relax the checking in fs/ntfs/super.c::is_boot_sector_ntfs() to
196 only emit a warning when the checksum is incorrect rather than
197 refusing the mount. Thanks to Bernd Casimir for pointing this
199 - Update attribute definition handling.
200 - Add NTFS_MAX_CLUSTER_SIZE and NTFS_MAX_PAGES_PER_CLUSTER constants.
201 - Use NTFS_MAX_CLUSTER_SIZE in super.c instead of hard coding 0x10000.
202 - Use MAX_BUF_PER_PAGE instead of variable sized array allocation for
203 better code generation and one less sparse warning in fs/ntfs/aops.c.
204 - Remove spurious void pointer casts from fs/ntfs/. (Pekka Enberg)
205 - Use C99 style structure initialization after memory allocation where
206 possible (fs/ntfs/{attrib.c,index.c,super.c}). Thanks to Al Viro and
208 - Stamp the transaction log ($UsnJrnl), aka user space journal, if it
209 is active on the volume and we are mounting read-write or remounting
210 from read-only to read-write.
211 - Fix a bug in address space operations error recovery code paths where
212 if the runlist was not mapped at all and a mapping error occured we
213 would leave the runlist locked on exit to the function so that the
214 next access to the same file would try to take the lock and deadlock.
215 - Detect the case when Windows has been suspended to disk on the volume
216 to be mounted and if this is the case do not allow (re)mounting
217 read-write. This is done by parsing hiberfil.sys if present.
218 - Fix several occurences of a bug where we would perform 'var & ~const'
219 with a 64-bit variable and a int, i.e. 32-bit, constant. This causes
220 the higher order 32-bits of the 64-bit variable to be zeroed. To fix
221 this cast the 'const' to the same 64-bit type as 'var'.
222 - Change the runlist terminator of the newly allocated cluster(s) to
223 LCN_ENOENT in ntfs_attr_make_non_resident(). Otherwise the runlist
225 - Add an extra parameter @last_vcn to ntfs_get_size_for_mapping_pairs()
226 and ntfs_mapping_pairs_build() to allow the runlist encoding to be
227 partial which is desirable when filling holes in sparse attributes.
229 - Change ntfs_map_runlist_nolock() to only decompress the mapping pairs
230 if the requested vcn is inside it. Otherwise we get into problems
231 when we try to map an out of bounds vcn because we then try to map
232 the already mapped runlist fragment which causes
233 ntfs_mapping_pairs_decompress() to fail and return error. Update
234 ntfs_attr_find_vcn_nolock() accordingly.
235 - Fix a nasty deadlock that appeared in recent kernels.
236 The situation: VFS inode X on a mounted ntfs volume is dirty. For
237 same inode X, the ntfs_inode is dirty and thus corresponding on-disk
238 inode, i.e. mft record, which is in a dirty PAGE_CACHE_PAGE belonging
239 to the table of inodes, i.e. $MFT, inode 0.
241 Process 1: sys_sync()/umount()/whatever... calls
242 __sync_single_inode() for $MFT -> do_writepages() -> write_page for
243 the dirty page containing the on-disk inode X, the page is now locked
244 -> ntfs_write_mst_block() which clears PageUptodate() on the page to
245 prevent anyone else getting hold of it whilst it does the write out.
246 This is necessary as the on-disk inode needs "fixups" applied before
247 the write to disk which are removed again after the write and
248 PageUptodate is then set again. It then analyses the page looking
249 for dirty on-disk inodes and when it finds one it calls
250 ntfs_may_write_mft_record() to see if it is safe to write this
251 on-disk inode. This then calls ilookup5() to check if the
252 corresponding VFS inode is in icache(). This in turn calls ifind()
253 which waits on the inode lock via wait_on_inode whilst holding the
255 Process 2: pdflush results in a call to __sync_single_inode for the
256 same VFS inode X on the ntfs volume. This locks the inode (I_LOCK)
257 then calls write-inode -> ntfs_write_inode -> map_mft_record() ->
258 read_cache_page() for the page (in page cache of table of inodes
259 $MFT, inode 0) containing the on-disk inode. This page has
260 PageUptodate() clear because of Process 1 (see above) so
261 read_cache_page() blocks when it tries to take the page lock for the
262 page so it can call ntfs_read_page().
263 Thus Process 1 is holding the page lock on the page containing the
264 on-disk inode X and it is waiting on the inode X to be unlocked in
265 ifind() so it can write the page out and then unlock the page.
266 And Process 2 is holding the inode lock on inode X and is waiting for
267 the page to be unlocked so it can call ntfs_readpage() or discover
268 that Process 1 set PageUptodate() again and use the page.
269 Thus we have a deadlock due to ifind() waiting on the inode lock.
270 The solution: The fix is to use the newly introduced
271 ilookup5_nowait() which does not wait on the inode's lock and hence
272 avoids the deadlock. This is safe as we do not care about the VFS
273 inode and only use the fact that it is in the VFS inode cache and the
274 fact that the vfs and ntfs inodes are one struct in memory to find
275 the ntfs inode in memory if present. Also, the ntfs inode has its
276 own locking so it does not matter if the vfs inode is locked.
277 - Fix bug in mft record writing where we forgot to set the device in
278 the buffers when mapping them after the VM had discarded them.
279 Thanks to Martin MOKREJŠ for the bug report.
281 2.1.22 - Many bug and race fixes and error handling improvements.
283 - Improve error handling in fs/ntfs/inode.c::ntfs_truncate().
284 - Change fs/ntfs/inode.c::ntfs_truncate() to return an error code
285 instead of void and provide a helper ntfs_truncate_vfs() for the
286 vfs ->truncate method.
287 - Add a new ntfs inode flag NInoTruncateFailed() and modify
288 fs/ntfs/inode.c::ntfs_truncate() to set and clear it appropriately.
289 - Fix min_size and max_size definitions in ATTR_DEF structure in
290 fs/ntfs/layout.h to be signed.
291 - Add attribute definition handling helpers to fs/ntfs/attrib.[hc]:
292 ntfs_attr_size_bounds_check(), ntfs_attr_can_be_non_resident(), and
293 ntfs_attr_can_be_resident(), which in turn use the new private helper
294 ntfs_attr_find_in_attrdef().
295 - In fs/ntfs/aops.c::mark_ntfs_record_dirty(), take the
296 mapping->private_lock around the dirtying of the buffer heads
297 analagous to the way it is done in __set_page_dirty_buffers().
298 - Ensure the mft record size does not exceed the PAGE_CACHE_SIZE at
299 mount time as this cannot work with the current implementation.
300 - Check for location of attribute name and improve error handling in
301 general in fs/ntfs/inode.c::ntfs_read_locked_inode() and friends.
302 - In fs/ntfs/aops.c::ntfs_writepage(), if the page is fully outside
303 i_size, i.e. race with truncate, invalidate the buffers on the page
304 so that they become freeable and hence the page does not leak.
305 - Remove unused function fs/ntfs/runlist.c::ntfs_rl_merge(). (Adrian
307 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_find() that resulted in
308 a NULL pointer dereference in the error code path when a corrupt
309 attribute was found. (Thanks to Domen Puncer for the bug report.)
310 - Add MODULE_VERSION() to fs/ntfs/super.c.
311 - Make several functions and variables static. (Adrian Bunk)
312 - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() so it allocates
313 buffers for the page if they are not present and then marks the
314 buffers belonging to the ntfs record dirty. This causes the buffers
315 to become busy and hence they are safe from removal until the page
316 has been written out.
317 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find() in the
318 error handling code path that resulted in a BUG() due to trying to
319 unmap an extent mft record when the mapping of it had failed and it
320 thus was not mapped. (Thanks to Ken MacFerrin for the bug report.)
321 - Drop the runlist lock after the vcn has been read in
322 fs/ntfs/lcnalloc.c::__ntfs_cluster_free().
323 - Rewrite handling of multi sector transfer errors. We now do not set
324 PageError() when such errors are detected in the async i/o handler
325 fs/ntfs/aops.c::ntfs_end_buffer_async_read(). All users of mst
326 protected attributes now check the magic of each ntfs record as they
327 use it and act appropriately. This has the effect of making errors
328 granular per ntfs record rather than per page which solves the case
329 where we cannot access any of the ntfs records in a page when a
330 single one of them had an mst error. (Thanks to Ken MacFerrin for
332 - Fix error handling in fs/ntfs/quota.c::ntfs_mark_quotas_out_of_date()
333 where we failed to release i_sem on the $Quota/$Q attribute inode.
334 - Fix bug in handling of bad inodes in fs/ntfs/namei.c::ntfs_lookup().
335 - Add mapping of unmapped buffers to all remaining code paths, i.e.
336 fs/ntfs/aops.c::ntfs_write_mst_block(), mft.c::ntfs_sync_mft_mirror(),
337 and write_mft_record_nolock(). From now on we require that the
338 complete runlist for the mft mirror is always mapped into memory.
339 - Add creation of buffers to fs/ntfs/mft.c::ntfs_sync_mft_mirror().
340 - Improve error handling in fs/ntfs/aops.c::ntfs_{read,write}_block().
341 - Cleanup fs/ntfs/aops.c::ntfs_{read,write}page() since we know that a
342 resident attribute will be smaller than a page which makes the code
343 simpler. Also make the code more tolerant to concurrent ->truncate.
345 2.1.21 - Fix some races and bugs, rewrite mft write code, add mft allocator.
347 - Implement extent mft record deallocation
348 fs/ntfs/mft.c::ntfs_extent_mft_record_free().
349 - Splitt runlist related functions off from attrib.[hc] to runlist.[hc].
350 - Add vol->mft_data_pos and initialize it at mount time.
351 - Rename init_runlist() to ntfs_init_runlist(), ntfs_vcn_to_lcn() to
352 ntfs_rl_vcn_to_lcn(), decompress_mapping_pairs() to
353 ntfs_mapping_pairs_decompress(), ntfs_merge_runlists() to
354 ntfs_runlists_merge() and adapt all callers.
