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.
25 2.1.24 - Lots of bug fixes and support more clean journal states.
27 - Support journals ($LogFile) which have been modified by chkdsk. This
28 means users can boot into Windows after we marked the volume dirty.
29 The Windows boot will run chkdsk and then reboot. The user can then
30 immediately boot into Linux rather than having to do a full Windows
31 boot first before rebooting into Linux and we will recognize such a
32 journal and empty it as it is clean by definition.
33 - Support journals ($LogFile) with only one restart page as well as
34 journals with two different restart pages. We sanity check both and
35 either use the only sane one or the more recent one of the two in the
36 case that both are valid.
37 - Modify fs/ntfs/malloc.h::ntfs_malloc_nofs() to do the kmalloc() based
38 allocations with __GFP_HIGHMEM, analogous to how the vmalloc() based
40 - Add fs/ntfs/malloc.h::ntfs_malloc_nofs_nofail() which is analogous to
41 ntfs_malloc_nofs() but it performs allocations with __GFP_NOFAIL and
43 - Use ntfs_malloc_nofs_nofail() in the two critical regions in
44 fs/ntfs/runlist.c::ntfs_runlists_merge(). This means we no longer
45 need to panic() if the allocation fails as it now cannot fail.
46 - Fix two nasty runlist merging bugs that had gone unnoticed so far.
47 Thanks to Stefano Picerno for the bug report.
48 - Remove two bogus BUG_ON()s from fs/ntfs/mft.c.
49 - Fix handling of valid but empty mapping pairs array in
50 fs/ntfs/runlist.c::ntfs_mapping_pairs_decompress().
51 - Report unrepresentable inodes during ntfs_readdir() as KERN_WARNING
52 messages and include the inode number. Thanks to Yura Pakhuchiy for
54 - Change ntfs_rl_truncate_nolock() to throw away the runlist if the new
56 - Add runlist.[hc]::ntfs_rl_punch_nolock() which punches a caller
57 specified hole into a runlist.
58 - Fix a bug in fs/ntfs/index.c::ntfs_index_lookup(). When the returned
59 index entry is in the index root, we forgot to set the @ir pointer in
60 the index context. Thanks to Yura Pakhuchiy for finding this bug.
61 - Remove bogus setting of PageError in ntfs_read_compressed_block().
62 - Add fs/ntfs/attrib.[hc]::ntfs_resident_attr_value_resize().
63 - Fix a bug in ntfs_map_runlist_nolock() where we forgot to protect
64 access to the allocated size in the ntfs inode with the size lock.
65 - Fix ntfs_attr_vcn_to_lcn_nolock() and ntfs_attr_find_vcn_nolock() to
66 return LCN_ENOENT when there is no runlist and the allocated size is
68 - Fix load_attribute_list() to handle the case of a NULL runlist.
69 - Fix handling of sparse attributes in ntfs_attr_make_non_resident().
70 - Add BUG() checks to ntfs_attr_make_non_resident() and ntfs_attr_set()
71 to ensure that these functions are never called for compressed or
73 - Fix cluster (de)allocators to work when the runlist is NULL and more
74 importantly to take a locked runlist rather than them locking it
75 which leads to lock reversal.
76 - Truncate {a,c,m}time to the ntfs supported time granularity when
77 updating the times in the inode in ntfs_setattr().
78 - Fixup handling of sparse, compressed, and encrypted attributes in
79 fs/ntfs/inode.c::ntfs_read_locked_{,attr_,index_}inode(),
80 fs/ntfs/aops.c::ntfs_{read,write}page().
81 - Make ntfs_write_block() not instantiate sparse blocks if they contain
83 - Optimize fs/ntfs/aops.c::ntfs_write_block() by extending the page
84 lock protection over the buffer submission for i/o which allows the
85 removal of the get_bh()/put_bh() pairs for each buffer.
86 - Fix fs/ntfs/aops.c::ntfs_{read,write}_block() to handle the case
87 where a concurrent truncate has truncated the runlist under our feet.
88 - Fix page_has_buffers()/page_buffers() handling in fs/ntfs/aops.c.
89 - In fs/ntfs/aops.c::ntfs_end_buffer_async_read(), use a bit spin lock
90 in the first buffer head instead of a driver global spin lock to
92 - Minor fix to error handling and error message display in
93 fs/ntfs/aops.c::ntfs_prepare_nonresident_write().
94 - Change the mount options {u,f,d}mask to always parse the number as
95 an octal number to conform to how chmod(1) works, too. Thanks to
96 Giuseppe Bilotta and Horst von Brand for pointing out the errors of
99 2.1.23 - Implement extension of resident files and make writing safe as well as
100 many bug fixes, cleanups, and enhancements...
102 - Add printk rate limiting for ntfs_warning() and ntfs_error() when
103 compiled without debug. This avoids a possible denial of service
104 attack. Thanks to Carl-Daniel Hailfinger from SuSE for pointing this
106 - Fix compilation warnings on ia64. (Randy Dunlap)
107 - Use i_size_{read,write}() instead of reading i_size by hand and cache
108 the value where apropriate.
109 - Add size_lock to the ntfs_inode structure. This is an rw spinlock
110 and it locks against access to the inode sizes. Note, ->size_lock
111 is also accessed from irq context so you must use the _irqsave and
112 _irqrestore lock and unlock functions, respectively. Protect all
113 accesses to allocated_size, initialized_size, and compressed_size.
114 - Minor optimization to fs/ntfs/super.c::ntfs_statfs() and its helpers.
115 - Implement extension of resident files in the regular file write code
116 paths (fs/ntfs/aops.c::ntfs_{prepare,commit}_write()). At present
117 this only works until the data attribute becomes too big for the mft
118 record after which we abort the write returning -EOPNOTSUPP from
119 ntfs_prepare_write().
120 - Add disable_sparse mount option together with a per volume sparse
121 enable bit which is set appropriately and a per inode sparse disable
122 bit which is preset on some system file inodes as appropriate.
123 - Enforce that sparse support is disabled on NTFS volumes pre 3.0.
124 - Fix a bug in fs/ntfs/runlist.c::ntfs_mapping_pairs_decompress() in
125 the creation of the unmapped runlist element for the base attribute
127 - Split ntfs_map_runlist() into ntfs_map_runlist() and a non-locking
128 helper ntfs_map_runlist_nolock() which is used by ntfs_map_runlist().
129 This allows us to map runlist fragments with the runlist lock already
130 held without having to drop and reacquire it around the call. Adapt
132 - Change ntfs_find_vcn() to ntfs_find_vcn_nolock() which takes a locked
133 runlist. This allows us to find runlist elements with the runlist
134 lock already held without having to drop and reacquire it around the
135 call. Adapt all callers.
136 - Change time to u64 in time.h::ntfs2utc() as it otherwise generates a
137 warning in the do_div() call on sparc32. Thanks to Meelis Roos for
138 the report and analysis of the warning.
139 - Fix a nasty runlist merge bug when merging two holes.
140 - Set the ntfs_inode->allocated_size to the real allocated size in the
141 mft record for resident attributes (fs/ntfs/inode.c).
142 - Small readability cleanup to use "a" instead of "ctx->attr"
143 everywhere (fs/ntfs/inode.c).
144 - Make fs/ntfs/namei.c::ntfs_get_{parent,dentry} static and move the
145 definition of ntfs_export_ops from fs/ntfs/super.c to namei.c. Also,
146 declare ntfs_export_ops in fs/ntfs/ntfs.h.
147 - Correct sparse file handling. The compressed values need to be
148 checked and set in the ntfs inode as done for compressed files and
149 the compressed size needs to be used for vfs inode->i_blocks instead
150 of the allocated size, again, as done for compressed files.
151 - Add AT_EA in addition to AT_DATA to whitelist for being allowed to be
152 non-resident in fs/ntfs/attrib.c::ntfs_attr_can_be_non_resident().
153 - Add fs/ntfs/attrib.c::ntfs_attr_vcn_to_lcn_nolock() used by the new
155 - Fix bug in fs/ntfs/attrib.c::ntfs_find_vcn_nolock() where after
156 dropping the read lock and taking the write lock we were not checking
157 whether someone else did not already do the work we wanted to do.
158 - Rename fs/ntfs/attrib.c::ntfs_find_vcn_nolock() to
159 ntfs_attr_find_vcn_nolock() and update all callers.
160 - Add fs/ntfs/attrib.[hc]::ntfs_attr_make_non_resident().
161 - Fix sign of various error return values to be negative in
163 - Modify ->readpage and ->writepage (fs/ntfs/aops.c) so they detect and
164 handle the case where an attribute is converted from resident to
165 non-resident by a concurrent file write.
166 - Remove checks for NULL before calling kfree() since kfree() does the
167 checking itself. (Jesper Juhl)
168 - Some utilities modify the boot sector but do not update the checksum.
169 Thus, relax the checking in fs/ntfs/super.c::is_boot_sector_ntfs() to
170 only emit a warning when the checksum is incorrect rather than
171 refusing the mount. Thanks to Bernd Casimir for pointing this
173 - Update attribute definition handling.
174 - Add NTFS_MAX_CLUSTER_SIZE and NTFS_MAX_PAGES_PER_CLUSTER constants.
175 - Use NTFS_MAX_CLUSTER_SIZE in super.c instead of hard coding 0x10000.
176 - Use MAX_BUF_PER_PAGE instead of variable sized array allocation for
177 better code generation and one less sparse warning in fs/ntfs/aops.c.
178 - Remove spurious void pointer casts from fs/ntfs/. (Pekka Enberg)
179 - Use C99 style structure initialization after memory allocation where
180 possible (fs/ntfs/{attrib.c,index.c,super.c}). Thanks to Al Viro and
182 - Stamp the transaction log ($UsnJrnl), aka user space journal, if it
183 is active on the volume and we are mounting read-write or remounting
184 from read-only to read-write.
185 - Fix a bug in address space operations error recovery code paths where
186 if the runlist was not mapped at all and a mapping error occured we
187 would leave the runlist locked on exit to the function so that the
188 next access to the same file would try to take the lock and deadlock.
189 - Detect the case when Windows has been suspended to disk on the volume
190 to be mounted and if this is the case do not allow (re)mounting
191 read-write. This is done by parsing hiberfil.sys if present.
192 - Fix several occurences of a bug where we would perform 'var & ~const'
193 with a 64-bit variable and a int, i.e. 32-bit, constant. This causes
194 the higher order 32-bits of the 64-bit variable to be zeroed. To fix
195 this cast the 'const' to the same 64-bit type as 'var'.
196 - Change the runlist terminator of the newly allocated cluster(s) to
197 LCN_ENOENT in ntfs_attr_make_non_resident(). Otherwise the runlist
199 - Add an extra parameter @last_vcn to ntfs_get_size_for_mapping_pairs()
200 and ntfs_mapping_pairs_build() to allow the runlist encoding to be
201 partial which is desirable when filling holes in sparse attributes.
