3 - The only places in the kernel where a file is resized are
4 ntfs_file_write*() and ntfs_truncate() for both of which i_sem is
5 held. Just have to be careful in read-/writepage and other helpers
6 not running under i_sem that we play nice... Also need to be careful
7 with initialized_size extension in ntfs_file_write*() and writepage.
8 UPDATE: The only things that need to be checked are the compressed
9 write and the other attribute resize/write cases like index
10 attributes, etc. For now none of these are implemented so are safe.
11 - Implement filling in of holes in aops.c::ntfs_writepage() and its
13 - Implement mft.c::sync_mft_mirror_umount(). We currently will just
14 leave the volume dirty on umount if the final iput(vol->mft_ino)
15 causes a write of any mirrored mft records due to the mft mirror
16 inode having been discarded already. Whether this can actually ever
17 happen is unclear however so it is worth waiting until someone hits
19 - Enable the code for setting the NT4 compatibility flag when we start
20 making NTFS 1.2 specific modifications.
22 2.1.25 - (Almost) fully implement write(2) and truncate(2).
24 - Change ntfs_map_runlist_nolock(), ntfs_attr_find_vcn_nolock() and
25 {__,}ntfs_cluster_free() to also take an optional attribute search
26 context as argument. This allows calling these functions with the
27 mft record mapped. Update all callers.
28 - Fix potential deadlock in ntfs_mft_data_extend_allocation_nolock()
29 error handling by passing in the active search context when calling
31 - Change ntfs_cluster_alloc() to take an extra boolean parameter
32 specifying whether the cluster are being allocated to extend an
33 attribute or to fill a hole.
34 - Change ntfs_attr_make_non_resident() to call ntfs_cluster_alloc()
35 with @is_extension set to TRUE and remove the runlist terminator
36 fixup code as this is now done by ntfs_cluster_alloc().
37 - Change ntfs_attr_make_non_resident to take the attribute value size
38 as an extra parameter. This is needed since we need to know the size
39 before we can map the mft record and our callers always know it. The
40 reason we cannot simply read the size from the vfs inode i_size is
41 that this is not necessarily uptodate. This happens when
42 ntfs_attr_make_non_resident() is called in the ->truncate call path.
43 - Fix ntfs_attr_make_non_resident() to update the vfs inode i_blocks
44 which is zero for a resident attribute but should no longer be zero
45 once the attribute is non-resident as it then has real clusters
47 - Add fs/ntfs/attrib.[hc]::ntfs_attr_extend_allocation(), a function to
48 extend the allocation of an attributes. Optionally, the data size,
49 but not the initialized size can be extended, too.
50 - Implement fs/ntfs/inode.[hc]::ntfs_truncate(). It only supports
51 uncompressed and unencrypted files and it never creates sparse files
52 at least for the moment (making a file sparse requires us to modify
53 its directory entries and we do not support directory operations at
54 the moment). Also, support for highly fragmented files, i.e. ones
55 whose data attribute is split across multiple extents, is severly
56 limited. When such a case is encountered, EOPNOTSUPP is returned.
57 - Enable ATTR_SIZE attribute changes in ntfs_setattr(). This completes
58 the initial implementation of file truncation. Now both open(2)ing
59 a file with the O_TRUNC flag and the {,f}truncate(2) system calls
60 will resize a file appropriately. The limitations are that only
61 uncompressed and unencrypted files are supported. Also, there is
62 only very limited support for highly fragmented files (the ones whose
63 $DATA attribute is split into multiple attribute extents).
64 - In attrib.c::ntfs_attr_set() call balance_dirty_pages_ratelimited()
65 and cond_resched() in the main loop as we could be dirtying a lot of
66 pages and this ensures we play nice with the VM and the system as a
68 - Implement file operations ->write, ->aio_write, ->writev for regular
69 files. This replaces the old use of generic_file_write(), et al and
70 the address space operations ->prepare_write and ->commit_write.
71 This means that both sparse and non-sparse (unencrypted and
72 uncompressed) files can now be extended using the normal write(2)
73 code path. There are two limitations at present and these are that
74 we never create sparse files and that we only have limited support
75 for highly fragmented files, i.e. ones whose data attribute is split
76 across multiple extents. When such a case is encountered,
77 EOPNOTSUPP is returned.
79 2.1.24 - Lots of bug fixes and support more clean journal states.