355 - Add fs/ntfs/runlist.[hc]::ntfs_get_nr_significant_bytes(),
356 ntfs_get_size_for_mapping_pairs(), ntfs_write_significant_bytes(),
357 and ntfs_mapping_pairs_build(), adapted from libntfs.
358 - Make fs/ntfs/lcnalloc.c::ntfs_cluster_free_from_rl_nolock() not
359 static and add a declaration for it to lcnalloc.h.
360 - Add fs/ntfs/lcnalloc.h::ntfs_cluster_free_from_rl() which is a static
361 inline wrapper for ntfs_cluster_free_from_rl_nolock() which takes the
362 cluster bitmap lock for the duration of the call.
363 - Add fs/ntfs/attrib.[hc]::ntfs_attr_record_resize().
364 - Implement the equivalent of memset() for an ntfs attribute in
365 fs/ntfs/attrib.[hc]::ntfs_attr_set() and switch
366 fs/ntfs/logfile.c::ntfs_empty_logfile() to using it.
367 - Remove unnecessary casts from LCN_* constants.
368 - Implement fs/ntfs/runlist.c::ntfs_rl_truncate_nolock().
369 - Add MFT_RECORD_OLD as a copy of MFT_RECORD in fs/ntfs/layout.h and
370 change MFT_RECORD to contain the NTFS 3.1+ specific fields.
371 - Add a helper function fs/ntfs/aops.c::mark_ntfs_record_dirty() which
372 marks all buffers belonging to an ntfs record dirty, followed by
373 marking the page the ntfs record is in dirty and also marking the vfs
374 inode containing the ntfs record dirty (I_DIRTY_PAGES).
375 - Switch fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to using the
376 new helper fs/ntfs/aops.c::mark_ntfs_record_dirty() and remove the no
377 longer needed fs/ntfs/index.[hc]::__ntfs_index_entry_mark_dirty().
378 - Move ntfs_{un,}map_page() from ntfs.h to aops.h and fix resulting
380 - Move the typedefs for runlist_element and runlist from types.h to
381 runlist.h and fix resulting include errors.
382 - Remove unused {__,}format_mft_record() from fs/ntfs/mft.c.
383 - Modify fs/ntfs/mft.c::__mark_mft_record_dirty() to use the helper
384 mark_ntfs_record_dirty() which also changes the behaviour in that we
385 now set the buffers belonging to the mft record dirty as well as the
387 - Update fs/ntfs/mft.c::write_mft_record_nolock() and sync_mft_mirror()
388 to cope with the fact that there now are dirty buffers in mft pages.
389 - Update fs/ntfs/inode.c::ntfs_write_inode() to also use the helper
390 mark_ntfs_record_dirty() and thus to set the buffers belonging to the
391 mft record dirty as well as the page itself.
392 - Fix compiler warnings on x86-64 in fs/ntfs/dir.c. (Randy Dunlap,
393 slightly modified by me)
394 - Add fs/ntfs/mft.c::try_map_mft_record() which fails with -EALREADY if
395 the mft record is already locked and otherwise behaves the same way
396 as fs/ntfs/mft.c::map_mft_record().
397 - Modify fs/ntfs/mft.c::write_mft_record_nolock() so that it only
398 writes the mft record if the buffers belonging to it are dirty.
399 Otherwise we assume that it was written out by other means already.
400 - Attempting to write outside initialized size is _not_ a bug so remove
401 the bug check from fs/ntfs/aops.c::ntfs_write_mst_block(). It is in
402 fact required to write outside initialized size when preparing to
403 extend the initialized size.
404 - Map the page instead of using page_address() before writing to it in
405 fs/ntfs/aops.c::ntfs_mft_writepage().
406 - Provide exclusion between opening an inode / mapping an mft record
407 and accessing the mft record in fs/ntfs/mft.c::ntfs_mft_writepage()
408 by setting the page not uptodate throughout ntfs_mft_writepage().
409 - Clear the page uptodate flag in fs/ntfs/aops.c::ntfs_write_mst_block()
410 to ensure noone can see the page whilst the mst fixups are applied.
411 - Add the helper fs/ntfs/mft.c::ntfs_may_write_mft_record() which
412 checks if an mft record may be written out safely obtaining any
413 necessary locks in the process. This is used by
414 fs/ntfs/aops.c::ntfs_write_mst_block().
415 - Modify fs/ntfs/aops.c::ntfs_write_mst_block() to also work for
416 writing mft records and improve its error handling in the process.
417 Now if any of the records in the page fail to be written out, all
418 other records will be written out instead of aborting completely.
419 - Remove ntfs_mft_aops and update all users to use ntfs_mst_aops.
420 - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to set the
421 ntfs_mst_aops for all inodes which are NInoMstProtected() and
422 ntfs_aops for all other inodes.
423 - Rename fs/ntfs/mft.c::sync_mft_mirror{,_umount}() to
424 ntfs_sync_mft_mirror{,_umount}() and change their parameters so they
425 no longer require an ntfs inode to be present. Update all callers.
426 - Cleanup the error handling in fs/ntfs/mft.c::ntfs_sync_mft_mirror().
427 - Clear the page uptodate flag in fs/ntfs/mft.c::ntfs_sync_mft_mirror()
428 to ensure noone can see the page whilst the mst fixups are applied.
429 - Remove the no longer needed fs/ntfs/mft.c::ntfs_mft_writepage() and
430 fs/ntfs/mft.c::try_map_mft_record().
431 - Fix callers of fs/ntfs/aops.c::mark_ntfs_record_dirty() to call it
432 with the ntfs inode which contains the page rather than the ntfs
433 inode the mft record of which is in the page.
434 - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by moving the
435 index inode bitmap inode release code from there to
436 fs/ntfs/inode.c::ntfs_clear_big_inode(). (Thanks to Christoph
437 Hellwig for spotting this.)
438 - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by taking the
439 inode semaphore around the code that sets ni->itype.index.bmp_ino to
440 NULL and reorganize the code to optimize it a bit. (Thanks to
441 Christoph Hellwig for spotting this.)
442 - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() to no longer take the
443 ntfs inode as a parameter as this is confusing and misleading and the
444 needed ntfs inode is available via NTFS_I(page->mapping->host).
445 Adapt all callers to this change.
446 - Modify fs/ntfs/mft.c::write_mft_record_nolock() and
447 fs/ntfs/aops.c::ntfs_write_mst_block() to only check the dirty state
448 of the first buffer in a record and to take this as the ntfs record
449 dirty state. We cannot look at the dirty state for subsequent
450 buffers because we might be racing with
451 fs/ntfs/aops.c::mark_ntfs_record_dirty().
452 - Move the static inline ntfs_init_big_inode() from fs/ntfs/inode.c to
453 inode.h and make fs/ntfs/inode.c::__ntfs_init_inode() non-static and
454 add a declaration for it to inode.h. Fix some compilation issues
455 that resulted due to #includes and header file interdependencies.
456 - Simplify setup of i_mode in fs/ntfs/inode.c::ntfs_read_locked_inode().
457 - Add helpers fs/ntfs/layout.h::MK_MREF() and MK_LE_MREF().
458 - Modify fs/ntfs/mft.c::map_extent_mft_record() to only verify the mft
459 record sequence number if it is specified (i.e. not zero).
460 - Add fs/ntfs/mft.[hc]::ntfs_mft_record_alloc() and various helper
461 functions used by it.
462 - Update Documentation/filesystems/ntfs.txt with instructions on how to
463 use the Device-Mapper driver with NTFS ftdisk/LDM raid. This removes
464 the linear raid problem with the Software RAID / MD driver when one
465 or more of the devices has an odd number of sectors.
467 2.1.20 - Fix two stupid bugs introduced in 2.1.18 release.
469 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_reinit_search_ctx()
470 where we did not clear ctx->al_entry but it was still set due to
471 changes in ntfs_attr_lookup() and ntfs_external_attr_find() in
473 - Fix another stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find()
474 where we forgot to unmap the extent mft record when we had finished
475 enumerating an attribute which caused a bug check to trigger when the
476 VFS calls ->clear_inode.
478 2.1.19 - Many cleanups, improvements, and a minor bug fix.
480 - Update ->setattr (fs/ntfs/inode.c::ntfs_setattr()) to refuse to
481 change the uid, gid, and mode of an inode as we do not support NTFS
483 - Remove BKL use from ntfs_setattr() syncing up with the rest of the
485 - Get rid of the ugly transparent union in fs/ntfs/dir.c::ntfs_readdir()
486 and ntfs_filldir() as per suggestion from Al Viro.
487 - Change '\0' and L'\0' to simply 0 as per advice from Linus Torvalds.
488 - Update ->truncate (fs/ntfs/inode.c::ntfs_truncate()) to check if the
489 inode size has changed and to only output an error if so.
490 - Rename fs/ntfs/attrib.h::attribute_value_length() to ntfs_attr_size().
491 - Add le{16,32,64} as well as sle{16,32,64} data types to
493 - Change ntfschar to be le16 instead of u16 in fs/ntfs/types.h.
494 - Add le versions of VCN, LCN, and LSN called leVCN, leLCN, and leLSN,
495 respectively, to fs/ntfs/types.h.
496 - Update endianness conversion macros in fs/ntfs/endian.h to use the
497 new types as appropriate.
498 - Do proper type casting when using sle64_to_cpup() in fs/ntfs/dir.c
500 - Add leMFT_REF data type to fs/ntfs/layout.h.
501 - Update all NTFS header files with the new little endian data types.
502 Affected files are fs/ntfs/layout.h, logfile.h, and time.h.
503 - Do proper type casting when using ntfs_is_*_recordp() in
504 fs/ntfs/logfile.c, mft.c, and super.c.
505 - Fix all the sparse bitwise warnings. Had to change all the typedef
506 enums storing little endian values to simple enums plus a typedef for
507 the datatype to make sparse happy.
508 - Fix a bug found by the new sparse bitwise warnings where the default
509 upcase table was defined as a pointer to wchar_t rather than ntfschar
510 in fs/ntfs/ntfs.h and super.c.