203 - Change ntfs_map_runlist_nolock() to only decompress the mapping pairs
204 if the requested vcn is inside it. Otherwise we get into problems
205 when we try to map an out of bounds vcn because we then try to map
206 the already mapped runlist fragment which causes
207 ntfs_mapping_pairs_decompress() to fail and return error. Update
208 ntfs_attr_find_vcn_nolock() accordingly.
209 - Fix a nasty deadlock that appeared in recent kernels.
210 The situation: VFS inode X on a mounted ntfs volume is dirty. For
211 same inode X, the ntfs_inode is dirty and thus corresponding on-disk
212 inode, i.e. mft record, which is in a dirty PAGE_CACHE_PAGE belonging
213 to the table of inodes, i.e. $MFT, inode 0.
215 Process 1: sys_sync()/umount()/whatever... calls
216 __sync_single_inode() for $MFT -> do_writepages() -> write_page for
217 the dirty page containing the on-disk inode X, the page is now locked
218 -> ntfs_write_mst_block() which clears PageUptodate() on the page to
219 prevent anyone else getting hold of it whilst it does the write out.
220 This is necessary as the on-disk inode needs "fixups" applied before
221 the write to disk which are removed again after the write and
222 PageUptodate is then set again. It then analyses the page looking
223 for dirty on-disk inodes and when it finds one it calls
224 ntfs_may_write_mft_record() to see if it is safe to write this
225 on-disk inode. This then calls ilookup5() to check if the
226 corresponding VFS inode is in icache(). This in turn calls ifind()
227 which waits on the inode lock via wait_on_inode whilst holding the
229 Process 2: pdflush results in a call to __sync_single_inode for the
230 same VFS inode X on the ntfs volume. This locks the inode (I_LOCK)
231 then calls write-inode -> ntfs_write_inode -> map_mft_record() ->
232 read_cache_page() for the page (in page cache of table of inodes
233 $MFT, inode 0) containing the on-disk inode. This page has
234 PageUptodate() clear because of Process 1 (see above) so
235 read_cache_page() blocks when it tries to take the page lock for the
236 page so it can call ntfs_read_page().
237 Thus Process 1 is holding the page lock on the page containing the
238 on-disk inode X and it is waiting on the inode X to be unlocked in
239 ifind() so it can write the page out and then unlock the page.
240 And Process 2 is holding the inode lock on inode X and is waiting for
241 the page to be unlocked so it can call ntfs_readpage() or discover
242 that Process 1 set PageUptodate() again and use the page.
243 Thus we have a deadlock due to ifind() waiting on the inode lock.
244 The solution: The fix is to use the newly introduced
245 ilookup5_nowait() which does not wait on the inode's lock and hence
246 avoids the deadlock. This is safe as we do not care about the VFS
247 inode and only use the fact that it is in the VFS inode cache and the
248 fact that the vfs and ntfs inodes are one struct in memory to find
249 the ntfs inode in memory if present. Also, the ntfs inode has its
250 own locking so it does not matter if the vfs inode is locked.
251 - Fix bug in mft record writing where we forgot to set the device in
252 the buffers when mapping them after the VM had discarded them.
253 Thanks to Martin MOKREJŠ for the bug report.
255 2.1.22 - Many bug and race fixes and error handling improvements.
257 - Improve error handling in fs/ntfs/inode.c::ntfs_truncate().
258 - Change fs/ntfs/inode.c::ntfs_truncate() to return an error code
259 instead of void and provide a helper ntfs_truncate_vfs() for the
260 vfs ->truncate method.
261 - Add a new ntfs inode flag NInoTruncateFailed() and modify
262 fs/ntfs/inode.c::ntfs_truncate() to set and clear it appropriately.
263 - Fix min_size and max_size definitions in ATTR_DEF structure in
264 fs/ntfs/layout.h to be signed.
265 - Add attribute definition handling helpers to fs/ntfs/attrib.[hc]:
266 ntfs_attr_size_bounds_check(), ntfs_attr_can_be_non_resident(), and
267 ntfs_attr_can_be_resident(), which in turn use the new private helper
268 ntfs_attr_find_in_attrdef().
269 - In fs/ntfs/aops.c::mark_ntfs_record_dirty(), take the
270 mapping->private_lock around the dirtying of the buffer heads
271 analagous to the way it is done in __set_page_dirty_buffers().
272 - Ensure the mft record size does not exceed the PAGE_CACHE_SIZE at
273 mount time as this cannot work with the current implementation.
274 - Check for location of attribute name and improve error handling in
275 general in fs/ntfs/inode.c::ntfs_read_locked_inode() and friends.
276 - In fs/ntfs/aops.c::ntfs_writepage(), if the page is fully outside
277 i_size, i.e. race with truncate, invalidate the buffers on the page
278 so that they become freeable and hence the page does not leak.
279 - Remove unused function fs/ntfs/runlist.c::ntfs_rl_merge(). (Adrian
281 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_find() that resulted in
282 a NULL pointer dereference in the error code path when a corrupt
283 attribute was found. (Thanks to Domen Puncer for the bug report.)
284 - Add MODULE_VERSION() to fs/ntfs/super.c.
285 - Make several functions and variables static. (Adrian Bunk)
286 - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() so it allocates
287 buffers for the page if they are not present and then marks the
288 buffers belonging to the ntfs record dirty. This causes the buffers
289 to become busy and hence they are safe from removal until the page
290 has been written out.
291 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find() in the
292 error handling code path that resulted in a BUG() due to trying to
293 unmap an extent mft record when the mapping of it had failed and it
294 thus was not mapped. (Thanks to Ken MacFerrin for the bug report.)
295 - Drop the runlist lock after the vcn has been read in
296 fs/ntfs/lcnalloc.c::__ntfs_cluster_free().
297 - Rewrite handling of multi sector transfer errors. We now do not set
298 PageError() when such errors are detected in the async i/o handler
299 fs/ntfs/aops.c::ntfs_end_buffer_async_read(). All users of mst
300 protected attributes now check the magic of each ntfs record as they
301 use it and act appropriately. This has the effect of making errors
302 granular per ntfs record rather than per page which solves the case
303 where we cannot access any of the ntfs records in a page when a
304 single one of them had an mst error. (Thanks to Ken MacFerrin for
306 - Fix error handling in fs/ntfs/quota.c::ntfs_mark_quotas_out_of_date()
307 where we failed to release i_sem on the $Quota/$Q attribute inode.
308 - Fix bug in handling of bad inodes in fs/ntfs/namei.c::ntfs_lookup().
309 - Add mapping of unmapped buffers to all remaining code paths, i.e.
310 fs/ntfs/aops.c::ntfs_write_mst_block(), mft.c::ntfs_sync_mft_mirror(),
311 and write_mft_record_nolock(). From now on we require that the
312 complete runlist for the mft mirror is always mapped into memory.
313 - Add creation of buffers to fs/ntfs/mft.c::ntfs_sync_mft_mirror().
314 - Improve error handling in fs/ntfs/aops.c::ntfs_{read,write}_block().
315 - Cleanup fs/ntfs/aops.c::ntfs_{read,write}page() since we know that a
316 resident attribute will be smaller than a page which makes the code
317 simpler. Also make the code more tolerant to concurrent ->truncate.
319 2.1.21 - Fix some races and bugs, rewrite mft write code, add mft allocator.
321 - Implement extent mft record deallocation
322 fs/ntfs/mft.c::ntfs_extent_mft_record_free().
323 - Splitt runlist related functions off from attrib.[hc] to runlist.[hc].
324 - Add vol->mft_data_pos and initialize it at mount time.
325 - Rename init_runlist() to ntfs_init_runlist(), ntfs_vcn_to_lcn() to
326 ntfs_rl_vcn_to_lcn(), decompress_mapping_pairs() to
327 ntfs_mapping_pairs_decompress(), ntfs_merge_runlists() to
328 ntfs_runlists_merge() and adapt all callers.
329 - Add fs/ntfs/runlist.[hc]::ntfs_get_nr_significant_bytes(),
330 ntfs_get_size_for_mapping_pairs(), ntfs_write_significant_bytes(),
331 and ntfs_mapping_pairs_build(), adapted from libntfs.
332 - Make fs/ntfs/lcnalloc.c::ntfs_cluster_free_from_rl_nolock() not
333 static and add a declaration for it to lcnalloc.h.
334 - Add fs/ntfs/lcnalloc.h::ntfs_cluster_free_from_rl() which is a static
335 inline wrapper for ntfs_cluster_free_from_rl_nolock() which takes the
336 cluster bitmap lock for the duration of the call.
337 - Add fs/ntfs/attrib.[hc]::ntfs_attr_record_resize().
338 - Implement the equivalent of memset() for an ntfs attribute in
339 fs/ntfs/attrib.[hc]::ntfs_attr_set() and switch
340 fs/ntfs/logfile.c::ntfs_empty_logfile() to using it.
341 - Remove unnecessary casts from LCN_* constants.
342 - Implement fs/ntfs/runlist.c::ntfs_rl_truncate_nolock().
343 - Add MFT_RECORD_OLD as a copy of MFT_RECORD in fs/ntfs/layout.h and
344 change MFT_RECORD to contain the NTFS 3.1+ specific fields.
345 - Add a helper function fs/ntfs/aops.c::mark_ntfs_record_dirty() which
346 marks all buffers belonging to an ntfs record dirty, followed by
347 marking the page the ntfs record is in dirty and also marking the vfs
348 inode containing the ntfs record dirty (I_DIRTY_PAGES).
349 - Switch fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to using the
350 new helper fs/ntfs/aops.c::mark_ntfs_record_dirty() and remove the no
351 longer needed fs/ntfs/index.[hc]::__ntfs_index_entry_mark_dirty().
352 - Move ntfs_{un,}map_page() from ntfs.h to aops.h and fix resulting
354 - Move the typedefs for runlist_element and runlist from types.h to
355 runlist.h and fix resulting include errors.
356 - Remove unused {__,}format_mft_record() from fs/ntfs/mft.c.
357 - Modify fs/ntfs/mft.c::__mark_mft_record_dirty() to use the helper
358 mark_ntfs_record_dirty() which also changes the behaviour in that we
359 now set the buffers belonging to the mft record dirty as well as the
361 - Update fs/ntfs/mft.c::write_mft_record_nolock() and sync_mft_mirror()
362 to cope with the fact that there now are dirty buffers in mft pages.
363 - Update fs/ntfs/inode.c::ntfs_write_inode() to also use the helper
364 mark_ntfs_record_dirty() and thus to set the buffers belonging to the
365 mft record dirty as well as the page itself.
366 - Fix compiler warnings on x86-64 in fs/ntfs/dir.c. (Randy Dunlap,
367 slightly modified by me)
368 - Add fs/ntfs/mft.c::try_map_mft_record() which fails with -EALREADY if
369 the mft record is already locked and otherwise behaves the same way
370 as fs/ntfs/mft.c::map_mft_record().
371 - Modify fs/ntfs/mft.c::write_mft_record_nolock() so that it only
372 writes the mft record if the buffers belonging to it are dirty.
373 Otherwise we assume that it was written out by other means already.
374 - Attempting to write outside initialized size is _not_ a bug so remove
375 the bug check from fs/ntfs/aops.c::ntfs_write_mst_block(). It is in
376 fact required to write outside initialized size when preparing to
377 extend the initialized size.