81 - Support journals ($LogFile) which have been modified by chkdsk. This
82 means users can boot into Windows after we marked the volume dirty.
83 The Windows boot will run chkdsk and then reboot. The user can then
84 immediately boot into Linux rather than having to do a full Windows
85 boot first before rebooting into Linux and we will recognize such a
86 journal and empty it as it is clean by definition. Note, this only
87 works if chkdsk left the journal in an obviously clean state.
88 - Support journals ($LogFile) with only one restart page as well as
89 journals with two different restart pages. We sanity check both and
90 either use the only sane one or the more recent one of the two in the
91 case that both are valid.
92 - Add fs/ntfs/malloc.h::ntfs_malloc_nofs_nofail() which is analogous to
93 ntfs_malloc_nofs() but it performs allocations with __GFP_NOFAIL and
95 - Use ntfs_malloc_nofs_nofail() in the two critical regions in
96 fs/ntfs/runlist.c::ntfs_runlists_merge(). This means we no longer
97 need to panic() if the allocation fails as it now cannot fail.
98 - Fix two nasty runlist merging bugs that had gone unnoticed so far.
99 Thanks to Stefano Picerno for the bug report.
100 - Remove two bogus BUG_ON()s from fs/ntfs/mft.c.
101 - Fix handling of valid but empty mapping pairs array in
102 fs/ntfs/runlist.c::ntfs_mapping_pairs_decompress().
103 - Report unrepresentable inodes during ntfs_readdir() as KERN_WARNING
104 messages and include the inode number. Thanks to Yura Pakhuchiy for
106 - Change ntfs_rl_truncate_nolock() to throw away the runlist if the new
108 - Add runlist.[hc]::ntfs_rl_punch_nolock() which punches a caller
109 specified hole into a runlist.
110 - Fix a bug in fs/ntfs/index.c::ntfs_index_lookup(). When the returned
111 index entry is in the index root, we forgot to set the @ir pointer in
112 the index context. Thanks to Yura Pakhuchiy for finding this bug.
113 - Remove bogus setting of PageError in ntfs_read_compressed_block().
114 - Add fs/ntfs/attrib.[hc]::ntfs_resident_attr_value_resize().
115 - Fix a bug in ntfs_map_runlist_nolock() where we forgot to protect
116 access to the allocated size in the ntfs inode with the size lock.
117 - Fix ntfs_attr_vcn_to_lcn_nolock() and ntfs_attr_find_vcn_nolock() to
118 return LCN_ENOENT when there is no runlist and the allocated size is
120 - Fix load_attribute_list() to handle the case of a NULL runlist.
121 - Fix handling of sparse attributes in ntfs_attr_make_non_resident().
122 - Add BUG() checks to ntfs_attr_make_non_resident() and ntfs_attr_set()
123 to ensure that these functions are never called for compressed or
124 encrypted attributes.
125 - Fix cluster (de)allocators to work when the runlist is NULL and more
126 importantly to take a locked runlist rather than them locking it
127 which leads to lock reversal.
128 - Truncate {a,c,m}time to the ntfs supported time granularity when
129 updating the times in the inode in ntfs_setattr().
130 - Fixup handling of sparse, compressed, and encrypted attributes in
131 fs/ntfs/inode.c::ntfs_read_locked_{,attr_,index_}inode(),
132 fs/ntfs/aops.c::ntfs_{read,write}page().
133 - Make ntfs_write_block() not instantiate sparse blocks if they contain
135 - Optimize fs/ntfs/aops.c::ntfs_write_block() by extending the page
136 lock protection over the buffer submission for i/o which allows the
137 removal of the get_bh()/put_bh() pairs for each buffer.
138 - Fix fs/ntfs/aops.c::ntfs_{read,write}_block() to handle the case
139 where a concurrent truncate has truncated the runlist under our feet.
140 - Fix page_has_buffers()/page_buffers() handling in fs/ntfs/aops.c.