511 - Change {const_,}cpu_to_le{16,32}(0) to just 0 as suggested by Al Viro.
513 2.1.18 - Fix scheduling latencies at mount time as well as an endianness bug.
515 - Remove vol->nr_mft_records as it was pretty meaningless and optimize
516 the calculation of total/free inodes as used by statfs().
517 - Fix scheduling latencies in ntfs_fill_super() by dropping the BKL
518 because the code itself is using the ntfs_lock semaphore which
519 provides safe locking. (Ingo Molnar)
520 - Fix a potential bug in fs/ntfs/mft.c::map_extent_mft_record() that
521 could occur in the future for when we start closing/freeing extent
522 inodes if we don't set base_ni->ext.extent_ntfs_inos to NULL after
524 - Rename {find,lookup}_attr() to ntfs_attr_{find,lookup}() as well as
525 find_external_attr() to ntfs_external_attr_find() to cleanup the
526 namespace a bit and to be more consistent with libntfs.
527 - Rename {{re,}init,get,put}_attr_search_ctx() to
528 ntfs_attr_{{re,}init,get,put}_search_ctx() as well as the type
529 attr_search_context to ntfs_attr_search_ctx.
530 - Force use of ntfs_attr_find() in ntfs_attr_lookup() when searching
531 for the attribute list attribute itself.
532 - Fix endianness bug in ntfs_external_attr_find().
533 - Change ntfs_{external_,}attr_find() to return 0 on success, -ENOENT
534 if the attribute is not found, and -EIO on real error. In the case
535 of -ENOENT, the search context is updated to describe the attribute
536 before which the attribute being searched for would need to be
537 inserted if such an action were to be desired and in the case of
538 ntfs_external_attr_find() the search context is also updated to
539 indicate the attribute list entry before which the attribute list
540 entry of the attribute being searched for would need to be inserted
541 if such an action were to be desired. Also make ntfs_find_attr()
542 static and remove its prototype from attrib.h as it is not used
543 anywhere other than attrib.c. Update ntfs_attr_lookup() and all
544 callers of ntfs_{external,}attr_{find,lookup}() for the new return
546 - Minor cleanup of fs/ntfs/inode.c::ntfs_init_locked_inode().
548 2.1.17 - Fix bugs in mount time error code paths and other updates.
550 - Implement bitmap modification code (fs/ntfs/bitmap.[hc]). This
551 includes functions to set/clear a single bit or a run of bits.
552 - Add fs/ntfs/attrib.[hc]::ntfs_find_vcn() which returns the locked
553 runlist element containing a particular vcn. It also takes care of
554 mapping any needed runlist fragments.
555 - Implement cluster (de-)allocation code (fs/ntfs/lcnalloc.[hc]).
556 - Load attribute definition table from $AttrDef at mount time.
557 - Fix bugs in mount time error code paths involving (de)allocation of
558 the default and volume upcase tables.
559 - Remove ntfs_nr_mounts as it is no longer used.
561 2.1.16 - Implement access time updates, file sync, async io, and read/writev.
563 - Add support for readv/writev and aio_read/aio_write (fs/ntfs/file.c).
564 This is done by setting the appropriate file operations pointers to
565 the generic helper functions provided by mm/filemap.c.
566 - Implement fsync, fdatasync, and msync both for files (fs/ntfs/file.c)
567 and directories (fs/ntfs/dir.c).
568 - Add support for {a,m,c}time updates to inode.c::ntfs_write_inode().
569 Note, except for the root directory and any other system files opened
570 by the user, the system files will not have their access times
571 updated as they are only accessed at the inode level an hence the
572 file level functions which cause the times to be updated are never
575 2.1.15 - Invalidate quotas when (re)mounting read-write.
577 - Add new element itype.index.collation_rule to the ntfs inode
578 structure and set it appropriately in ntfs_read_locked_inode().
579 - Implement a new inode type "index" to allow efficient access to the
580 indices found in various system files and adapt inode handling
581 accordingly (fs/ntfs/inode.[hc]). An index inode is essentially an
582 attribute inode (NInoAttr() is true) with an attribute type of
583 AT_INDEX_ALLOCATION. As such, it is no longer allowed to call
584 ntfs_attr_iget() with an attribute type of AT_INDEX_ALLOCATION as
585 there would be no way to distinguish between normal attribute inodes
586 and index inodes. The function to obtain an index inode is
587 ntfs_index_iget() and it uses the helper function
588 ntfs_read_locked_index_inode(). Note, we do not overload
589 ntfs_attr_iget() as indices consist of multiple attributes so using
590 ntfs_attr_iget() to obtain an index inode would be confusing.
591 - Ensure that there is no overflow when doing page->index <<
592 PAGE_CACHE_SHIFT by casting page->index to s64 in fs/ntfs/aops.c.
593 - Use atomic kmap instead of kmap() in fs/ntfs/aops.c::ntfs_read_page()
594 and ntfs_read_block().
595 - Use case sensitive attribute lookups instead of case insensitive ones.
596 - Lock all page cache pages belonging to mst protected attributes while
597 accessing them to ensure we never see corrupt data while the page is
599 - Add framework for generic ntfs collation (fs/ntfs/collation.[hc]).
600 We have ntfs_is_collation_rule_supported() to check if the collation
601 rule you want to use is supported and ntfs_collation() which actually
602 collates two data items. We currently only support COLLATION_BINARY
603 and COLLATION_NTOFS_ULONG but support for other collation rules will
604 be added as the need arises.
605 - Add a new type, ntfs_index_context, to allow retrieval of an index
606 entry using the corresponding index key. To get an index context,
607 use ntfs_index_ctx_get() and to release it, use ntfs_index_ctx_put().
608 This also adds a new slab cache for the index contexts. To lookup a
609 key in an index inode, use ntfs_index_lookup(). After modifying an
610 index entry, call ntfs_index_entry_flush_dcache_page() followed by
611 ntfs_index_entry_mark_dirty() to ensure the changes are written out
612 to disk. For details see fs/ntfs/index.[hc]. Note, at present, if
613 an index entry is in the index allocation attribute rather than the
614 index root attribute it will not be written out (you will get a
615 warning message about discarded changes instead).
616 - Load the quota file ($Quota) and check if quota tracking is enabled
617 and if so, mark the quotas out of date. This causes windows to
618 rescan the volume on boot and update all quota entries.
619 - Add a set_page_dirty address space operation for ntfs_m[fs]t_aops.
620 It is simply set to __set_page_dirty_nobuffers() to make sure that
621 running set_page_dirty() on a page containing mft/ntfs records will
622 not affect the dirty state of the page buffers.
623 - Add fs/ntfs/index.c::__ntfs_index_entry_mark_dirty() which sets all
624 buffers that are inside the ntfs record in the page dirty after which
625 it sets the page dirty. This allows ->writepage to only write the
626 dirty index records rather than having to write all the records in
627 the page. Modify fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to
628 use this rather than __set_page_dirty_nobuffers().
629 - Implement fs/ntfs/aops.c::ntfs_write_mst_block() which enables the
630 writing of page cache pages belonging to mst protected attributes
631 like the index allocation attribute in directory indices and other
632 indices like $Quota/$Q, etc. This means that the quota is now marked
633 out of date on all volumes rather than only on ones where the quota
634 defaults entry is in the index root attribute of the $Quota/$Q index.
636 2.1.14 - Fix an NFSd caused deadlock reported by several users.
638 - Modify fs/ntfs/ntfs_readdir() to copy the index root attribute value
639 to a buffer so that we can put the search context and unmap the mft
640 record before calling the filldir() callback. We need to do this
641 because of NFSd which calls ->lookup() from its filldir callback()
642 and this causes NTFS to deadlock as ntfs_lookup() maps the mft record
643 of the directory and since ntfs_readdir() has got it mapped already
644 ntfs_lookup() deadlocks.
646 2.1.13 - Enable overwriting of resident files and housekeeping of system files.
648 - Implement writing of mft records (fs/ntfs/mft.[hc]), which includes
649 keeping the mft mirror in sync with the mft when mirrored mft records
650 are written. The functions are write_mft_record{,_nolock}(). The
651 implementation is quite rudimentary for now with lots of things not
652 implemented yet but I am not sure any of them can actually occur so
653 I will wait for people to hit each one and only then implement it.
654 - Commit open system inodes at umount time. This should make it
655 virtually impossible for sync_mft_mirror_umount() to ever be needed.
656 - Implement ->write_inode (fs/ntfs/inode.c::ntfs_write_inode()) for the
657 ntfs super operations. This gives us inode writing via the VFS inode
658 dirty code paths. Note: Access time updates are not implemented yet.
659 - Implement fs/ntfs/mft.[hc]::{,__}mark_mft_record_dirty() and make
660 fs/ntfs/aops.c::ntfs_writepage() and ntfs_commit_write() use it, thus
661 finally enabling resident file overwrite! (-8 This also includes a
662 placeholder for ->writepage (ntfs_mft_writepage()), which for now
663 just redirties the page and returns. Also, at umount time, we for
664 now throw away all mft data page cache pages after the last call to
665 ntfs_commit_inode() in the hope that all inodes will have been
666 written out by then and hence no dirty (meta)data will be lost. We
667 also check for this case and emit an error message telling the user
669 - Use set_page_writeback() and end_page_writeback() in the resident
670 attribute code path of fs/ntfs/aops.c::ntfs_writepage() otherwise
671 the radix-tree tag PAGECACHE_TAG_DIRTY remains set even though the
673 - Implement ntfs_mft_writepage() so it now checks if any of the mft
674 records in the page are dirty and if so redirties the page and
675 returns. Otherwise it just returns (after doing set_page_writeback(),
676 unlock_page(), end_page_writeback() or the radix-tree tag
677 PAGECACHE_TAG_DIRTY remains set even though the page is clean), thus
678 alowing the VM to do with the page as it pleases. Also, at umount
679 time, now only throw away dirty mft (meta)data pages if dirty inodes
680 are present and ask the user to email us if they see this happening.