378 - Map the page instead of using page_address() before writing to it in
379 fs/ntfs/aops.c::ntfs_mft_writepage().
380 - Provide exclusion between opening an inode / mapping an mft record
381 and accessing the mft record in fs/ntfs/mft.c::ntfs_mft_writepage()
382 by setting the page not uptodate throughout ntfs_mft_writepage().
383 - Clear the page uptodate flag in fs/ntfs/aops.c::ntfs_write_mst_block()
384 to ensure noone can see the page whilst the mst fixups are applied.
385 - Add the helper fs/ntfs/mft.c::ntfs_may_write_mft_record() which
386 checks if an mft record may be written out safely obtaining any
387 necessary locks in the process. This is used by
388 fs/ntfs/aops.c::ntfs_write_mst_block().
389 - Modify fs/ntfs/aops.c::ntfs_write_mst_block() to also work for
390 writing mft records and improve its error handling in the process.
391 Now if any of the records in the page fail to be written out, all
392 other records will be written out instead of aborting completely.
393 - Remove ntfs_mft_aops and update all users to use ntfs_mst_aops.
394 - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to set the
395 ntfs_mst_aops for all inodes which are NInoMstProtected() and
396 ntfs_aops for all other inodes.
397 - Rename fs/ntfs/mft.c::sync_mft_mirror{,_umount}() to
398 ntfs_sync_mft_mirror{,_umount}() and change their parameters so they
399 no longer require an ntfs inode to be present. Update all callers.
400 - Cleanup the error handling in fs/ntfs/mft.c::ntfs_sync_mft_mirror().
401 - Clear the page uptodate flag in fs/ntfs/mft.c::ntfs_sync_mft_mirror()
402 to ensure noone can see the page whilst the mst fixups are applied.
403 - Remove the no longer needed fs/ntfs/mft.c::ntfs_mft_writepage() and
404 fs/ntfs/mft.c::try_map_mft_record().
405 - Fix callers of fs/ntfs/aops.c::mark_ntfs_record_dirty() to call it
406 with the ntfs inode which contains the page rather than the ntfs
407 inode the mft record of which is in the page.
408 - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by moving the
409 index inode bitmap inode release code from there to
410 fs/ntfs/inode.c::ntfs_clear_big_inode(). (Thanks to Christoph
411 Hellwig for spotting this.)
412 - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by taking the
413 inode semaphore around the code that sets ni->itype.index.bmp_ino to
414 NULL and reorganize the code to optimize it a bit. (Thanks to
415 Christoph Hellwig for spotting this.)
416 - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() to no longer take the
417 ntfs inode as a parameter as this is confusing and misleading and the
418 needed ntfs inode is available via NTFS_I(page->mapping->host).
419 Adapt all callers to this change.
420 - Modify fs/ntfs/mft.c::write_mft_record_nolock() and
421 fs/ntfs/aops.c::ntfs_write_mst_block() to only check the dirty state
422 of the first buffer in a record and to take this as the ntfs record
423 dirty state. We cannot look at the dirty state for subsequent
424 buffers because we might be racing with
425 fs/ntfs/aops.c::mark_ntfs_record_dirty().
426 - Move the static inline ntfs_init_big_inode() from fs/ntfs/inode.c to
427 inode.h and make fs/ntfs/inode.c::__ntfs_init_inode() non-static and
428 add a declaration for it to inode.h. Fix some compilation issues
429 that resulted due to #includes and header file interdependencies.
430 - Simplify setup of i_mode in fs/ntfs/inode.c::ntfs_read_locked_inode().
431 - Add helpers fs/ntfs/layout.h::MK_MREF() and MK_LE_MREF().
432 - Modify fs/ntfs/mft.c::map_extent_mft_record() to only verify the mft
433 record sequence number if it is specified (i.e. not zero).
434 - Add fs/ntfs/mft.[hc]::ntfs_mft_record_alloc() and various helper
435 functions used by it.
436 - Update Documentation/filesystems/ntfs.txt with instructions on how to
437 use the Device-Mapper driver with NTFS ftdisk/LDM raid. This removes
438 the linear raid problem with the Software RAID / MD driver when one
439 or more of the devices has an odd number of sectors.
441 2.1.20 - Fix two stupid bugs introduced in 2.1.18 release.
443 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_reinit_search_ctx()
444 where we did not clear ctx->al_entry but it was still set due to
445 changes in ntfs_attr_lookup() and ntfs_external_attr_find() in
447 - Fix another stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find()
448 where we forgot to unmap the extent mft record when we had finished
449 enumerating an attribute which caused a bug check to trigger when the
450 VFS calls ->clear_inode.
452 2.1.19 - Many cleanups, improvements, and a minor bug fix.
454 - Update ->setattr (fs/ntfs/inode.c::ntfs_setattr()) to refuse to
455 change the uid, gid, and mode of an inode as we do not support NTFS
457 - Remove BKL use from ntfs_setattr() syncing up with the rest of the
459 - Get rid of the ugly transparent union in fs/ntfs/dir.c::ntfs_readdir()
460 and ntfs_filldir() as per suggestion from Al Viro.
461 - Change '\0' and L'\0' to simply 0 as per advice from Linus Torvalds.
462 - Update ->truncate (fs/ntfs/inode.c::ntfs_truncate()) to check if the
463 inode size has changed and to only output an error if so.
464 - Rename fs/ntfs/attrib.h::attribute_value_length() to ntfs_attr_size().
465 - Add le{16,32,64} as well as sle{16,32,64} data types to
467 - Change ntfschar to be le16 instead of u16 in fs/ntfs/types.h.
468 - Add le versions of VCN, LCN, and LSN called leVCN, leLCN, and leLSN,
469 respectively, to fs/ntfs/types.h.
470 - Update endianness conversion macros in fs/ntfs/endian.h to use the
471 new types as appropriate.
472 - Do proper type casting when using sle64_to_cpup() in fs/ntfs/dir.c
474 - Add leMFT_REF data type to fs/ntfs/layout.h.
475 - Update all NTFS header files with the new little endian data types.
476 Affected files are fs/ntfs/layout.h, logfile.h, and time.h.
477 - Do proper type casting when using ntfs_is_*_recordp() in
478 fs/ntfs/logfile.c, mft.c, and super.c.
479 - Fix all the sparse bitwise warnings. Had to change all the typedef
480 enums storing little endian values to simple enums plus a typedef for
481 the datatype to make sparse happy.
482 - Fix a bug found by the new sparse bitwise warnings where the default
483 upcase table was defined as a pointer to wchar_t rather than ntfschar
484 in fs/ntfs/ntfs.h and super.c.
485 - Change {const_,}cpu_to_le{16,32}(0) to just 0 as suggested by Al Viro.
487 2.1.18 - Fix scheduling latencies at mount time as well as an endianness bug.
489 - Remove vol->nr_mft_records as it was pretty meaningless and optimize
490 the calculation of total/free inodes as used by statfs().
491 - Fix scheduling latencies in ntfs_fill_super() by dropping the BKL
492 because the code itself is using the ntfs_lock semaphore which
493 provides safe locking. (Ingo Molnar)
494 - Fix a potential bug in fs/ntfs/mft.c::map_extent_mft_record() that
495 could occur in the future for when we start closing/freeing extent
496 inodes if we don't set base_ni->ext.extent_ntfs_inos to NULL after
498 - Rename {find,lookup}_attr() to ntfs_attr_{find,lookup}() as well as
499 find_external_attr() to ntfs_external_attr_find() to cleanup the
500 namespace a bit and to be more consistent with libntfs.
501 - Rename {{re,}init,get,put}_attr_search_ctx() to
502 ntfs_attr_{{re,}init,get,put}_search_ctx() as well as the type
503 attr_search_context to ntfs_attr_search_ctx.
504 - Force use of ntfs_attr_find() in ntfs_attr_lookup() when searching
505 for the attribute list attribute itself.
506 - Fix endianness bug in ntfs_external_attr_find().
507 - Change ntfs_{external_,}attr_find() to return 0 on success, -ENOENT
508 if the attribute is not found, and -EIO on real error. In the case
509 of -ENOENT, the search context is updated to describe the attribute
510 before which the attribute being searched for would need to be
511 inserted if such an action were to be desired and in the case of
512 ntfs_external_attr_find() the search context is also updated to
513 indicate the attribute list entry before which the attribute list
514 entry of the attribute being searched for would need to be inserted
515 if such an action were to be desired. Also make ntfs_find_attr()
516 static and remove its prototype from attrib.h as it is not used
517 anywhere other than attrib.c. Update ntfs_attr_lookup() and all
518 callers of ntfs_{external,}attr_{find,lookup}() for the new return
520 - Minor cleanup of fs/ntfs/inode.c::ntfs_init_locked_inode().
522 2.1.17 - Fix bugs in mount time error code paths and other updates.
524 - Implement bitmap modification code (fs/ntfs/bitmap.[hc]). This
525 includes functions to set/clear a single bit or a run of bits.
526 - Add fs/ntfs/attrib.[hc]::ntfs_find_vcn() which returns the locked
527 runlist element containing a particular vcn. It also takes care of
528 mapping any needed runlist fragments.
529 - Implement cluster (de-)allocation code (fs/ntfs/lcnalloc.[hc]).
530 - Load attribute definition table from $AttrDef at mount time.
531 - Fix bugs in mount time error code paths involving (de)allocation of
532 the default and volume upcase tables.
533 - Remove ntfs_nr_mounts as it is no longer used.
535 2.1.16 - Implement access time updates, file sync, async io, and read/writev.
537 - Add support for readv/writev and aio_read/aio_write (fs/ntfs/file.c).
538 This is done by setting the appropriate file operations pointers to
539 the generic helper functions provided by mm/filemap.c.
540 - Implement fsync, fdatasync, and msync both for files (fs/ntfs/file.c)
541 and directories (fs/ntfs/dir.c).
542 - Add support for {a,m,c}time updates to inode.c::ntfs_write_inode().
543 Note, except for the root directory and any other system files opened
544 by the user, the system files will not have their access times
545 updated as they are only accessed at the inode level an hence the
546 file level functions which cause the times to be updated are never
549 2.1.15 - Invalidate quotas when (re)mounting read-write.
551 - Add new element itype.index.collation_rule to the ntfs inode
552 structure and set it appropriately in ntfs_read_locked_inode().
553 - Implement a new inode type "index" to allow efficient access to the
554 indices found in various system files and adapt inode handling
555 accordingly (fs/ntfs/inode.[hc]). An index inode is essentially an
556 attribute inode (NInoAttr() is true) with an attribute type of
557 AT_INDEX_ALLOCATION. As such, it is no longer allowed to call
558 ntfs_attr_iget() with an attribute type of AT_INDEX_ALLOCATION as
559 there would be no way to distinguish between normal attribute inodes
560 and index inodes. The function to obtain an index inode is
561 ntfs_index_iget() and it uses the helper function
562 ntfs_read_locked_index_inode(). Note, we do not overload
563 ntfs_attr_iget() as indices consist of multiple attributes so using
564 ntfs_attr_iget() to obtain an index inode would be confusing.