141 - In fs/ntfs/aops.c::ntfs_end_buffer_async_read(), use a bit spin lock
142 in the first buffer head instead of a driver global spin lock to
144 - Minor fix to error handling and error message display in
145 fs/ntfs/aops.c::ntfs_prepare_nonresident_write().
146 - Change the mount options {u,f,d}mask to always parse the number as
147 an octal number to conform to how chmod(1) works, too. Thanks to
148 Giuseppe Bilotta and Horst von Brand for pointing out the errors of
150 - Fix various bugs in the runlist merging code. (Based on libntfs
151 changes by Richard Russon.)
152 - Fix sparse warnings that have crept in over time.
153 - Change ntfs_cluster_free() to require a write locked runlist on entry
154 since we otherwise get into a lock reversal deadlock if a read locked
155 runlist is passed in. In the process also change it to take an ntfs
156 inode instead of a vfs inode as parameter.
157 - Fix the definition of the CHKD ntfs record magic. It had an off by
158 two error causing it to be CHKB instead of CHKD.
159 - Fix a stupid bug in __ntfs_bitmap_set_bits_in_run() which caused the
160 count to become negative and hence we had a wild memset() scribbling
161 all over the system's ram.
163 2.1.23 - Implement extension of resident files and make writing safe as well as
164 many bug fixes, cleanups, and enhancements...
166 - Add printk rate limiting for ntfs_warning() and ntfs_error() when
167 compiled without debug. This avoids a possible denial of service
168 attack. Thanks to Carl-Daniel Hailfinger from SuSE for pointing this
170 - Fix compilation warnings on ia64. (Randy Dunlap)
171 - Use i_size_{read,write}() instead of reading i_size by hand and cache
172 the value where apropriate.
173 - Add size_lock to the ntfs_inode structure. This is an rw spinlock
174 and it locks against access to the inode sizes. Note, ->size_lock
175 is also accessed from irq context so you must use the _irqsave and
176 _irqrestore lock and unlock functions, respectively. Protect all
177 accesses to allocated_size, initialized_size, and compressed_size.
178 - Minor optimization to fs/ntfs/super.c::ntfs_statfs() and its helpers.
179 - Implement extension of resident files in the regular file write code
180 paths (fs/ntfs/aops.c::ntfs_{prepare,commit}_write()). At present
181 this only works until the data attribute becomes too big for the mft
182 record after which we abort the write returning -EOPNOTSUPP from
183 ntfs_prepare_write().
184 - Add disable_sparse mount option together with a per volume sparse
185 enable bit which is set appropriately and a per inode sparse disable
186 bit which is preset on some system file inodes as appropriate.
187 - Enforce that sparse support is disabled on NTFS volumes pre 3.0.
188 - Fix a bug in fs/ntfs/runlist.c::ntfs_mapping_pairs_decompress() in
189 the creation of the unmapped runlist element for the base attribute
191 - Split ntfs_map_runlist() into ntfs_map_runlist() and a non-locking
192 helper ntfs_map_runlist_nolock() which is used by ntfs_map_runlist().
193 This allows us to map runlist fragments with the runlist lock already
194 held without having to drop and reacquire it around the call. Adapt
196 - Change ntfs_find_vcn() to ntfs_find_vcn_nolock() which takes a locked
197 runlist. This allows us to find runlist elements with the runlist
198 lock already held without having to drop and reacquire it around the
199 call. Adapt all callers.
200 - Change time to u64 in time.h::ntfs2utc() as it otherwise generates a
201 warning in the do_div() call on sparc32. Thanks to Meelis Roos for
202 the report and analysis of the warning.
203 - Fix a nasty runlist merge bug when merging two holes.
204 - Set the ntfs_inode->allocated_size to the real allocated size in the
205 mft record for resident attributes (fs/ntfs/inode.c).
206 - Small readability cleanup to use "a" instead of "ctx->attr"
207 everywhere (fs/ntfs/inode.c).
208 - Make fs/ntfs/namei.c::ntfs_get_{parent,dentry} static and move the
209 definition of ntfs_export_ops from fs/ntfs/super.c to namei.c. Also,
210 declare ntfs_export_ops in fs/ntfs/ntfs.h.