681 - Add functions ntfs_{clear,set}_volume_flags(), to modify the volume
682 information flags (fs/ntfs/super.c).
683 - Mark the volume dirty when (re)mounting read-write and mark it clean
684 when unmounting or remounting read-only. If any volume errors are
685 found, the volume is left marked dirty to force chkdsk to run.
686 - Add code to set the NT4 compatibility flag when (re)mounting
687 read-write for newer NTFS versions but leave it commented out for now
688 since we do not make any modifications that are NTFS 1.2 specific yet
689 and since setting this flag breaks Captive-NTFS which is not nice.
690 This code must be enabled once we start writing NTFS 1.2 specific
691 changes otherwise Windows NTFS driver might crash / cause corruption.
693 2.1.12 - Fix the second fix to the decompression engine and some cleanups.
695 - Add a new address space operations struct, ntfs_mst_aops, for mst
696 protected attributes. This is because the default ntfs_aops do not
697 make sense with mst protected data and were they to write anything to
698 such an attribute they would cause data corruption so we provide
699 ntfs_mst_aops which does not have any write related operations set.
700 - Cleanup dirty ntfs inode handling (fs/ntfs/inode.[hc]) which also
701 includes an adapted ntfs_commit_inode() and an implementation of
702 ntfs_write_inode() which for now just cleans dirty inodes without
703 writing them (it does emit a warning that this is happening).
704 - Undo the second decompression engine fix (see 2.1.9 release ChangeLog
705 entry) as it was only fixing a theoretical bug but at the same time
706 it badly broke the handling of sparse and uncompressed compression
709 2.1.11 - Driver internal cleanups.
711 - Only build logfile.o if building the driver with read-write support.
712 - Really final white space cleanups.
713 - Use generic_ffs() instead of ffs() in logfile.c which allows the
714 log_page_size variable to be optimized by gcc into a constant.
715 - Rename uchar_t to ntfschar everywhere as uchar_t is unsigned 1-byte
716 char as defined by POSIX and as found on some systems.
718 2.1.10 - Force read-only (re)mounting of volumes with unsupported volume flags.
720 - Finish off the white space cleanups (remove trailing spaces, etc).
721 - Clean up ntfs_fill_super() and ntfs_read_inode_mount() by removing
722 the kludges around the first iget(). Instead of (re)setting ->s_op
723 we have the $MFT inode set up by explicit new_inode() / set ->i_ino /
724 insert_inode_hash() / call ntfs_read_inode_mount() directly. This
725 kills the need for second super_operations and allows to return error
726 from ntfs_read_inode_mount() without resorting to ugly "poisoning"
728 - Force read-only (re)mounting if any of the following bits are set in
729 the volume information flags:
730 VOLUME_IS_DIRTY, VOLUME_RESIZE_LOG_FILE,
731 VOLUME_UPGRADE_ON_MOUNT, VOLUME_DELETE_USN_UNDERWAY,
732 VOLUME_REPAIR_OBJECT_ID, VOLUME_MODIFIED_BY_CHKDSK
733 To make this easier we define VOLUME_MUST_MOUNT_RO_MASK with all the
734 above bits set so the test is made easy.
736 2.1.9 - Fix two bugs in decompression engine.
738 - Fix a bug where we would not always detect that we have reached the
739 end of a compression block because we were ending at minus one byte
740 which is effectively the same as being at the end. The fix is to
741 check whether the uncompressed buffer has been fully filled and if so
742 we assume we have reached the end of the compression block. A big
743 thank you to Marcin Gibuła for the bug report, the assistance in
744 tracking down the bug and testing the fix.
745 - Fix a possible bug where when a compressed read is truncated to the
746 end of the file, the offset inside the last page was not truncated.
748 2.1.8 - Handle $MFT mirror and $LogFile, improve time handling, and cleanups.
750 - Use get_bh() instead of manual atomic_inc() in fs/ntfs/compress.c.
751 - Modify fs/ntfs/time.c::ntfs2utc(), get_current_ntfs_time(), and
752 utc2ntfs() to work with struct timespec instead of time_t on the
753 Linux UTC time side thus preserving the full precision of the NTFS
754 time and only loosing up to 99 nano-seconds in the Linux UTC time.
755 - Move fs/ntfs/time.c to fs/ntfs/time.h and make the time functions
757 - Remove unused ntfs_dirty_inode().
758 - Cleanup super operations declaration in fs/ntfs/super.c.
759 - Wrap flush_dcache_mft_record_page() in #ifdef NTFS_RW.
760 - Add NInoTestSetFoo() and NInoTestClearFoo() macro magic to
761 fs/ntfs/inode.h and use it to declare NInoTest{Set,Clear}Dirty.
762 - Move typedefs for ntfs_attr and test_t from fs/ntfs/inode.c to
763 fs/ntfs/inode.h so they can be used elsewhere.
764 - Determine the mft mirror size as the number of mirrored mft records
765 and store it in ntfs_volume->mftmirr_size (fs/ntfs/super.c).
766 - Load the mft mirror at mount time and compare the mft records stored
767 in it to the ones in the mft. Force a read-only mount if the two do
768 not match (fs/ntfs/super.c).
769 - Fix type casting related warnings on 64-bit architectures. Thanks
770 to Meelis Roos for reporting them.
771 - Move %L to %ll as %L is floating point and %ll is integer which is
773 - Read the journal ($LogFile) and determine if the volume has been
774 shutdown cleanly and force a read-only mount if not (fs/ntfs/super.c
775 and fs/ntfs/logfile.c). This is a little bit of a crude check in
776 that we only look at the restart areas and not at the actual log
777 records so that there will be a very small number of cases where we
778 think that a volume is dirty when in fact it is clean. This should
779 only affect volumes that have not been shutdown cleanly and did not
780 have any pending, non-check-pointed i/o.
781 - If the $LogFile indicates a clean shutdown and a read-write (re)mount
782 is requested, empty $LogFile by overwriting it with 0xff bytes to
783 ensure that Windows cannot cause data corruption by replaying a stale
784 journal after Linux has written to the volume.
786 2.1.7 - Enable NFS exporting of mounted NTFS volumes.
788 - Set i_generation in the VFS inode from the seq_no of the NTFS inode.
789 - Make ntfs_lookup() NFS export safe, i.e. use d_splice_alias(), etc.
790 - Implement ->get_dentry() in fs/ntfs/namei.c::ntfs_get_dentry() as the
791 default doesn't allow inode number 0 which is a valid inode on NTFS
792 and even if it did allow that it uses iget() instead of ntfs_iget()
793 which makes it useless for us.
794 - Implement ->get_parent() in fs/ntfs/namei.c::ntfs_get_parent() as the
795 default just returns -EACCES which is not very useful.
796 - Define export operations (->s_export_op) for NTFS (ntfs_export_ops)
797 and set them up in the super block at mount time (super.c) this
798 allows mounted NTFS volumes to be exported via NFS.
799 - Add missing return -EOPNOTSUPP; in
800 fs/ntfs/aops.c::ntfs_commit_nonresident_write().
801 - Enforce no atime and no dir atime updates at mount/remount time as
802 they are not implemented yet anyway.
803 - Move a few assignments in fs/ntfs/attrib.c::load_attribute_list() to
804 after a NULL check. Thanks to Dave Jones for pointing this out.
806 2.1.6 - Fix minor bug in handling of compressed directories.
808 - Fix bug in handling of compressed directories. A compressed
809 directory is not really compressed so when we set the ->i_blocks
810 field of a compressed directory inode we were setting it from the
811 non-existing field ni->itype.compressed.size which gave random
812 results... For directories we now always use ni->allocated_size.
814 2.1.5 - Fix minor bug in attribute list attribute handling.
816 - Fix bug in attribute list handling. Actually it is not as much a bug
817 as too much protection in that we were not allowing attribute lists
818 which waste space on disk while Windows XP clearly allows it and in
819 fact creates such attribute lists so our driver was failing.
820 - Update NTFS documentation ready for 2.6 kernel release.
822 2.1.4 - Reduce compiler requirements.
824 - Remove all uses of unnamed structs and unions in the driver to make
825 old and newer gcc versions happy. Makes it a bit uglier IMO but at
826 least people will stop hassling me about it.
828 2.1.3 - Important bug fixes in corner cases.
830 - super.c::parse_ntfs_boot_sector(): Correct the check for 64-bit
831 clusters. (Philipp Thomas)
832 - attrib.c::load_attribute_list(): Fix bug when initialized_size is a
833 multiple of the block_size but not the cluster size. (Szabolcs
834 Szakacsits <szaka@sienet.hu>)
836 2.1.2 - Important bug fixes aleviating the hangs in statfs.
838 - Fix buggy free cluster and free inode determination logic.
840 2.1.1 - Minor updates.
842 - Add handling for initialized_size != data_size in compressed files.
843 - Reduce function local stack usage from 0x3d4 bytes to just noise in
844 fs/ntfs/upcase.c. (Randy Dunlap <rddunlap@osdl.ord>)
845 - Remove compiler warnings for newer gcc.
846 - Pages are no longer kmapped by mm/filemap.c::generic_file_write()
847 around calls to ->{prepare,commit}_write. Adapt NTFS appropriately
848 in fs/ntfs/aops.c::ntfs_prepare_nonresident_write() by using
849 kmap_atomic(KM_USER0).
851 2.1.0 - First steps towards write support: implement file overwrite.
853 - Add configuration option for developmental write support with an
854 appropriately scary configuration help text.
855 - Initial implementation of fs/ntfs/aops.c::ntfs_writepage() and its
856 helper fs/ntfs/aops.c::ntfs_write_block(). This enables mmap(2) based
857 overwriting of existing files on ntfs. Note: Resident files are
858 only written into memory, and not written out to disk at present, so
859 avoid writing to files smaller than about 1kiB.