565 - Ensure that there is no overflow when doing page->index <<
566 PAGE_CACHE_SHIFT by casting page->index to s64 in fs/ntfs/aops.c.
567 - Use atomic kmap instead of kmap() in fs/ntfs/aops.c::ntfs_read_page()
568 and ntfs_read_block().
569 - Use case sensitive attribute lookups instead of case insensitive ones.
570 - Lock all page cache pages belonging to mst protected attributes while
571 accessing them to ensure we never see corrupt data while the page is
573 - Add framework for generic ntfs collation (fs/ntfs/collation.[hc]).
574 We have ntfs_is_collation_rule_supported() to check if the collation
575 rule you want to use is supported and ntfs_collation() which actually
576 collates two data items. We currently only support COLLATION_BINARY
577 and COLLATION_NTOFS_ULONG but support for other collation rules will
578 be added as the need arises.
579 - Add a new type, ntfs_index_context, to allow retrieval of an index
580 entry using the corresponding index key. To get an index context,
581 use ntfs_index_ctx_get() and to release it, use ntfs_index_ctx_put().
582 This also adds a new slab cache for the index contexts. To lookup a
583 key in an index inode, use ntfs_index_lookup(). After modifying an
584 index entry, call ntfs_index_entry_flush_dcache_page() followed by
585 ntfs_index_entry_mark_dirty() to ensure the changes are written out
586 to disk. For details see fs/ntfs/index.[hc]. Note, at present, if
587 an index entry is in the index allocation attribute rather than the
588 index root attribute it will not be written out (you will get a
589 warning message about discarded changes instead).
590 - Load the quota file ($Quota) and check if quota tracking is enabled
591 and if so, mark the quotas out of date. This causes windows to
592 rescan the volume on boot and update all quota entries.
593 - Add a set_page_dirty address space operation for ntfs_m[fs]t_aops.
594 It is simply set to __set_page_dirty_nobuffers() to make sure that
595 running set_page_dirty() on a page containing mft/ntfs records will
596 not affect the dirty state of the page buffers.
597 - Add fs/ntfs/index.c::__ntfs_index_entry_mark_dirty() which sets all
598 buffers that are inside the ntfs record in the page dirty after which
599 it sets the page dirty. This allows ->writepage to only write the
600 dirty index records rather than having to write all the records in
601 the page. Modify fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to
602 use this rather than __set_page_dirty_nobuffers().
603 - Implement fs/ntfs/aops.c::ntfs_write_mst_block() which enables the
604 writing of page cache pages belonging to mst protected attributes
605 like the index allocation attribute in directory indices and other
606 indices like $Quota/$Q, etc. This means that the quota is now marked
607 out of date on all volumes rather than only on ones where the quota
608 defaults entry is in the index root attribute of the $Quota/$Q index.
610 2.1.14 - Fix an NFSd caused deadlock reported by several users.
612 - Modify fs/ntfs/ntfs_readdir() to copy the index root attribute value
613 to a buffer so that we can put the search context and unmap the mft
614 record before calling the filldir() callback. We need to do this
615 because of NFSd which calls ->lookup() from its filldir callback()
616 and this causes NTFS to deadlock as ntfs_lookup() maps the mft record
617 of the directory and since ntfs_readdir() has got it mapped already
618 ntfs_lookup() deadlocks.
620 2.1.13 - Enable overwriting of resident files and housekeeping of system files.
622 - Implement writing of mft records (fs/ntfs/mft.[hc]), which includes
623 keeping the mft mirror in sync with the mft when mirrored mft records
624 are written. The functions are write_mft_record{,_nolock}(). The
625 implementation is quite rudimentary for now with lots of things not
626 implemented yet but I am not sure any of them can actually occur so
627 I will wait for people to hit each one and only then implement it.
628 - Commit open system inodes at umount time. This should make it
629 virtually impossible for sync_mft_mirror_umount() to ever be needed.
630 - Implement ->write_inode (fs/ntfs/inode.c::ntfs_write_inode()) for the
631 ntfs super operations. This gives us inode writing via the VFS inode
632 dirty code paths. Note: Access time updates are not implemented yet.
633 - Implement fs/ntfs/mft.[hc]::{,__}mark_mft_record_dirty() and make
634 fs/ntfs/aops.c::ntfs_writepage() and ntfs_commit_write() use it, thus
635 finally enabling resident file overwrite! (-8 This also includes a
636 placeholder for ->writepage (ntfs_mft_writepage()), which for now
637 just redirties the page and returns. Also, at umount time, we for
638 now throw away all mft data page cache pages after the last call to
639 ntfs_commit_inode() in the hope that all inodes will have been
640 written out by then and hence no dirty (meta)data will be lost. We
641 also check for this case and emit an error message telling the user
643 - Use set_page_writeback() and end_page_writeback() in the resident
644 attribute code path of fs/ntfs/aops.c::ntfs_writepage() otherwise
645 the radix-tree tag PAGECACHE_TAG_DIRTY remains set even though the
647 - Implement ntfs_mft_writepage() so it now checks if any of the mft
648 records in the page are dirty and if so redirties the page and
649 returns. Otherwise it just returns (after doing set_page_writeback(),
650 unlock_page(), end_page_writeback() or the radix-tree tag
651 PAGECACHE_TAG_DIRTY remains set even though the page is clean), thus
652 alowing the VM to do with the page as it pleases. Also, at umount
653 time, now only throw away dirty mft (meta)data pages if dirty inodes
654 are present and ask the user to email us if they see this happening.
655 - Add functions ntfs_{clear,set}_volume_flags(), to modify the volume
656 information flags (fs/ntfs/super.c).
657 - Mark the volume dirty when (re)mounting read-write and mark it clean
658 when unmounting or remounting read-only. If any volume errors are
659 found, the volume is left marked dirty to force chkdsk to run.
660 - Add code to set the NT4 compatibility flag when (re)mounting
661 read-write for newer NTFS versions but leave it commented out for now
662 since we do not make any modifications that are NTFS 1.2 specific yet
663 and since setting this flag breaks Captive-NTFS which is not nice.
664 This code must be enabled once we start writing NTFS 1.2 specific
665 changes otherwise Windows NTFS driver might crash / cause corruption.
667 2.1.12 - Fix the second fix to the decompression engine and some cleanups.
669 - Add a new address space operations struct, ntfs_mst_aops, for mst
670 protected attributes. This is because the default ntfs_aops do not
671 make sense with mst protected data and were they to write anything to
672 such an attribute they would cause data corruption so we provide
673 ntfs_mst_aops which does not have any write related operations set.
674 - Cleanup dirty ntfs inode handling (fs/ntfs/inode.[hc]) which also
675 includes an adapted ntfs_commit_inode() and an implementation of
676 ntfs_write_inode() which for now just cleans dirty inodes without
677 writing them (it does emit a warning that this is happening).
678 - Undo the second decompression engine fix (see 2.1.9 release ChangeLog
679 entry) as it was only fixing a theoretical bug but at the same time
680 it badly broke the handling of sparse and uncompressed compression
683 2.1.11 - Driver internal cleanups.
685 - Only build logfile.o if building the driver with read-write support.
686 - Really final white space cleanups.
687 - Use generic_ffs() instead of ffs() in logfile.c which allows the
688 log_page_size variable to be optimized by gcc into a constant.
689 - Rename uchar_t to ntfschar everywhere as uchar_t is unsigned 1-byte
690 char as defined by POSIX and as found on some systems.
692 2.1.10 - Force read-only (re)mounting of volumes with unsupported volume flags.
694 - Finish off the white space cleanups (remove trailing spaces, etc).
695 - Clean up ntfs_fill_super() and ntfs_read_inode_mount() by removing
696 the kludges around the first iget(). Instead of (re)setting ->s_op
697 we have the $MFT inode set up by explicit new_inode() / set ->i_ino /
698 insert_inode_hash() / call ntfs_read_inode_mount() directly. This
699 kills the need for second super_operations and allows to return error
700 from ntfs_read_inode_mount() without resorting to ugly "poisoning"
702 - Force read-only (re)mounting if any of the following bits are set in
703 the volume information flags:
704 VOLUME_IS_DIRTY, VOLUME_RESIZE_LOG_FILE,
705 VOLUME_UPGRADE_ON_MOUNT, VOLUME_DELETE_USN_UNDERWAY,
706 VOLUME_REPAIR_OBJECT_ID, VOLUME_MODIFIED_BY_CHKDSK
707 To make this easier we define VOLUME_MUST_MOUNT_RO_MASK with all the
708 above bits set so the test is made easy.
710 2.1.9 - Fix two bugs in decompression engine.
712 - Fix a bug where we would not always detect that we have reached the
713 end of a compression block because we were ending at minus one byte
714 which is effectively the same as being at the end. The fix is to
715 check whether the uncompressed buffer has been fully filled and if so
716 we assume we have reached the end of the compression block. A big
717 thank you to Marcin Gibuła for the bug report, the assistance in
718 tracking down the bug and testing the fix.
719 - Fix a possible bug where when a compressed read is truncated to the
720 end of the file, the offset inside the last page was not truncated.
722 2.1.8 - Handle $MFT mirror and $LogFile, improve time handling, and cleanups.
724 - Use get_bh() instead of manual atomic_inc() in fs/ntfs/compress.c.
725 - Modify fs/ntfs/time.c::ntfs2utc(), get_current_ntfs_time(), and
726 utc2ntfs() to work with struct timespec instead of time_t on the
727 Linux UTC time side thus preserving the full precision of the NTFS
728 time and only loosing up to 99 nano-seconds in the Linux UTC time.
729 - Move fs/ntfs/time.c to fs/ntfs/time.h and make the time functions
731 - Remove unused ntfs_dirty_inode().
732 - Cleanup super operations declaration in fs/ntfs/super.c.
733 - Wrap flush_dcache_mft_record_page() in #ifdef NTFS_RW.
734 - Add NInoTestSetFoo() and NInoTestClearFoo() macro magic to
735 fs/ntfs/inode.h and use it to declare NInoTest{Set,Clear}Dirty.
736 - Move typedefs for ntfs_attr and test_t from fs/ntfs/inode.c to
737 fs/ntfs/inode.h so they can be used elsewhere.
738 - Determine the mft mirror size as the number of mirrored mft records
739 and store it in ntfs_volume->mftmirr_size (fs/ntfs/super.c).
740 - Load the mft mirror at mount time and compare the mft records stored
741 in it to the ones in the mft. Force a read-only mount if the two do
742 not match (fs/ntfs/super.c).
743 - Fix type casting related warnings on 64-bit architectures. Thanks
744 to Meelis Roos for reporting them.
745 - Move %L to %ll as %L is floating point and %ll is integer which is
747 - Read the journal ($LogFile) and determine if the volume has been
748 shutdown cleanly and force a read-only mount if not (fs/ntfs/super.c
749 and fs/ntfs/logfile.c). This is a little bit of a crude check in
750 that we only look at the restart areas and not at the actual log
751 records so that there will be a very small number of cases where we
752 think that a volume is dirty when in fact it is clean. This should
753 only affect volumes that have not been shutdown cleanly and did not
754 have any pending, non-check-pointed i/o.
