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