860 - Initial implementation of fs/ntfs/aops.c::ntfs_prepare_write(), its
861 helper fs/ntfs/aops.c::ntfs_prepare_nonresident_write() and their
862 counterparts, fs/ntfs/aops.c::ntfs_commit_write(), and
863 fs/ntfs/aops.c::ntfs_commit_nonresident_write(), respectively. Also,
864 add generic_file_write() to the ntfs file operations (fs/ntfs/file.c).
865 This enables write(2) based overwriting of existing files on ntfs.
866 Note: As with mmap(2) based overwriting, resident files are only
867 written into memory, and not written out to disk at present, so avoid
868 writing to files smaller than about 1kiB.
869 - Implement ->truncate (fs/ntfs/inode.c::ntfs_truncate()) and
870 ->setattr() (fs/ntfs/inode.c::ntfs_setattr()) inode operations for
871 files with the purpose of intercepting and aborting all i_size
872 changes which we do not support yet. ntfs_truncate() actually only
873 emits a warning message but AFAICS our interception of i_size changes
874 elsewhere means ntfs_truncate() never gets called for i_size changes.
875 It is only called from generic_file_write() when we fail in
876 ntfs_prepare_{,nonresident_}write() in order to discard any
877 instantiated buffers beyond i_size. Thus i_size is not actually
878 changed so our warning message is enough. Unfortunately it is not
879 possible to easily determine if i_size is being changed or not hence
880 we just emit an appropriately worded error message.
882 2.0.25 - Small bug fixes and cleanups.
884 - Unlock the page in an out of memory error code path in
885 fs/ntfs/aops.c::ntfs_read_block().
886 - If fs/ntfs/aops.c::ntfs_read_page() is called on an uptodate page,
887 just unlock the page and return. (This can happen due to ->writepage
888 clearing PageUptodate() during write out of MstProtected()
890 - Remove leaked write code again.
894 - Treat BUG_ON() as ASSERT() not VERIFY(), i.e. do not use side effects
895 inside BUG_ON(). (Adam J. Richter)
896 - Split logical OR expressions inside BUG_ON() into individual BUG_ON()
897 calls for improved debugging. (Adam J. Richter)
898 - Add errors flag to the ntfs volume state, accessed via
899 NVol{,Set,Clear}Errors(vol).
900 - Do not allow read-write remounts of read-only volumes with errors.
901 - Clarify comment for ntfs file operation sendfile which was added by
902 Christoph Hellwig a while ago (just using generic_file_sendfile())
903 to say that ntfs ->sendfile is only used for the case where the
904 source data is on the ntfs partition and the destination is
905 somewhere else, i.e. nothing we need to concern ourselves with.
906 - Add generic_file_write() as our ntfs file write operation.
908 2.0.23 - Major bug fixes (races, deadlocks, non-i386 architectures).
910 - Massive internal locking changes to mft record locking. Fixes lock
911 recursion and replaces the mrec_lock read/write semaphore with a
912 mutex. Also removes the now superfluous mft_count. This fixes several
913 race conditions and deadlocks, especially in the future write code.
914 - Fix ntfs over loopback for compressed files by adding an
915 optimization barrier. (gcc was screwing up otherwise ?)
916 - Miscellaneous cleanups all over the code and a fix or two in error
918 Thanks go to Christoph Hellwig for pointing out the following two:
919 - Remove now unused function fs/ntfs/malloc.h::vmalloc_nofs().
920 - Fix ntfs_free() for ia64 and parisc by checking for VMALLOC_END, too.
922 2.0.22 - Cleanups, mainly to ntfs_readdir(), and use C99 initializers.
924 - Change fs/ntfs/dir.c::ntfs_reddir() to only read/write ->f_pos once
925 at entry/exit respectively.
926 - Use C99 initializers for structures.
927 - Remove unused variable blocks from fs/ntfs/aops.c::ntfs_read_block().
929 2.0.21 - Check for, and refuse to work with too large files/directories/volumes.
931 - Limit volume size at mount time to 2TiB on architectures where
932 unsigned long is 32-bits (fs/ntfs/super.c::parse_ntfs_boot_sector()).
933 This is the most we can do without overflowing the 32-bit limit of
934 the block device size imposed on us by sb_bread() and sb_getblk()
936 - Limit file/directory size at open() time to 16TiB on architectures
937 where unsigned long is 32-bits (fs/ntfs/file.c::ntfs_file_open() and
938 fs/ntfs/dir.c::ntfs_dir_open()). This is the most we can do without
939 overflowing the page cache page index.
941 2.0.20 - Support non-resident directory index bitmaps, fix page leak in readdir.
943 - Move the directory index bitmap to use an attribute inode instead of
944 having special fields for it inside the ntfs inode structure. This
945 means that the index bitmaps now use the page cache for i/o, too,
946 and also as a side effect we get support for non-resident index
948 - Simplify/cleanup error handling in fs/ntfs/dir.c::ntfs_readdir() and
949 fix a page leak that manifested itself in some cases.
950 - Add fs/ntfs/inode.c::ntfs_put_inode(), which we need to release the
951 index bitmap inode on the final iput().
953 2.0.19 - Fix race condition, improvements, and optimizations in i/o interface.
955 - Apply block optimization added to fs/ntfs/aops.c::ntfs_read_block()
956 to fs/ntfs/compress.c::ntfs_file_read_compressed_block() as well.
957 - Drop the "file" from ntfs_file_read_compressed_block().
958 - Rename fs/ntfs/aops.c::ntfs_enb_buffer_read_async() to
959 ntfs_end_buffer_async_read() (more like the fs/buffer.c counterpart).
960 - Update ntfs_end_buffer_async_read() with the improved logic from
961 its updated counterpart fs/buffer.c::end_buffer_async_read(). Apply
962 further logic improvements to better determine when we set PageError.
963 - Update submission of buffers in fs/ntfs/aops.c::ntfs_read_block() to
964 check for the buffers being uptodate first in line with the updated
965 fs/buffer.c::block_read_full_page(). This plugs a small race
968 2.0.18 - Fix race condition in reading of compressed files.
970 - There was a narrow window between checking a buffer head for being
971 uptodate and locking it in ntfs_file_read_compressed_block(). We now
972 lock the buffer and then check whether it is uptodate or not.
974 2.0.17 - Cleanups and optimizations - shrinking the ToDo list.
976 - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to return an error
977 code and update callers, i.e. ntfs_iget(), to pass that error code
978 up instead of just using -EIO.
979 - Modifications to super.c to ensure that both mount and remount
980 cannot set any write related options when the driver is compiled
982 - Optimize block resolution in fs/ntfs/aops.c::ntfs_read_block() to
983 cache the current runlist element. This should improve performance
984 when reading very large and/or very fragmented data.
986 2.0.16 - Convert access to $MFT/$BITMAP to attribute inode API.
988 - Fix a stupid bug introduced in 2.0.15 where we were unmapping the
989 wrong inode in fs/ntfs/inode.c::ntfs_attr_iget().
990 - Fix debugging check in fs/ntfs/aops.c::ntfs_read_block().
991 - Convert $MFT/$BITMAP access to attribute inode API and remove all
992 remnants of the ugly mftbmp address space and operations hack. This
993 means we finally have only one readpage function as well as only one
994 async io completion handler. Yey! The mft bitmap is now just an
995 attribute inode and is accessed from vol->mftbmp_ino just as if it
996 were a normal file. Fake inodes rule. (-:
998 2.0.15 - Fake inodes based attribute i/o via the pagecache, fixes and cleanups.
1000 - Fix silly bug in fs/ntfs/super.c::parse_options() which was causing
1001 remounts to fail when the partition had an entry in /etc/fstab and
1002 the entry specified the nls= option.
1003 - Apply same macro magic used in fs/ntfs/inode.h to fs/ntfs/volume.h to
1004 expand all the helper functions NVolFoo(), NVolSetFoo(), and
1006 - Move copyright statement from driver initialisation message to
1007 module description (fs/super.c). This makes the initialisation
1008 message fit on one line and fits in better with rest of kernel.
1009 - Update fs/ntfs/attrib.c::map_run_list() to work on both real and
1010 attribute inodes, and both for files and directories.
1011 - Implement fake attribute inodes allowing all attribute i/o to go via
1012 the page cache and to use all the normal vfs/mm functionality:
1013 - Add ntfs_attr_iget() and its helper ntfs_read_locked_attr_inode()
1015 - Add needed cleanup code to ntfs_clear_big_inode().
1016 - Merge address space operations for files and directories (aops.c),
1017 now just have ntfs_aops:
1019 end_buffer_read_attr_async() -> ntfs_end_buffer_read_async(),
1020 ntfs_attr_read_block() -> ntfs_read_block(),
1021 ntfs_file_read_page() -> ntfs_readpage().
1022 - Rewrite fs/ntfs/aops.c::ntfs_readpage() to work on both real and
1023 attribute inodes, and both for files and directories.
1024 - Remove obsolete fs/ntfs/aops.c::ntfs_mst_readpage().
1026 2.0.14 - Run list merging code cleanup, minor locking changes, typo fixes.
1028 - Change fs/ntfs/super.c::ntfs_statfs() to not rely on BKL by moving
1029 the locking out of super.c::get_nr_free_mft_records() and taking and
1030 dropping the mftbmp_lock rw_semaphore in ntfs_statfs() itself.
1031 - Bring attribute runlist merging code (fs/ntfs/attrib.c) in sync with
1032 current userspace ntfs library code. This means that if a merge
1033 fails the original runlists are always left unmodified instead of
1034 being silently corrupted.
1037 2.0.13 - Use iget5_locked() in preparation for fake inodes and small cleanups.
1039 - Remove nr_mft_bits and the now superfluous union with nr_mft_records
1040 from ntfs_volume structure.
1041 - Remove nr_lcn_bits and the now superfluous union with nr_clusters
1042 from ntfs_volume structure.
1043 - Use iget5_locked() and friends instead of conventional iget(). Wrap
1044 the call in fs/ntfs/inode.c::ntfs_iget() and update callers of iget()
1045 to use ntfs_iget(). Leave only one iget() call at mount time so we
1046 don't need an ntfs_iget_mount().