755 - If the $LogFile indicates a clean shutdown and a read-write (re)mount
756 is requested, empty $LogFile by overwriting it with 0xff bytes to
757 ensure that Windows cannot cause data corruption by replaying a stale
758 journal after Linux has written to the volume.
760 2.1.7 - Enable NFS exporting of mounted NTFS volumes.
762 - Set i_generation in the VFS inode from the seq_no of the NTFS inode.
763 - Make ntfs_lookup() NFS export safe, i.e. use d_splice_alias(), etc.
764 - Implement ->get_dentry() in fs/ntfs/namei.c::ntfs_get_dentry() as the
765 default doesn't allow inode number 0 which is a valid inode on NTFS
766 and even if it did allow that it uses iget() instead of ntfs_iget()
767 which makes it useless for us.
768 - Implement ->get_parent() in fs/ntfs/namei.c::ntfs_get_parent() as the
769 default just returns -EACCES which is not very useful.
770 - Define export operations (->s_export_op) for NTFS (ntfs_export_ops)
771 and set them up in the super block at mount time (super.c) this
772 allows mounted NTFS volumes to be exported via NFS.
773 - Add missing return -EOPNOTSUPP; in
774 fs/ntfs/aops.c::ntfs_commit_nonresident_write().
775 - Enforce no atime and no dir atime updates at mount/remount time as
776 they are not implemented yet anyway.
777 - Move a few assignments in fs/ntfs/attrib.c::load_attribute_list() to
778 after a NULL check. Thanks to Dave Jones for pointing this out.
780 2.1.6 - Fix minor bug in handling of compressed directories.
782 - Fix bug in handling of compressed directories. A compressed
783 directory is not really compressed so when we set the ->i_blocks
784 field of a compressed directory inode we were setting it from the
785 non-existing field ni->itype.compressed.size which gave random
786 results... For directories we now always use ni->allocated_size.
788 2.1.5 - Fix minor bug in attribute list attribute handling.
790 - Fix bug in attribute list handling. Actually it is not as much a bug
791 as too much protection in that we were not allowing attribute lists
792 which waste space on disk while Windows XP clearly allows it and in
793 fact creates such attribute lists so our driver was failing.
794 - Update NTFS documentation ready for 2.6 kernel release.
796 2.1.4 - Reduce compiler requirements.
798 - Remove all uses of unnamed structs and unions in the driver to make
799 old and newer gcc versions happy. Makes it a bit uglier IMO but at
800 least people will stop hassling me about it.
802 2.1.3 - Important bug fixes in corner cases.
804 - super.c::parse_ntfs_boot_sector(): Correct the check for 64-bit
805 clusters. (Philipp Thomas)
806 - attrib.c::load_attribute_list(): Fix bug when initialized_size is a
807 multiple of the block_size but not the cluster size. (Szabolcs
808 Szakacsits <szaka@sienet.hu>)
810 2.1.2 - Important bug fixes aleviating the hangs in statfs.
812 - Fix buggy free cluster and free inode determination logic.
814 2.1.1 - Minor updates.
816 - Add handling for initialized_size != data_size in compressed files.
817 - Reduce function local stack usage from 0x3d4 bytes to just noise in
818 fs/ntfs/upcase.c. (Randy Dunlap <rddunlap@osdl.ord>)
819 - Remove compiler warnings for newer gcc.
820 - Pages are no longer kmapped by mm/filemap.c::generic_file_write()
821 around calls to ->{prepare,commit}_write. Adapt NTFS appropriately
822 in fs/ntfs/aops.c::ntfs_prepare_nonresident_write() by using
823 kmap_atomic(KM_USER0).
825 2.1.0 - First steps towards write support: implement file overwrite.
827 - Add configuration option for developmental write support with an
828 appropriately scary configuration help text.
829 - Initial implementation of fs/ntfs/aops.c::ntfs_writepage() and its
830 helper fs/ntfs/aops.c::ntfs_write_block(). This enables mmap(2) based
831 overwriting of existing files on ntfs. Note: Resident files are
832 only written into memory, and not written out to disk at present, so
833 avoid writing to files smaller than about 1kiB.
834 - Initial implementation of fs/ntfs/aops.c::ntfs_prepare_write(), its
835 helper fs/ntfs/aops.c::ntfs_prepare_nonresident_write() and their
836 counterparts, fs/ntfs/aops.c::ntfs_commit_write(), and
837 fs/ntfs/aops.c::ntfs_commit_nonresident_write(), respectively. Also,
838 add generic_file_write() to the ntfs file operations (fs/ntfs/file.c).
839 This enables write(2) based overwriting of existing files on ntfs.
840 Note: As with mmap(2) based overwriting, resident files are only
841 written into memory, and not written out to disk at present, so avoid
842 writing to files smaller than about 1kiB.
843 - Implement ->truncate (fs/ntfs/inode.c::ntfs_truncate()) and
844 ->setattr() (fs/ntfs/inode.c::ntfs_setattr()) inode operations for
845 files with the purpose of intercepting and aborting all i_size
846 changes which we do not support yet. ntfs_truncate() actually only
847 emits a warning message but AFAICS our interception of i_size changes
848 elsewhere means ntfs_truncate() never gets called for i_size changes.
849 It is only called from generic_file_write() when we fail in
850 ntfs_prepare_{,nonresident_}write() in order to discard any
851 instantiated buffers beyond i_size. Thus i_size is not actually
852 changed so our warning message is enough. Unfortunately it is not
853 possible to easily determine if i_size is being changed or not hence
854 we just emit an appropriately worded error message.
856 2.0.25 - Small bug fixes and cleanups.
858 - Unlock the page in an out of memory error code path in
859 fs/ntfs/aops.c::ntfs_read_block().
860 - If fs/ntfs/aops.c::ntfs_read_page() is called on an uptodate page,
861 just unlock the page and return. (This can happen due to ->writepage
862 clearing PageUptodate() during write out of MstProtected()
864 - Remove leaked write code again.
868 - Treat BUG_ON() as ASSERT() not VERIFY(), i.e. do not use side effects
869 inside BUG_ON(). (Adam J. Richter)
870 - Split logical OR expressions inside BUG_ON() into individual BUG_ON()
871 calls for improved debugging. (Adam J. Richter)
872 - Add errors flag to the ntfs volume state, accessed via
873 NVol{,Set,Clear}Errors(vol).
874 - Do not allow read-write remounts of read-only volumes with errors.
875 - Clarify comment for ntfs file operation sendfile which was added by
876 Christoph Hellwig a while ago (just using generic_file_sendfile())
877 to say that ntfs ->sendfile is only used for the case where the
878 source data is on the ntfs partition and the destination is
879 somewhere else, i.e. nothing we need to concern ourselves with.
880 - Add generic_file_write() as our ntfs file write operation.
882 2.0.23 - Major bug fixes (races, deadlocks, non-i386 architectures).
884 - Massive internal locking changes to mft record locking. Fixes lock
885 recursion and replaces the mrec_lock read/write semaphore with a
886 mutex. Also removes the now superfluous mft_count. This fixes several
887 race conditions and deadlocks, especially in the future write code.
888 - Fix ntfs over loopback for compressed files by adding an
889 optimization barrier. (gcc was screwing up otherwise ?)
890 - Miscellaneous cleanups all over the code and a fix or two in error
892 Thanks go to Christoph Hellwig for pointing out the following two:
893 - Remove now unused function fs/ntfs/malloc.h::vmalloc_nofs().
894 - Fix ntfs_free() for ia64 and parisc by checking for VMALLOC_END, too.
896 2.0.22 - Cleanups, mainly to ntfs_readdir(), and use C99 initializers.
898 - Change fs/ntfs/dir.c::ntfs_reddir() to only read/write ->f_pos once
899 at entry/exit respectively.
900 - Use C99 initializers for structures.
901 - Remove unused variable blocks from fs/ntfs/aops.c::ntfs_read_block().
903 2.0.21 - Check for, and refuse to work with too large files/directories/volumes.
905 - Limit volume size at mount time to 2TiB on architectures where
906 unsigned long is 32-bits (fs/ntfs/super.c::parse_ntfs_boot_sector()).
907 This is the most we can do without overflowing the 32-bit limit of
908 the block device size imposed on us by sb_bread() and sb_getblk()
910 - Limit file/directory size at open() time to 16TiB on architectures
911 where unsigned long is 32-bits (fs/ntfs/file.c::ntfs_file_open() and
912 fs/ntfs/dir.c::ntfs_dir_open()). This is the most we can do without
913 overflowing the page cache page index.
915 2.0.20 - Support non-resident directory index bitmaps, fix page leak in readdir.
917 - Move the directory index bitmap to use an attribute inode instead of
918 having special fields for it inside the ntfs inode structure. This
919 means that the index bitmaps now use the page cache for i/o, too,
920 and also as a side effect we get support for non-resident index
922 - Simplify/cleanup error handling in fs/ntfs/dir.c::ntfs_readdir() and
923 fix a page leak that manifested itself in some cases.
924 - Add fs/ntfs/inode.c::ntfs_put_inode(), which we need to release the
925 index bitmap inode on the final iput().
927 2.0.19 - Fix race condition, improvements, and optimizations in i/o interface.
929 - Apply block optimization added to fs/ntfs/aops.c::ntfs_read_block()
930 to fs/ntfs/compress.c::ntfs_file_read_compressed_block() as well.
931 - Drop the "file" from ntfs_file_read_compressed_block().
932 - Rename fs/ntfs/aops.c::ntfs_enb_buffer_read_async() to
933 ntfs_end_buffer_async_read() (more like the fs/buffer.c counterpart).
934 - Update ntfs_end_buffer_async_read() with the improved logic from
935 its updated counterpart fs/buffer.c::end_buffer_async_read(). Apply
936 further logic improvements to better determine when we set PageError.
937 - Update submission of buffers in fs/ntfs/aops.c::ntfs_read_block() to
938 check for the buffers being uptodate first in line with the updated
939 fs/buffer.c::block_read_full_page(). This plugs a small race
942 2.0.18 - Fix race condition in reading of compressed files.
944 - There was a narrow window between checking a buffer head for being
945 uptodate and locking it in ntfs_file_read_compressed_block(). We now
946 lock the buffer and then check whether it is uptodate or not.
948 2.0.17 - Cleanups and optimizations - shrinking the ToDo list.
950 - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to return an error
951 code and update callers, i.e. ntfs_iget(), to pass that error code
952 up instead of just using -EIO.
953 - Modifications to super.c to ensure that both mount and remount
954 cannot set any write related options when the driver is compiled
956 - Optimize block resolution in fs/ntfs/aops.c::ntfs_read_block() to
957 cache the current runlist element. This should improve performance
958 when reading very large and/or very fragmented data.
960 2.0.16 - Convert access to $MFT/$BITMAP to attribute inode API.