211 - Correct sparse file handling. The compressed values need to be
212 checked and set in the ntfs inode as done for compressed files and
213 the compressed size needs to be used for vfs inode->i_blocks instead
214 of the allocated size, again, as done for compressed files.
215 - Add AT_EA in addition to AT_DATA to whitelist for being allowed to be
216 non-resident in fs/ntfs/attrib.c::ntfs_attr_can_be_non_resident().
217 - Add fs/ntfs/attrib.c::ntfs_attr_vcn_to_lcn_nolock() used by the new
219 - Fix bug in fs/ntfs/attrib.c::ntfs_find_vcn_nolock() where after
220 dropping the read lock and taking the write lock we were not checking
221 whether someone else did not already do the work we wanted to do.
222 - Rename fs/ntfs/attrib.c::ntfs_find_vcn_nolock() to
223 ntfs_attr_find_vcn_nolock() and update all callers.
224 - Add fs/ntfs/attrib.[hc]::ntfs_attr_make_non_resident().
225 - Fix sign of various error return values to be negative in
227 - Modify ->readpage and ->writepage (fs/ntfs/aops.c) so they detect and
228 handle the case where an attribute is converted from resident to
229 non-resident by a concurrent file write.
230 - Remove checks for NULL before calling kfree() since kfree() does the
231 checking itself. (Jesper Juhl)
232 - Some utilities modify the boot sector but do not update the checksum.
233 Thus, relax the checking in fs/ntfs/super.c::is_boot_sector_ntfs() to
234 only emit a warning when the checksum is incorrect rather than
235 refusing the mount. Thanks to Bernd Casimir for pointing this
237 - Update attribute definition handling.
238 - Add NTFS_MAX_CLUSTER_SIZE and NTFS_MAX_PAGES_PER_CLUSTER constants.
239 - Use NTFS_MAX_CLUSTER_SIZE in super.c instead of hard coding 0x10000.
240 - Use MAX_BUF_PER_PAGE instead of variable sized array allocation for
241 better code generation and one less sparse warning in fs/ntfs/aops.c.
242 - Remove spurious void pointer casts from fs/ntfs/. (Pekka Enberg)
243 - Use C99 style structure initialization after memory allocation where
244 possible (fs/ntfs/{attrib.c,index.c,super.c}). Thanks to Al Viro and
246 - Stamp the transaction log ($UsnJrnl), aka user space journal, if it
247 is active on the volume and we are mounting read-write or remounting
248 from read-only to read-write.
249 - Fix a bug in address space operations error recovery code paths where
250 if the runlist was not mapped at all and a mapping error occured we
251 would leave the runlist locked on exit to the function so that the
252 next access to the same file would try to take the lock and deadlock.
253 - Detect the case when Windows has been suspended to disk on the volume
254 to be mounted and if this is the case do not allow (re)mounting
255 read-write. This is done by parsing hiberfil.sys if present.
256 - Fix several occurences of a bug where we would perform 'var & ~const'
257 with a 64-bit variable and a int, i.e. 32-bit, constant. This causes
258 the higher order 32-bits of the 64-bit variable to be zeroed. To fix
259 this cast the 'const' to the same 64-bit type as 'var'.
260 - Change the runlist terminator of the newly allocated cluster(s) to
261 LCN_ENOENT in ntfs_attr_make_non_resident(). Otherwise the runlist
263 - Add an extra parameter @last_vcn to ntfs_get_size_for_mapping_pairs()
264 and ntfs_mapping_pairs_build() to allow the runlist encoding to be
265 partial which is desirable when filling holes in sparse attributes.
267 - Change ntfs_map_runlist_nolock() to only decompress the mapping pairs
268 if the requested vcn is inside it. Otherwise we get into problems
269 when we try to map an out of bounds vcn because we then try to map
270 the already mapped runlist fragment which causes
271 ntfs_mapping_pairs_decompress() to fail and return error. Update
272 ntfs_attr_find_vcn_nolock() accordingly.
273 - Fix a nasty deadlock that appeared in recent kernels.