1047 - Change fs/ntfs/inode.c::ntfs_new_extent_inode() to take mft_no as an
1048 additional argument.
1050 2.0.12 - Initial cleanup of address space operations following 2.0.11 changes.
1052 - Merge fs/ntfs/aops.c::end_buffer_read_mst_async() and
1053 fs/ntfs/aops.c::end_buffer_read_file_async() into one function
1054 fs/ntfs/aops.c::end_buffer_read_attr_async() using NInoMstProtected()
1055 to determine whether to apply mst fixups or not.
1056 - Above change allows merging fs/ntfs/aops.c::ntfs_file_read_block()
1057 and fs/ntfs/aops.c::ntfs_mst_readpage() into one function
1058 fs/ntfs/aops.c::ntfs_attr_read_block(). Also, create a tiny wrapper
1059 fs/ntfs/aops.c::ntfs_mst_readpage() to transform the parameters from
1060 the VFS readpage function prototype to the ntfs_attr_read_block()
1063 2.0.11 - Initial preparations for fake inode based attribute i/o.
1065 - Move definition of ntfs_inode_state_bits to fs/ntfs/inode.h and
1066 do some macro magic (adapted from include/linux/buffer_head.h) to
1067 expand all the helper functions NInoFoo(), NInoSetFoo(), and
1069 - Add new flag to ntfs_inode_state_bits: NI_Sparse.
1070 - Add new fields to ntfs_inode structure to allow use of fake inodes
1071 for attribute i/o: type, name, name_len. Also add new state bits:
1072 NI_Attr, which, if set, indicates the inode is a fake inode, and
1073 NI_MstProtected, which, if set, indicates the attribute uses multi
1074 sector transfer protection, i.e. fixups need to be applied after
1075 reads and before/after writes.
1076 - Rename fs/ntfs/inode.c::ntfs_{new,clear,destroy}_inode() to
1077 ntfs_{new,clear,destroy}_extent_inode() and update callers.
1078 - Use ntfs_clear_extent_inode() in fs/ntfs/inode.c::__ntfs_clear_inode()
1079 instead of ntfs_destroy_extent_inode().
1080 - Cleanup memory deallocations in {__,}ntfs_clear_{,big_}inode().
1081 - Make all operations on ntfs inode state bits use the NIno* functions.
1082 - Set up the new ntfs inode fields and state bits in
1083 fs/ntfs/inode.c::ntfs_read_inode() and add appropriate cleanup of
1084 allocated memory to __ntfs_clear_inode().
1085 - Cleanup ntfs_inode structure a bit for better ordering of elements
1086 w.r.t. their size to allow better packing of the structure in memory.
1088 2.0.10 - There can only be 2^32 - 1 inodes on an NTFS volume.
1090 - Add check at mount time to verify that the number of inodes on the
1091 volume does not exceed 2^32 - 1, which is the maximum allowed for
1092 NTFS according to Microsoft.
1093 - Change mft_no member of ntfs_inode structure to be unsigned long.
1094 Update all users. This makes ntfs_inode->mft_no just a copy of struct
1095 inode->i_ino. But we can't just always use struct inode->i_ino and
1096 remove mft_no because extent inodes do not have an attached struct
1099 2.0.9 - Decompression engine now uses a single buffer and other cleanups.
1101 - Change decompression engine to use a single buffer protected by a
1102 spin lock instead of per-CPU buffers. (Rusty Russell)
1103 - Do not update cb_pos when handling a partial final page during
1104 decompression of a sparse compression block, as the value is later
1105 reset without being read/used. (Rusty Russell)
1106 - Switch to using the new KM_BIO_SRC_IRQ for atomic kmap()s. (Andrew
1108 - Change buffer size in ntfs_readdir()/ntfs_filldir() to use
1109 NLS_MAX_CHARSET_SIZE which makes the buffers almost 1kiB each but
1110 it also makes everything safer so it is a good thing.
1111 - Miscellaneous minor cleanups to comments.
1113 2.0.8 - Major updates for handling of case sensitivity and dcache aliasing.
1115 Big thanks go to Al Viro and other inhabitants of #kernel for investing
1116 their time to discuss the case sensitivity and dcache aliasing issues.
1118 - Remove unused source file fs/ntfs/attraops.c.
1119 - Remove show_inodes mount option(s), thus dropping support for
1120 displaying of short file names.
1121 - Remove deprecated mount option posix.
1122 - Restore show_sys_files mount option.
1123 - Add new mount option case_sensitive, to determine if the driver
1124 treats file names as case sensitive or not. If case sensitive, create
1125 file names in the POSIX namespace. Otherwise create file names in the
1126 LONG/WIN32 namespace. Note, files remain accessible via their short
1127 file name, if it exists.
1128 - Remove really dumb logic bug in boot sector recovery code.
1129 - Fix dcache aliasing issues wrt short/long file names via changes
1130 to fs/ntfs/dir.c::ntfs_lookup_inode_by_name() and
1131 fs/ntfs/namei.c::ntfs_lookup():
1132 - Add additional argument to ntfs_lookup_inode_by_name() in which we
1133 return information about the matching file name if the case is not
1134 matching or the match is a short file name. See comments above the
1135 function definition for details.
1136 - Change ntfs_lookup() to only create dcache entries for the correctly
1137 cased file name and only for the WIN32 namespace counterpart of DOS
1138 namespace file names. This ensures we have only one dentry per
1139 directory and also removes all dcache aliasing issues between short
1140 and long file names once we add write support. See comments above
1141 function for details.
1142 - Fix potential 1 byte overflow in fs/ntfs/unistr.c::ntfs_ucstonls().
1144 2.0.7 - Minor cleanups and updates for changes in core kernel code.
1146 - Remove much of the NULL struct element initializers.
1147 - Various updates to make compatible with recent kernels.
1148 - Remove defines of MAX_BUF_PER_PAGE and include linux/buffer_head.h
1149 in fs/ntfs/ntfs.h instead.
1150 - Remove no longer needed KERNEL_VERSION checks. We are now in the
1151 kernel proper so they are no longer needed.
1153 2.0.6 - Major bugfix to make compatible with other kernel changes.
1155 - Initialize the mftbmp address space properly now that there are more
1156 fields in the struct address_space. This was leading to hangs and
1157 oopses on umount since 2.5.12 because of changes to other parts of
1158 the kernel. We probably want a kernel generic init_address_space()
1160 - Drop BKL from ntfs_readdir() after consultation with Al Viro. The
1161 only caller of ->readdir() is vfs_readdir() which holds i_sem during
1162 the call, and i_sem is sufficient protection against changes in the
1163 directory inode (including ->i_size).
1164 - Use generic_file_llseek() for directories (as opposed to
1165 default_llseek()) as this downs i_sem instead of the BKL which is
1166 what we now need for exclusion against ->f_pos changes considering we
1167 no longer take the BKL in ntfs_readdir().
1169 2.0.5 - Major bugfix. Buffer overflow in extent inode handling.
1171 - No need to set old blocksize in super.c::ntfs_fill_super() as the
1172 VFS does so via invocation of deactivate_super() calling
1173 fs->fill_super() calling block_kill_super() which does it.
1174 - BKL moved from VFS into dir.c::ntfs_readdir(). (Linus Torvalds)
1175 -> Do we really need it? I don't think so as we have exclusion on
1176 the directory ntfs_inode rw_semaphore mrec_lock. We mmight have to
1177 move the ->f_pos accesses under the mrec_lock though. Check this...
1178 - Fix really, really, really stupid buffer overflow in extent inode
1179 handling in mft.c::map_extent_mft_record().
1181 2.0.4 - Cleanups and updates for kernel 2.5.11.
1183 - Add documentation on how to use the MD driver to be able to use NTFS
1184 stripe and volume sets in Linux and generally cleanup documentation
1186 Remove all uses of kdev_t in favour of struct block_device *:
1187 - Change compress.c::ntfs_file_read_compressed_block() to use
1188 sb_getblk() instead of getblk().
1189 - Change super.c::ntfs_fill_super() to use bdev_hardsect_size() instead
1190 of get_hardsect_size().
1191 - No need to get old blocksize in super.c::ntfs_fill_super() as
1192 fs/super.c::get_sb_bdev() already does this.
1193 - Set bh->b_bdev instead of bh->b_dev throughout aops.c.
1195 2.0.3 - Small bug fixes, cleanups, and performance improvements.
1197 - Remove some dead code from mft.c.
1198 - Optimize readpage and read_block functions throughout aops.c so that
1199 only initialized blocks are read. Non-initialized ones have their
1200 buffer head mapped, zeroed, and set up to date, without scheduling
1201 any i/o. Thanks to Al Viro for advice on how to avoid the device i/o.
1202 Thanks go to Andrew Morton for spotting the below:
1203 - Fix buglet in allocate_compression_buffers() error code path.
1204 - Call flush_dcache_page() after modifying page cache page contents in
1205 ntfs_file_readpage().
1206 - Check for existence of page buffers throughout aops.c before calling
1207 create_empty_buffers(). This happens when an I/O error occurs and the
1208 read is retried. (It also happens once writing is implemented so that
1209 needed doing anyway but I had left it for later...)
1210 - Don't BUG_ON() uptodate and/or mapped buffers throughout aops.c in
1211 readpage and read_block functions. Reasoning same as above (i.e. I/O
1212 error retries and future write code paths.)
1214 2.0.2 - Minor updates and cleanups.
1216 - Cleanup: rename mst.c::__post_read_mst_fixup to post_write_mst_fixup
1217 and cleanup the code a bit, removing the unused size parameter.
1218 - Change default fmask to 0177 and update documentation.
1219 - Change attrib.c::get_attr_search_ctx() to return the search context
1220 directly instead of taking the address of a pointer. A return value
1221 of NULL means the allocation failed. Updated all callers
1223 - Update to 2.5.9 kernel (preserving backwards compatibility) by
1224 replacing all occurences of page->buffers with page_buffers(page).
1225 - Fix minor bugs in runlist merging, also minor cleanup.