962 - Fix a stupid bug introduced in 2.0.15 where we were unmapping the
963 wrong inode in fs/ntfs/inode.c::ntfs_attr_iget().
964 - Fix debugging check in fs/ntfs/aops.c::ntfs_read_block().
965 - Convert $MFT/$BITMAP access to attribute inode API and remove all
966 remnants of the ugly mftbmp address space and operations hack. This
967 means we finally have only one readpage function as well as only one
968 async io completion handler. Yey! The mft bitmap is now just an
969 attribute inode and is accessed from vol->mftbmp_ino just as if it
970 were a normal file. Fake inodes rule. (-:
972 2.0.15 - Fake inodes based attribute i/o via the pagecache, fixes and cleanups.
974 - Fix silly bug in fs/ntfs/super.c::parse_options() which was causing
975 remounts to fail when the partition had an entry in /etc/fstab and
976 the entry specified the nls= option.
977 - Apply same macro magic used in fs/ntfs/inode.h to fs/ntfs/volume.h to
978 expand all the helper functions NVolFoo(), NVolSetFoo(), and
980 - Move copyright statement from driver initialisation message to
981 module description (fs/super.c). This makes the initialisation
982 message fit on one line and fits in better with rest of kernel.
983 - Update fs/ntfs/attrib.c::map_run_list() to work on both real and
984 attribute inodes, and both for files and directories.
985 - Implement fake attribute inodes allowing all attribute i/o to go via
986 the page cache and to use all the normal vfs/mm functionality:
987 - Add ntfs_attr_iget() and its helper ntfs_read_locked_attr_inode()
989 - Add needed cleanup code to ntfs_clear_big_inode().
990 - Merge address space operations for files and directories (aops.c),
991 now just have ntfs_aops:
993 end_buffer_read_attr_async() -> ntfs_end_buffer_read_async(),
994 ntfs_attr_read_block() -> ntfs_read_block(),
995 ntfs_file_read_page() -> ntfs_readpage().
996 - Rewrite fs/ntfs/aops.c::ntfs_readpage() to work on both real and
997 attribute inodes, and both for files and directories.
998 - Remove obsolete fs/ntfs/aops.c::ntfs_mst_readpage().
1000 2.0.14 - Run list merging code cleanup, minor locking changes, typo fixes.
1002 - Change fs/ntfs/super.c::ntfs_statfs() to not rely on BKL by moving
1003 the locking out of super.c::get_nr_free_mft_records() and taking and
1004 dropping the mftbmp_lock rw_semaphore in ntfs_statfs() itself.
1005 - Bring attribute runlist merging code (fs/ntfs/attrib.c) in sync with
1006 current userspace ntfs library code. This means that if a merge
1007 fails the original runlists are always left unmodified instead of
1008 being silently corrupted.
1011 2.0.13 - Use iget5_locked() in preparation for fake inodes and small cleanups.
1013 - Remove nr_mft_bits and the now superfluous union with nr_mft_records
1014 from ntfs_volume structure.
1015 - Remove nr_lcn_bits and the now superfluous union with nr_clusters
1016 from ntfs_volume structure.
1017 - Use iget5_locked() and friends instead of conventional iget(). Wrap
1018 the call in fs/ntfs/inode.c::ntfs_iget() and update callers of iget()
1019 to use ntfs_iget(). Leave only one iget() call at mount time so we
1020 don't need an ntfs_iget_mount().
1021 - Change fs/ntfs/inode.c::ntfs_new_extent_inode() to take mft_no as an
1022 additional argument.
1024 2.0.12 - Initial cleanup of address space operations following 2.0.11 changes.
1026 - Merge fs/ntfs/aops.c::end_buffer_read_mst_async() and
1027 fs/ntfs/aops.c::end_buffer_read_file_async() into one function
1028 fs/ntfs/aops.c::end_buffer_read_attr_async() using NInoMstProtected()
1029 to determine whether to apply mst fixups or not.
1030 - Above change allows merging fs/ntfs/aops.c::ntfs_file_read_block()
1031 and fs/ntfs/aops.c::ntfs_mst_readpage() into one function
1032 fs/ntfs/aops.c::ntfs_attr_read_block(). Also, create a tiny wrapper
1033 fs/ntfs/aops.c::ntfs_mst_readpage() to transform the parameters from
1034 the VFS readpage function prototype to the ntfs_attr_read_block()
1037 2.0.11 - Initial preparations for fake inode based attribute i/o.
1039 - Move definition of ntfs_inode_state_bits to fs/ntfs/inode.h and
1040 do some macro magic (adapted from include/linux/buffer_head.h) to
1041 expand all the helper functions NInoFoo(), NInoSetFoo(), and
1043 - Add new flag to ntfs_inode_state_bits: NI_Sparse.
1044 - Add new fields to ntfs_inode structure to allow use of fake inodes
1045 for attribute i/o: type, name, name_len. Also add new state bits:
1046 NI_Attr, which, if set, indicates the inode is a fake inode, and
1047 NI_MstProtected, which, if set, indicates the attribute uses multi
1048 sector transfer protection, i.e. fixups need to be applied after
1049 reads and before/after writes.
1050 - Rename fs/ntfs/inode.c::ntfs_{new,clear,destroy}_inode() to
1051 ntfs_{new,clear,destroy}_extent_inode() and update callers.
1052 - Use ntfs_clear_extent_inode() in fs/ntfs/inode.c::__ntfs_clear_inode()
1053 instead of ntfs_destroy_extent_inode().
1054 - Cleanup memory deallocations in {__,}ntfs_clear_{,big_}inode().
1055 - Make all operations on ntfs inode state bits use the NIno* functions.
1056 - Set up the new ntfs inode fields and state bits in
1057 fs/ntfs/inode.c::ntfs_read_inode() and add appropriate cleanup of
1058 allocated memory to __ntfs_clear_inode().
1059 - Cleanup ntfs_inode structure a bit for better ordering of elements
1060 w.r.t. their size to allow better packing of the structure in memory.
1062 2.0.10 - There can only be 2^32 - 1 inodes on an NTFS volume.
1064 - Add check at mount time to verify that the number of inodes on the
1065 volume does not exceed 2^32 - 1, which is the maximum allowed for
1066 NTFS according to Microsoft.
1067 - Change mft_no member of ntfs_inode structure to be unsigned long.
1068 Update all users. This makes ntfs_inode->mft_no just a copy of struct
1069 inode->i_ino. But we can't just always use struct inode->i_ino and
1070 remove mft_no because extent inodes do not have an attached struct
1073 2.0.9 - Decompression engine now uses a single buffer and other cleanups.
1075 - Change decompression engine to use a single buffer protected by a
1076 spin lock instead of per-CPU buffers. (Rusty Russell)
1077 - Do not update cb_pos when handling a partial final page during
1078 decompression of a sparse compression block, as the value is later
1079 reset without being read/used. (Rusty Russell)
1080 - Switch to using the new KM_BIO_SRC_IRQ for atomic kmap()s. (Andrew
1082 - Change buffer size in ntfs_readdir()/ntfs_filldir() to use
1083 NLS_MAX_CHARSET_SIZE which makes the buffers almost 1kiB each but
1084 it also makes everything safer so it is a good thing.
1085 - Miscellaneous minor cleanups to comments.
1087 2.0.8 - Major updates for handling of case sensitivity and dcache aliasing.
1089 Big thanks go to Al Viro and other inhabitants of #kernel for investing
1090 their time to discuss the case sensitivity and dcache aliasing issues.
1092 - Remove unused source file fs/ntfs/attraops.c.
1093 - Remove show_inodes mount option(s), thus dropping support for
1094 displaying of short file names.
1095 - Remove deprecated mount option posix.
1096 - Restore show_sys_files mount option.
1097 - Add new mount option case_sensitive, to determine if the driver
1098 treats file names as case sensitive or not. If case sensitive, create
1099 file names in the POSIX namespace. Otherwise create file names in the
1100 LONG/WIN32 namespace. Note, files remain accessible via their short
1101 file name, if it exists.
1102 - Remove really dumb logic bug in boot sector recovery code.
1103 - Fix dcache aliasing issues wrt short/long file names via changes
1104 to fs/ntfs/dir.c::ntfs_lookup_inode_by_name() and
1105 fs/ntfs/namei.c::ntfs_lookup():
1106 - Add additional argument to ntfs_lookup_inode_by_name() in which we
1107 return information about the matching file name if the case is not
1108 matching or the match is a short file name. See comments above the
1109 function definition for details.
1110 - Change ntfs_lookup() to only create dcache entries for the correctly
1111 cased file name and only for the WIN32 namespace counterpart of DOS
1112 namespace file names. This ensures we have only one dentry per
1113 directory and also removes all dcache aliasing issues between short
1114 and long file names once we add write support. See comments above
1115 function for details.
1116 - Fix potential 1 byte overflow in fs/ntfs/unistr.c::ntfs_ucstonls().
1118 2.0.7 - Minor cleanups and updates for changes in core kernel code.
1120 - Remove much of the NULL struct element initializers.
1121 - Various updates to make compatible with recent kernels.
1122 - Remove defines of MAX_BUF_PER_PAGE and include linux/buffer_head.h
1123 in fs/ntfs/ntfs.h instead.
1124 - Remove no longer needed KERNEL_VERSION checks. We are now in the
1125 kernel proper so they are no longer needed.
1127 2.0.6 - Major bugfix to make compatible with other kernel changes.
1129 - Initialize the mftbmp address space properly now that there are more
1130 fields in the struct address_space. This was leading to hangs and
1131 oopses on umount since 2.5.12 because of changes to other parts of
1132 the kernel. We probably want a kernel generic init_address_space()
1134 - Drop BKL from ntfs_readdir() after consultation with Al Viro. The
1135 only caller of ->readdir() is vfs_readdir() which holds i_sem during
1136 the call, and i_sem is sufficient protection against changes in the
1137 directory inode (including ->i_size).
1138 - Use generic_file_llseek() for directories (as opposed to
1139 default_llseek()) as this downs i_sem instead of the BKL which is
1140 what we now need for exclusion against ->f_pos changes considering we
1141 no longer take the BKL in ntfs_readdir().
1143 2.0.5 - Major bugfix. Buffer overflow in extent inode handling.
1145 - No need to set old blocksize in super.c::ntfs_fill_super() as the
1146 VFS does so via invocation of deactivate_super() calling
1147 fs->fill_super() calling block_kill_super() which does it.
1148 - BKL moved from VFS into dir.c::ntfs_readdir(). (Linus Torvalds)
1149 -> Do we really need it? I don't think so as we have exclusion on
1150 the directory ntfs_inode rw_semaphore mrec_lock. We mmight have to
1151 move the ->f_pos accesses under the mrec_lock though. Check this...
1152 - Fix really, really, really stupid buffer overflow in extent inode
1153 handling in mft.c::map_extent_mft_record().
1155 2.0.4 - Cleanups and updates for kernel 2.5.11.
1157 - Add documentation on how to use the MD driver to be able to use NTFS
1158 stripe and volume sets in Linux and generally cleanup documentation
1160 Remove all uses of kdev_t in favour of struct block_device *:
1161 - Change compress.c::ntfs_file_read_compressed_block() to use
1162 sb_getblk() instead of getblk().