1226 - Updates to bootsector layout and mft mirror contents descriptions.
1227 - Small bug fix in error detection in unistr.c and some cleanups.
1228 - Grow name buffer allocations in unistr.c in aligned mutlipled of 64
1231 2.0.1 - Minor updates.
1233 - Make default umask correspond to documentation.
1234 - Improve documentation.
1235 - Set default mode to include execute bit. The {u,f,d}mask can be used
1236 to take it away if desired. This allows binaries to be executed from
1237 a mounted ntfs partition.
1239 2.0.0 - New version number. Remove TNG from the name. Now in the kernel.
1241 - Add kill_super, just keeping up with the vfs changes in the kernel.
1242 - Repeat some changes from tng-0.0.8 that somehow got lost on the way
1243 from the CVS import into BitKeeper.
1244 - Begin to implement proper handling of allocated_size vs
1245 initialized_size vs data_size (i.e. i_size). Done are
1246 mft.c::ntfs_mft_readpage(), aops.c::end_buffer_read_index_async(),
1247 and attrib.c::load_attribute_list().
1248 - Lock the runlist in attrib.c::load_attribute_list() while using it.
1249 - Fix memory leak in ntfs_file_read_compressed_block() and generally
1250 clean up compress.c a little, removing some uncommented/unused debug
1252 - Tidy up dir.c a little bit.
1253 - Don't bother getting the runlist in inode.c::ntfs_read_inode().
1254 - Merge mft.c::ntfs_mft_readpage() and aops.c::ntfs_index_readpage()
1255 creating aops.c::ntfs_mst_readpage(), improving the handling of
1256 holes and overflow in the process and implementing the correct
1257 equivalent of ntfs_file_get_block() in ntfs_mst_readpage() itself.
1258 I am aiming for correctness at the moment. Modularisation can come
1260 - Rename aops.c::end_buffer_read_index_async() to
1261 end_buffer_read_mst_async() and optimize the overflow checking and
1263 - Use the host of the mftbmp address space mapping to hold the ntfs
1264 volume. This is needed so the async i/o completion handler can
1265 retrieve a pointer to the volume. Hopefully this will not cause
1266 problems elsewhere in the kernel... Otherwise will need to use a
1268 - Complete implementation of proper handling of allocated_size vs
1269 initialized_size vs data_size (i.e. i_size) in whole driver.
1270 Basically aops.c is now completely rewritten.
1271 - Change NTFS driver name to just NTFS and set version number to 2.0.0
1272 to make a clear distinction from the old driver which is still on
1275 tng-0.0.8 - 08/03/2002 - Now using BitKeeper, http://linux-ntfs.bkbits.net/
1277 - Replace bdevname(sb->s_dev) with sb->s_id.
1278 - Remove now superfluous new-line characters in all callers of
1280 - Apply kludge in ntfs_read_inode(), setting i_nlink to 1 for
1281 directories. Without this the "find" utility gets very upset which is
1282 fair enough as Linux/Unix do not support directory hard links.
1283 - Further runlist merging work. (Richard Russon)
1284 - Backwards compatibility for gcc-2.95. (Richard Russon)
1285 - Update to kernel 2.5.5-pre1 and rediff the now tiny patch.
1286 - Convert to new filesystem declaration using ->ntfs_get_sb() and
1287 replacing ntfs_read_super() with ntfs_fill_super().
1288 - Set s_maxbytes to MAX_LFS_FILESIZE to avoid page cache page index
1289 overflow on 32-bit architectures.
1290 - Cleanup upcase loading code to use ntfs_(un)map_page().
1291 - Disable/reenable preemtion in critical sections of compession engine.
1292 - Replace device size determination in ntfs_fill_super() with
1293 sb->s_bdev->bd_inode->i_size (in bytes) and remove now superfluous
1294 function super.c::get_nr_blocks().
1295 - Implement a mount time option (show_inodes) allowing choice of which
1296 types of inode names readdir() returns and modify ntfs_filldir()
1297 accordingly. There are several parameters to show_inodes:
1298 system: system files
1299 win32: long file names (including POSIX file names) [DEFAULT]
1301 dos: short file names only (excluding POSIX file names)
1303 posix: same as both win32 and dos
1305 Note that the options are additive, i.e. specifying:
1306 -o show_inodes=system,show_inodes=win32,show_inodes=dos
1307 is the same as specifying:
1309 Note that the "posix" and "all" options will show all directory
1310 names, BUT the link count on each directory inode entry is set to 1,
1311 due to Linux not supporting directory hard links. This may well
1312 confuse some userspace applications, since the directory names will
1313 have the same inode numbers. Thus it is NOT advisable to use the
1314 "posix" or "all" options. We provide them only for completeness sake.
1315 - Add copies of allocated_size, initialized_size, and compressed_size to
1316 the ntfs inode structure and set them up in
1317 inode.c::ntfs_read_inode(). These reflect the unnamed data attribute
1318 for files and the index allocation attribute for directories.
1319 - Add copies of allocated_size and initialized_size to ntfs inode for
1320 $BITMAP attribute of large directories and set them up in
1321 inode.c::ntfs_read_inode().
1322 - Add copies of allocated_size and initialized_size to ntfs volume for
1323 $BITMAP attribute of $MFT and set them up in
1324 super.c::load_system_files().
1325 - Parse deprecated ntfs driver options (iocharset, show_sys_files,
1326 posix, and utf8) and tell user what the new options to use are. Note
1327 we still do support them but they will be removed with kernel 2.7.x.
1328 - Change all occurences of integer long long printf formatting to hex
1329 as printk() will not support long long integer format if/when the
1330 div64 patch goes into the kernel.
1331 - Make slab caches have stable names and change the names to what they
1332 were intended to be. These changes are required/made possible by the
1333 new slab cache name handling which removes the length limitation by
1334 requiring the caller of kmem_cache_create() to supply a stable name
1335 which is then referenced but not copied.
1336 - Rename run_list structure to run_list_element and create a new
1337 run_list structure containing a pointer to a run_list_element
1338 structure and a read/write semaphore. Adapt all users of runlists
1339 to new scheme and take and release the lock as needed. This fixes a
1340 nasty race as the run_list changes even when inodes are locked for
1341 reading and even when the inode isn't locked at all, so we really
1342 needed the serialization. We use a semaphore rather than a spinlock
1343 as memory allocations can sleep and doing everything GFP_ATOMIC
1345 - Cleanup read_inode() removing all code checking for lowest_vcn != 0.
1346 This can never happen due to the nature of lookup_attr() and how we
1347 support attribute lists. If it did happen it would imply the inode
1349 - Check for lowest_vcn != 0 in ntfs_read_inode() and mark the inode as
1351 - Update to 2.5.6-pre2 changes in struct address_space.
1352 - Use parent_ino() when accessing d_parent inode number in dir.c.
1353 - Import Sourceforge CVS repository into BitKeeper repository:
1354 http://linux-ntfs.bkbits.net/ntfs-tng-2.5
1355 - Update fs/Makefile, fs/Config.help, fs/Config.in, and
1356 Documentation/filesystems/ntfs.txt for NTFS TNG.
1357 - Create kernel configuration option controlling whether debugging
1359 - Add the required export of end_buffer_io_sync() from the patches
1360 directory to the kernel code.
1361 - Update inode.c::ntfs_show_options() with show_inodes mount option.
1362 - Update errors mount option.
1364 tng-0.0.7 - 13/02/2002 - The driver is now feature complete for read-only!
1366 - Cleanup mft.c and it's debug/error output in particular. Fix a minor
1367 bug in mapping of extent inodes. Update all the comments to fit all
1368 the recent code changes.
1369 - Modify vcn_to_lcn() to cope with entirely unmapped runlists.
1370 - Cleanups in compress.c, mostly comments and folding help.
1371 - Implement attrib.c::map_run_list() as a generic helper.
1372 - Make compress.c::ntfs_file_read_compressed_block() use map_run_list()
1373 thus making code shorter and enabling attribute list support.
1374 - Cleanup incorrect use of [su]64 with %L printf format specifier in
1375 all source files. Type casts to [unsigned] long long added to correct
1376 the mismatches (important for architectures which have long long not
1378 - Merge async io completion handlers for directory indexes and $MFT
1379 data into one by setting the index_block_size{_bits} of the ntfs
1380 inode for $MFT to the mft_record_size{_bits} of the ntfs_volume.
1381 - Cleanup aops.c, update comments.
1382 - Make ntfs_file_get_block() use map_run_list() so all files now
1383 support attribute lists.
1384 - Make ntfs_dir_readpage() almost verbatim copy of
1385 block_read_full_page() by using ntfs_file_get_block() with only real
1386 difference being the use of our own async io completion handler
1387 rather than the default one, thus reducing the amount of code and
1388 automatically enabling attribute list support for directory indices.
1389 - Fix bug in load_attribute_list() - forgot to call brelse in error
1391 - Change parameters to find_attr() and lookup_attr(). We no longer
1392 pass in the upcase table and its length. These can be gotten from
1393 ctx->ntfs_ino->vol->upcase{_len}. Update all callers.
1394 - Cleanups in attrib.c.
1395 - Implement merging of runlists, attrib.c::merge_run_lists() and its
1396 helpers. (Richard Russon)
1397 - Attribute lists part 2, attribute extents and multi part runlists:
1398 enable proper support for LCN_RL_NOT_MAPPED and automatic mapping of
1399 further runlist parts via attrib.c::map_run_list().
1400 - Tiny endianness bug fix in decompress_mapping_pairs().
1402 tng-0.0.6 - Encrypted directories, bug fixes, cleanups, debugging enhancements.
1404 - Enable encrypted directories. (Their index root is marked encrypted
1405 to indicate that new files in that directory should be created
1407 - Fix bug in NInoBmpNonResident() macro. (Cut and paste error.)