1163 - Change super.c::ntfs_fill_super() to use bdev_hardsect_size() instead
1164 of get_hardsect_size().
1165 - No need to get old blocksize in super.c::ntfs_fill_super() as
1166 fs/super.c::get_sb_bdev() already does this.
1167 - Set bh->b_bdev instead of bh->b_dev throughout aops.c.
1169 2.0.3 - Small bug fixes, cleanups, and performance improvements.
1171 - Remove some dead code from mft.c.
1172 - Optimize readpage and read_block functions throughout aops.c so that
1173 only initialized blocks are read. Non-initialized ones have their
1174 buffer head mapped, zeroed, and set up to date, without scheduling
1175 any i/o. Thanks to Al Viro for advice on how to avoid the device i/o.
1176 Thanks go to Andrew Morton for spotting the below:
1177 - Fix buglet in allocate_compression_buffers() error code path.
1178 - Call flush_dcache_page() after modifying page cache page contents in
1179 ntfs_file_readpage().
1180 - Check for existence of page buffers throughout aops.c before calling
1181 create_empty_buffers(). This happens when an I/O error occurs and the
1182 read is retried. (It also happens once writing is implemented so that
1183 needed doing anyway but I had left it for later...)
1184 - Don't BUG_ON() uptodate and/or mapped buffers throughout aops.c in
1185 readpage and read_block functions. Reasoning same as above (i.e. I/O
1186 error retries and future write code paths.)
1188 2.0.2 - Minor updates and cleanups.
1190 - Cleanup: rename mst.c::__post_read_mst_fixup to post_write_mst_fixup
1191 and cleanup the code a bit, removing the unused size parameter.
1192 - Change default fmask to 0177 and update documentation.
1193 - Change attrib.c::get_attr_search_ctx() to return the search context
1194 directly instead of taking the address of a pointer. A return value
1195 of NULL means the allocation failed. Updated all callers
1197 - Update to 2.5.9 kernel (preserving backwards compatibility) by
1198 replacing all occurences of page->buffers with page_buffers(page).
1199 - Fix minor bugs in runlist merging, also minor cleanup.
1200 - Updates to bootsector layout and mft mirror contents descriptions.
1201 - Small bug fix in error detection in unistr.c and some cleanups.
1202 - Grow name buffer allocations in unistr.c in aligned mutlipled of 64
1205 2.0.1 - Minor updates.
1207 - Make default umask correspond to documentation.
1208 - Improve documentation.
1209 - Set default mode to include execute bit. The {u,f,d}mask can be used
1210 to take it away if desired. This allows binaries to be executed from
1211 a mounted ntfs partition.
1213 2.0.0 - New version number. Remove TNG from the name. Now in the kernel.
1215 - Add kill_super, just keeping up with the vfs changes in the kernel.
1216 - Repeat some changes from tng-0.0.8 that somehow got lost on the way
1217 from the CVS import into BitKeeper.
1218 - Begin to implement proper handling of allocated_size vs
1219 initialized_size vs data_size (i.e. i_size). Done are
1220 mft.c::ntfs_mft_readpage(), aops.c::end_buffer_read_index_async(),
1221 and attrib.c::load_attribute_list().
1222 - Lock the runlist in attrib.c::load_attribute_list() while using it.
1223 - Fix memory leak in ntfs_file_read_compressed_block() and generally
1224 clean up compress.c a little, removing some uncommented/unused debug
1226 - Tidy up dir.c a little bit.
1227 - Don't bother getting the runlist in inode.c::ntfs_read_inode().
1228 - Merge mft.c::ntfs_mft_readpage() and aops.c::ntfs_index_readpage()
1229 creating aops.c::ntfs_mst_readpage(), improving the handling of
1230 holes and overflow in the process and implementing the correct
1231 equivalent of ntfs_file_get_block() in ntfs_mst_readpage() itself.
1232 I am aiming for correctness at the moment. Modularisation can come
1234 - Rename aops.c::end_buffer_read_index_async() to
1235 end_buffer_read_mst_async() and optimize the overflow checking and
1237 - Use the host of the mftbmp address space mapping to hold the ntfs
1238 volume. This is needed so the async i/o completion handler can
1239 retrieve a pointer to the volume. Hopefully this will not cause
1240 problems elsewhere in the kernel... Otherwise will need to use a
1242 - Complete implementation of proper handling of allocated_size vs
1243 initialized_size vs data_size (i.e. i_size) in whole driver.
1244 Basically aops.c is now completely rewritten.
1245 - Change NTFS driver name to just NTFS and set version number to 2.0.0
1246 to make a clear distinction from the old driver which is still on
1249 tng-0.0.8 - 08/03/2002 - Now using BitKeeper, http://linux-ntfs.bkbits.net/
1251 - Replace bdevname(sb->s_dev) with sb->s_id.
1252 - Remove now superfluous new-line characters in all callers of
1254 - Apply kludge in ntfs_read_inode(), setting i_nlink to 1 for
1255 directories. Without this the "find" utility gets very upset which is
1256 fair enough as Linux/Unix do not support directory hard links.
1257 - Further runlist merging work. (Richard Russon)
1258 - Backwards compatibility for gcc-2.95. (Richard Russon)
1259 - Update to kernel 2.5.5-pre1 and rediff the now tiny patch.
1260 - Convert to new filesystem declaration using ->ntfs_get_sb() and
1261 replacing ntfs_read_super() with ntfs_fill_super().
1262 - Set s_maxbytes to MAX_LFS_FILESIZE to avoid page cache page index
1263 overflow on 32-bit architectures.
1264 - Cleanup upcase loading code to use ntfs_(un)map_page().
1265 - Disable/reenable preemtion in critical sections of compession engine.
1266 - Replace device size determination in ntfs_fill_super() with
1267 sb->s_bdev->bd_inode->i_size (in bytes) and remove now superfluous
1268 function super.c::get_nr_blocks().
1269 - Implement a mount time option (show_inodes) allowing choice of which
1270 types of inode names readdir() returns and modify ntfs_filldir()
1271 accordingly. There are several parameters to show_inodes:
1272 system: system files
1273 win32: long file names (including POSIX file names) [DEFAULT]
1275 dos: short file names only (excluding POSIX file names)
1277 posix: same as both win32 and dos
1279 Note that the options are additive, i.e. specifying:
1280 -o show_inodes=system,show_inodes=win32,show_inodes=dos
1281 is the same as specifying:
1283 Note that the "posix" and "all" options will show all directory
1284 names, BUT the link count on each directory inode entry is set to 1,
1285 due to Linux not supporting directory hard links. This may well
1286 confuse some userspace applications, since the directory names will
1287 have the same inode numbers. Thus it is NOT advisable to use the
1288 "posix" or "all" options. We provide them only for completeness sake.
1289 - Add copies of allocated_size, initialized_size, and compressed_size to
1290 the ntfs inode structure and set them up in
1291 inode.c::ntfs_read_inode(). These reflect the unnamed data attribute
1292 for files and the index allocation attribute for directories.
1293 - Add copies of allocated_size and initialized_size to ntfs inode for
1294 $BITMAP attribute of large directories and set them up in
1295 inode.c::ntfs_read_inode().
1296 - Add copies of allocated_size and initialized_size to ntfs volume for
1297 $BITMAP attribute of $MFT and set them up in
1298 super.c::load_system_files().
1299 - Parse deprecated ntfs driver options (iocharset, show_sys_files,
1300 posix, and utf8) and tell user what the new options to use are. Note
1301 we still do support them but they will be removed with kernel 2.7.x.
1302 - Change all occurences of integer long long printf formatting to hex
1303 as printk() will not support long long integer format if/when the
1304 div64 patch goes into the kernel.
1305 - Make slab caches have stable names and change the names to what they
1306 were intended to be. These changes are required/made possible by the
1307 new slab cache name handling which removes the length limitation by
1308 requiring the caller of kmem_cache_create() to supply a stable name
1309 which is then referenced but not copied.
1310 - Rename run_list structure to run_list_element and create a new
1311 run_list structure containing a pointer to a run_list_element
1312 structure and a read/write semaphore. Adapt all users of runlists
1313 to new scheme and take and release the lock as needed. This fixes a
1314 nasty race as the run_list changes even when inodes are locked for
1315 reading and even when the inode isn't locked at all, so we really
1316 needed the serialization. We use a semaphore rather than a spinlock
1317 as memory allocations can sleep and doing everything GFP_ATOMIC
1319 - Cleanup read_inode() removing all code checking for lowest_vcn != 0.
1320 This can never happen due to the nature of lookup_attr() and how we
1321 support attribute lists. If it did happen it would imply the inode
1323 - Check for lowest_vcn != 0 in ntfs_read_inode() and mark the inode as
1325 - Update to 2.5.6-pre2 changes in struct address_space.
1326 - Use parent_ino() when accessing d_parent inode number in dir.c.
1327 - Import Sourceforge CVS repository into BitKeeper repository:
1328 http://linux-ntfs.bkbits.net/ntfs-tng-2.5
1329 - Update fs/Makefile, fs/Config.help, fs/Config.in, and
1330 Documentation/filesystems/ntfs.txt for NTFS TNG.
1331 - Create kernel configuration option controlling whether debugging
1333 - Add the required export of end_buffer_io_sync() from the patches
1334 directory to the kernel code.
1335 - Update inode.c::ntfs_show_options() with show_inodes mount option.
1336 - Update errors mount option.
1338 tng-0.0.7 - 13/02/2002 - The driver is now feature complete for read-only!
1340 - Cleanup mft.c and it's debug/error output in particular. Fix a minor
1341 bug in mapping of extent inodes. Update all the comments to fit all
1342 the recent code changes.
1343 - Modify vcn_to_lcn() to cope with entirely unmapped runlists.
1344 - Cleanups in compress.c, mostly comments and folding help.
1345 - Implement attrib.c::map_run_list() as a generic helper.
1346 - Make compress.c::ntfs_file_read_compressed_block() use map_run_list()
1347 thus making code shorter and enabling attribute list support.
1348 - Cleanup incorrect use of [su]64 with %L printf format specifier in
1349 all source files. Type casts to [unsigned] long long added to correct
1350 the mismatches (important for architectures which have long long not
1352 - Merge async io completion handlers for directory indexes and $MFT
1353 data into one by setting the index_block_size{_bits} of the ntfs
1354 inode for $MFT to the mft_record_size{_bits} of the ntfs_volume.
1355 - Cleanup aops.c, update comments.
1356 - Make ntfs_file_get_block() use map_run_list() so all files now
1357 support attribute lists.
1358 - Make ntfs_dir_readpage() almost verbatim copy of
1359 block_read_full_page() by using ntfs_file_get_block() with only real
1360 difference being the use of our own async io completion handler
1361 rather than the default one, thus reducing the amount of code and
1362 automatically enabling attribute list support for directory indices.
1363 - Fix bug in load_attribute_list() - forgot to call brelse in error
1365 - Change parameters to find_attr() and lookup_attr(). We no longer
1366 pass in the upcase table and its length. These can be gotten from
1367 ctx->ntfs_ino->vol->upcase{_len}. Update all callers.
1368 - Cleanups in attrib.c.
1369 - Implement merging of runlists, attrib.c::merge_run_lists() and its
1370 helpers. (Richard Russon)
1371 - Attribute lists part 2, attribute extents and multi part runlists:
1372 enable proper support for LCN_RL_NOT_MAPPED and automatic mapping of
1373 further runlist parts via attrib.c::map_run_list().
1374 - Tiny endianness bug fix in decompress_mapping_pairs().
1376 tng-0.0.6 - Encrypted directories, bug fixes, cleanups, debugging enhancements.
1378 - Enable encrypted directories. (Their index root is marked encrypted
1379 to indicate that new files in that directory should be created
1381 - Fix bug in NInoBmpNonResident() macro. (Cut and paste error.)
1382 - Enable $Extend system directory. Most (if not all) extended system
1383 files do not have unnamed data attributes so ntfs_read_inode() had to
1384 special case them but that is ok, as the special casing recovery
1385 happens inside an error code path so there is zero slow down in the
1386 normal fast path. The special casing is done by introducing a new
1387 function inode.c::ntfs_is_extended_system_file() which checks if any
1388 of the hard links in the inode point to $Extend as being their parent
1389 directory and if they do we assume this is an extended system file.
1390 - Create a sysctl/proc interface to allow {dis,en}abling of debug output
1391 when compiled with -DDEBUG. Default is debug messages to be disabled.
1392 To enable them, one writes a non-zero value to /proc/sys/fs/ntfs-debug
1393 (if /proc is enabled) or uses sysctl(2) to effect the same (if sysctl
1394 interface is enabled). Inspired by old ntfs driver.
1395 - Add debug_msgs insmod/kernel boot parameter to set whether debug
1396 messages are {dis,en}abled. This is useful to enable debug messages
1397 during ntfs initialization and is the only way to activate debugging
1398 when the sysctl interface is not enabled.
1399 - Cleanup debug output in various places.
1400 - Remove all dollar signs ($) from the source (except comments) to
1401 enable compilation on architectures whose gcc compiler does not
1402 support dollar signs in the names of variables/constants. Attribute
1403 types now start with AT_ instead of $ and $I30 is now just I30.
1404 - Cleanup ntfs_lookup() and add consistency check of sequence numbers.
1405 - Load complete runlist for $MFT/$BITMAP during mount and cleanup
1406 access functions. This means we now cope with $MFT/$BITMAP being
1407 spread accross several mft records.
1408 - Disable modification of mft_zone_multiplier on remount. We can always
1409 reenable this later on if we really want to, but we will need to make
1410 sure we readjust the mft_zone size / layout accordingly.
1412 tng-0.0.5 - Modernize for 2.5.x and further in line-ing with Al Viro's comments.
1414 - Use sb_set_blocksize() instead of set_blocksize() and verify the
1416 - Use sb_bread() instead of bread() throughout.
1417 - Add index_vcn_size{_bits} to ntfs_inode structure to store the size
1418 of a directory index block vcn. Apply resulting simplifications in
1420 - Fix a small bug somewhere (but forgot what it was).
1421 - Change ntfs_{debug,error,warning} to enable gcc to do type checking
1422 on the printf-format parameter list and fix bugs reported by gcc
1423 as a result. (Richard Russon)
1424 - Move inode allocation strategy to Al's new stuff but maintain the
1425 divorce of ntfs_inode from struct inode. To achieve this we have two
1426 separate slab caches, one for big ntfs inodes containing a struct
1427 inode and pure ntfs inodes and at the same time fix some faulty
1428 error code paths in ntfs_read_inode().
1429 - Show mount options in proc (inode.c::ntfs_show_options()).
1431 tng-0.0.4 - Big changes, getting in line with Al Viro's comments.
1433 - Modified (un)map_mft_record functions to be common for read and write
1434 case. To specify which is which, added extra parameter at front of
1435 parameter list. Pass either READ or WRITE to this, each has the
1437 - General cleanups to allow for easier folding in vi.
1438 - attrib.c::decompress_mapping_pairs() now accepts the old runlist
1439 argument, and invokes attrib.c::merge_run_lists() to merge the old
1440 and the new runlists.
1441 - Removed attrib.c::find_first_attr().
1442 - Implemented loading of attribute list and complete runlist for $MFT.
1443 This means we now cope with $MFT being spread across several mft
1445 - Adapt to 2.5.2-pre9 and the changed create_empty_buffers() syntax.
1446 - Adapt major/minor/kdev_t/[bk]devname stuff to new 2.5.x kernels.
1447 - Make ntfs_volume be allocated via kmalloc() instead of using a slab
1448 cache. There are too little ntfs_volume structures at any one time
1449 to justify a private slab cache.
1450 - Fix bogus kmap() use in async io completion. Now use kmap_atomic().
1451 Use KM_BIO_IRQ on advice from IRC/kernel...
1452 - Use ntfs_map_page() in map_mft_record() and create ->readpage method
1453 for reading $MFT (ntfs_mft_readpage). In the process create dedicated
1454 address space operations (ntfs_mft_aops) for $MFT inode mapping. Also
1455 removed the now superfluous exports from the kernel core patch.
1456 - Fix a bug where kfree() was used insted of ntfs_free().
1457 - Change map_mft_record() to take ntfs_inode as argument instead of
1458 vfs inode. Dito for unmap_mft_record(). Adapt all callers.
1459 - Add pointer to ntfs_volume to ntfs_inode.
1460 - Add mft record number and sequence number to ntfs_inode. Stop using
1461 i_ino and i_generation for in-driver purposes.
1462 - Implement attrib.c::merge_run_lists(). (Richard Russon)
1463 - Remove use of proper inodes by extent inodes. Move i_ino and
1464 i_generation to ntfs_inode to do this. Apply simplifications that
1465 result and remove iget_no_wait(), etc.
1466 - Pass ntfs_inode everywhere in the driver (used to be struct inode).
1467 - Add reference counting in ntfs_inode for the ntfs inode itself and
1468 for the mapped mft record.
1469 - Extend mft record mapping so we can (un)map extent mft records (new
1470 functions (un)map_extent_mft_record), and so mappings are reference
1471 counted and don't have to happen twice if already mapped - just ref
1473 - Add -o iocharset as alias to -o nls for backwards compatibility.
1474 - The latest core patch is now tiny. In fact just a single additional
1475 export is necessary over the base kernel.
1477 tng-0.0.3 - Cleanups, enhancements, bug fixes.
1479 - Work on attrib.c::decompress_mapping_pairs() to detect base extents
1480 and setup the runlist appropriately using knowledge provided by the
1481 sizes in the base attribute record.
1482 - Balance the get_/put_attr_search_ctx() calls so we don't leak memory
1484 - Introduce ntfs_malloc_nofs() and ntfs_free() to allocate/free a single
1485 page or use vmalloc depending on the amount of memory requested.
1486 - Cleanup error output. The __FUNCTION__ "(): " is now added
1487 automatically. Introduced a new header file debug.h to support this
1488 and also moved ntfs_debug() function into it.
1489 - Make reading of compressed files more intelligent and especially get
1490 rid of the vmalloc_nofs() from readpage(). This now uses per CPU
1491 buffers (allocated at first mount with cluster size <= 4kiB and
1492 deallocated on last umount with cluster size <= 4kiB), and
1493 asynchronous io for the compressed data using a list of buffer heads.
1494 Er, we use synchronous io as async io only works on whole pages
1495 covered by buffers and not on individual buffer heads...
1496 - Bug fix for reading compressed files with sparse compression blocks.
1498 tng-0.0.2 - Now handles larger/fragmented/compressed volumes/files/dirs.
1500 - Fixed handling of directories when cluster size exceeds index block
1502 - Hide DOS only name space directory entries from readdir() but allow
1503 them in lookup(). This should fix the problem that Linux doesn't
1504 support directory hard links, while still allowing access to entries
1505 via their short file name. This also has the benefit of mimicking
1506 what Windows users are used to, so it is the ideal solution.
1507 - Implemented sync_page everywhere so no more hangs in D state when
1509 - Stop using bforget() in favour of brelse().
1510 - Stop locking buffers unnecessarily.
1511 - Implemented compressed files (inode->mapping contains uncompressed
1512 data, raw compressed data is currently bread() into a vmalloc()ed
1514 - Enable compressed directories. (Their index root is marked compressed
1515 to indicate that new files in that directory should be created
1517 - Use vsnprintf rather than vsprintf in the ntfs_error and ntfs_warning
1518 functions. (Thanks to Will Dyson for pointing this out.)
1519 - Moved the ntfs_inode and ntfs_volume (the former ntfs_inode_info and
1520 ntfs_sb_info) out of the common inode and super_block structures and
1521 started using the generic_ip and generic_sbp pointers instead. This
1522 makes ntfs entirely private with respect to the kernel tree.
1523 - Detect compiler version and abort with error message if gcc less than
1525 - Fix bug in name comparison function in unistr.c.
1526 - Implement attribute lists part 1, the infrastructure: search contexts
1527 and operations, find_external_attr(), lookup_attr()) and make the
1528 code use the infrastructure.
1529 - Fix stupid buffer overflow bug that became apparent on larger run
1530 list containing attributes.
1531 - Fix bugs in readdir() that became apparent on larger directories.
1533 The driver is now really useful and survives the test
1534 find . -type f -exec md5sum "{}" \;
1535 without any error messages on a over 1GiB sized partition with >16k
1536 files on it, including compressed files and directories and many files
1537 and directories with attribute lists.
1539 tng-0.0.1 - The first useful version.
1541 - Added ntfs_lookup().
1542 - Added default upcase generation and handling.
1543 - Added compile options to be shown on module init.
1544 - Many bug fixes that were "hidden" before.
1545 - Update to latest kernel.
1546 - Added ntfs_readdir().
1547 - Added file operations for mmap(), read(), open() and llseek(). We just
1548 use the generic ones. The whole point of going through implementing
1549 readpage() methods and where possible get_block() call backs is that
1550 this allows us to make use of the generic high level methods provided
1553 The driver is now actually useful! Yey. (-: It undoubtedly has got bugs
1554 though and it doesn't implement accesssing compressed files yet. Also,
1555 accessing files with attribute list attributes is not implemented yet
1556 either. But for small or simple filesystems it should work and allow
1557 you to list directories, use stat on directory entries and the file
1558 system, open, read, mmap and llseek around in files. A big mile stone
1561 tng-0.0.0 - Initial version tag.
1563 Initial driver implementation. The driver can mount and umount simple
1564 NTFS filesystems (i.e. ones without attribute lists in the system
1565 files). If the mount fails there might be problems in the error handling
1566 code paths, so be warned. Otherwise it seems to be loading the system
1567 files nicely and the mft record read mapping/unmapping seems to be
1568 working nicely, too. Proof of inode metadata in the page cache and non-
1569 resident file unnamed stream data in the page cache concepts is thus
projects / mv-sheeva.git / blob