1408 - Enable $Extend system directory. Most (if not all) extended system
1409 files do not have unnamed data attributes so ntfs_read_inode() had to
1410 special case them but that is ok, as the special casing recovery
1411 happens inside an error code path so there is zero slow down in the
1412 normal fast path. The special casing is done by introducing a new
1413 function inode.c::ntfs_is_extended_system_file() which checks if any
1414 of the hard links in the inode point to $Extend as being their parent
1415 directory and if they do we assume this is an extended system file.
1416 - Create a sysctl/proc interface to allow {dis,en}abling of debug output
1417 when compiled with -DDEBUG. Default is debug messages to be disabled.
1418 To enable them, one writes a non-zero value to /proc/sys/fs/ntfs-debug
1419 (if /proc is enabled) or uses sysctl(2) to effect the same (if sysctl
1420 interface is enabled). Inspired by old ntfs driver.
1421 - Add debug_msgs insmod/kernel boot parameter to set whether debug
1422 messages are {dis,en}abled. This is useful to enable debug messages
1423 during ntfs initialization and is the only way to activate debugging
1424 when the sysctl interface is not enabled.
1425 - Cleanup debug output in various places.
1426 - Remove all dollar signs ($) from the source (except comments) to
1427 enable compilation on architectures whose gcc compiler does not
1428 support dollar signs in the names of variables/constants. Attribute
1429 types now start with AT_ instead of $ and $I30 is now just I30.
1430 - Cleanup ntfs_lookup() and add consistency check of sequence numbers.
1431 - Load complete runlist for $MFT/$BITMAP during mount and cleanup
1432 access functions. This means we now cope with $MFT/$BITMAP being
1433 spread accross several mft records.
1434 - Disable modification of mft_zone_multiplier on remount. We can always
1435 reenable this later on if we really want to, but we will need to make
1436 sure we readjust the mft_zone size / layout accordingly.
1438 tng-0.0.5 - Modernize for 2.5.x and further in line-ing with Al Viro's comments.
1440 - Use sb_set_blocksize() instead of set_blocksize() and verify the
1442 - Use sb_bread() instead of bread() throughout.
1443 - Add index_vcn_size{_bits} to ntfs_inode structure to store the size
1444 of a directory index block vcn. Apply resulting simplifications in
1446 - Fix a small bug somewhere (but forgot what it was).
1447 - Change ntfs_{debug,error,warning} to enable gcc to do type checking
1448 on the printf-format parameter list and fix bugs reported by gcc
1449 as a result. (Richard Russon)
1450 - Move inode allocation strategy to Al's new stuff but maintain the
1451 divorce of ntfs_inode from struct inode. To achieve this we have two
1452 separate slab caches, one for big ntfs inodes containing a struct
1453 inode and pure ntfs inodes and at the same time fix some faulty
1454 error code paths in ntfs_read_inode().
1455 - Show mount options in proc (inode.c::ntfs_show_options()).
1457 tng-0.0.4 - Big changes, getting in line with Al Viro's comments.
1459 - Modified (un)map_mft_record functions to be common for read and write
1460 case. To specify which is which, added extra parameter at front of
1461 parameter list. Pass either READ or WRITE to this, each has the
1463 - General cleanups to allow for easier folding in vi.
1464 - attrib.c::decompress_mapping_pairs() now accepts the old runlist
1465 argument, and invokes attrib.c::merge_run_lists() to merge the old
1466 and the new runlists.
1467 - Removed attrib.c::find_first_attr().
1468 - Implemented loading of attribute list and complete runlist for $MFT.
1469 This means we now cope with $MFT being spread across several mft
1471 - Adapt to 2.5.2-pre9 and the changed create_empty_buffers() syntax.
1472 - Adapt major/minor/kdev_t/[bk]devname stuff to new 2.5.x kernels.
1473 - Make ntfs_volume be allocated via kmalloc() instead of using a slab
1474 cache. There are too little ntfs_volume structures at any one time
1475 to justify a private slab cache.
1476 - Fix bogus kmap() use in async io completion. Now use kmap_atomic().
1477 Use KM_BIO_IRQ on advice from IRC/kernel...
1478 - Use ntfs_map_page() in map_mft_record() and create ->readpage method
1479 for reading $MFT (ntfs_mft_readpage). In the process create dedicated
1480 address space operations (ntfs_mft_aops) for $MFT inode mapping. Also
1481 removed the now superfluous exports from the kernel core patch.
1482 - Fix a bug where kfree() was used insted of ntfs_free().
1483 - Change map_mft_record() to take ntfs_inode as argument instead of
1484 vfs inode. Dito for unmap_mft_record(). Adapt all callers.
1485 - Add pointer to ntfs_volume to ntfs_inode.
1486 - Add mft record number and sequence number to ntfs_inode. Stop using
1487 i_ino and i_generation for in-driver purposes.
1488 - Implement attrib.c::merge_run_lists(). (Richard Russon)
1489 - Remove use of proper inodes by extent inodes. Move i_ino and
1490 i_generation to ntfs_inode to do this. Apply simplifications that
1491 result and remove iget_no_wait(), etc.
1492 - Pass ntfs_inode everywhere in the driver (used to be struct inode).
1493 - Add reference counting in ntfs_inode for the ntfs inode itself and
1494 for the mapped mft record.
1495 - Extend mft record mapping so we can (un)map extent mft records (new
1496 functions (un)map_extent_mft_record), and so mappings are reference
1497 counted and don't have to happen twice if already mapped - just ref
1499 - Add -o iocharset as alias to -o nls for backwards compatibility.
1500 - The latest core patch is now tiny. In fact just a single additional
1501 export is necessary over the base kernel.
1503 tng-0.0.3 - Cleanups, enhancements, bug fixes.
1505 - Work on attrib.c::decompress_mapping_pairs() to detect base extents
1506 and setup the runlist appropriately using knowledge provided by the
1507 sizes in the base attribute record.
1508 - Balance the get_/put_attr_search_ctx() calls so we don't leak memory
1510 - Introduce ntfs_malloc_nofs() and ntfs_free() to allocate/free a single
1511 page or use vmalloc depending on the amount of memory requested.
1512 - Cleanup error output. The __FUNCTION__ "(): " is now added
1513 automatically. Introduced a new header file debug.h to support this
1514 and also moved ntfs_debug() function into it.
1515 - Make reading of compressed files more intelligent and especially get
1516 rid of the vmalloc_nofs() from readpage(). This now uses per CPU
1517 buffers (allocated at first mount with cluster size <= 4kiB and
1518 deallocated on last umount with cluster size <= 4kiB), and
1519 asynchronous io for the compressed data using a list of buffer heads.
1520 Er, we use synchronous io as async io only works on whole pages
1521 covered by buffers and not on individual buffer heads...
1522 - Bug fix for reading compressed files with sparse compression blocks.
1524 tng-0.0.2 - Now handles larger/fragmented/compressed volumes/files/dirs.
1526 - Fixed handling of directories when cluster size exceeds index block
1528 - Hide DOS only name space directory entries from readdir() but allow
1529 them in lookup(). This should fix the problem that Linux doesn't
1530 support directory hard links, while still allowing access to entries
1531 via their short file name. This also has the benefit of mimicking
1532 what Windows users are used to, so it is the ideal solution.
1533 - Implemented sync_page everywhere so no more hangs in D state when
1535 - Stop using bforget() in favour of brelse().
1536 - Stop locking buffers unnecessarily.
1537 - Implemented compressed files (inode->mapping contains uncompressed
1538 data, raw compressed data is currently bread() into a vmalloc()ed
1540 - Enable compressed directories. (Their index root is marked compressed
1541 to indicate that new files in that directory should be created
1543 - Use vsnprintf rather than vsprintf in the ntfs_error and ntfs_warning
1544 functions. (Thanks to Will Dyson for pointing this out.)
1545 - Moved the ntfs_inode and ntfs_volume (the former ntfs_inode_info and
1546 ntfs_sb_info) out of the common inode and super_block structures and
1547 started using the generic_ip and generic_sbp pointers instead. This
1548 makes ntfs entirely private with respect to the kernel tree.
1549 - Detect compiler version and abort with error message if gcc less than
1551 - Fix bug in name comparison function in unistr.c.
1552 - Implement attribute lists part 1, the infrastructure: search contexts
1553 and operations, find_external_attr(), lookup_attr()) and make the
1554 code use the infrastructure.
1555 - Fix stupid buffer overflow bug that became apparent on larger run
1556 list containing attributes.
1557 - Fix bugs in readdir() that became apparent on larger directories.
1559 The driver is now really useful and survives the test
1560 find . -type f -exec md5sum "{}" \;
1561 without any error messages on a over 1GiB sized partition with >16k
1562 files on it, including compressed files and directories and many files
1563 and directories with attribute lists.
1565 tng-0.0.1 - The first useful version.
1567 - Added ntfs_lookup().
1568 - Added default upcase generation and handling.
1569 - Added compile options to be shown on module init.
1570 - Many bug fixes that were "hidden" before.
1571 - Update to latest kernel.
1572 - Added ntfs_readdir().
1573 - Added file operations for mmap(), read(), open() and llseek(). We just
1574 use the generic ones. The whole point of going through implementing
1575 readpage() methods and where possible get_block() call backs is that
1576 this allows us to make use of the generic high level methods provided
1579 The driver is now actually useful! Yey. (-: It undoubtedly has got bugs
1580 though and it doesn't implement accesssing compressed files yet. Also,
1581 accessing files with attribute list attributes is not implemented yet
1582 either. But for small or simple filesystems it should work and allow
1583 you to list directories, use stat on directory entries and the file
1584 system, open, read, mmap and llseek around in files. A big mile stone
1587 tng-0.0.0 - Initial version tag.
1589 Initial driver implementation. The driver can mount and umount simple
1590 NTFS filesystems (i.e. ones without attribute lists in the system
1591 files). If the mount fails there might be problems in the error handling
1592 code paths, so be warned. Otherwise it seems to be loading the system
1593 files nicely and the mft record read mapping/unmapping seems to be
1594 working nicely, too. Proof of inode metadata in the page cache and non-
1595 resident file unnamed stream data in the page cache concepts is thus
projects / linux-2.6 / blob