274 The situation: VFS inode X on a mounted ntfs volume is dirty. For
275 same inode X, the ntfs_inode is dirty and thus corresponding on-disk
276 inode, i.e. mft record, which is in a dirty PAGE_CACHE_PAGE belonging
277 to the table of inodes, i.e. $MFT, inode 0.
279 Process 1: sys_sync()/umount()/whatever... calls
280 __sync_single_inode() for $MFT -> do_writepages() -> write_page for
281 the dirty page containing the on-disk inode X, the page is now locked
282 -> ntfs_write_mst_block() which clears PageUptodate() on the page to
283 prevent anyone else getting hold of it whilst it does the write out.
284 This is necessary as the on-disk inode needs "fixups" applied before
285 the write to disk which are removed again after the write and
286 PageUptodate is then set again. It then analyses the page looking
287 for dirty on-disk inodes and when it finds one it calls
288 ntfs_may_write_mft_record() to see if it is safe to write this
289 on-disk inode. This then calls ilookup5() to check if the
290 corresponding VFS inode is in icache(). This in turn calls ifind()
291 which waits on the inode lock via wait_on_inode whilst holding the
293 Process 2: pdflush results in a call to __sync_single_inode for the
294 same VFS inode X on the ntfs volume. This locks the inode (I_LOCK)
295 then calls write-inode -> ntfs_write_inode -> map_mft_record() ->
296 read_cache_page() for the page (in page cache of table of inodes
297 $MFT, inode 0) containing the on-disk inode. This page has
298 PageUptodate() clear because of Process 1 (see above) so
299 read_cache_page() blocks when it tries to take the page lock for the
300 page so it can call ntfs_read_page().
301 Thus Process 1 is holding the page lock on the page containing the
302 on-disk inode X and it is waiting on the inode X to be unlocked in
303 ifind() so it can write the page out and then unlock the page.
304 And Process 2 is holding the inode lock on inode X and is waiting for
305 the page to be unlocked so it can call ntfs_readpage() or discover
306 that Process 1 set PageUptodate() again and use the page.
307 Thus we have a deadlock due to ifind() waiting on the inode lock.
308 The solution: The fix is to use the newly introduced
309 ilookup5_nowait() which does not wait on the inode's lock and hence
310 avoids the deadlock. This is safe as we do not care about the VFS
311 inode and only use the fact that it is in the VFS inode cache and the
312 fact that the vfs and ntfs inodes are one struct in memory to find
313 the ntfs inode in memory if present. Also, the ntfs inode has its
314 own locking so it does not matter if the vfs inode is locked.
315 - Fix bug in mft record writing where we forgot to set the device in
316 the buffers when mapping them after the VM had discarded them.
317 Thanks to Martin MOKREJŠ for the bug report.
319 2.1.22 - Many bug and race fixes and error handling improvements.
321 - Improve error handling in fs/ntfs/inode.c::ntfs_truncate().
322 - Change fs/ntfs/inode.c::ntfs_truncate() to return an error code
323 instead of void and provide a helper ntfs_truncate_vfs() for the
324 vfs ->truncate method.
325 - Add a new ntfs inode flag NInoTruncateFailed() and modify
326 fs/ntfs/inode.c::ntfs_truncate() to set and clear it appropriately.
327 - Fix min_size and max_size definitions in ATTR_DEF structure in
328 fs/ntfs/layout.h to be signed.
329 - Add attribute definition handling helpers to fs/ntfs/attrib.[hc]:
330 ntfs_attr_size_bounds_check(), ntfs_attr_can_be_non_resident(), and
331 ntfs_attr_can_be_resident(), which in turn use the new private helper
332 ntfs_attr_find_in_attrdef().
333 - In fs/ntfs/aops.c::mark_ntfs_record_dirty(), take the
334 mapping->private_lock around the dirtying of the buffer heads
335 analagous to the way it is done in __set_page_dirty_buffers().
336 - Ensure the mft record size does not exceed the PAGE_CACHE_SIZE at
337 mount time as this cannot work with the current implementation.
338 - Check for location of attribute name and improve error handling in
339 general in fs/ntfs/inode.c::ntfs_read_locked_inode() and friends.
340 - In fs/ntfs/aops.c::ntfs_writepage(), if the page is fully outside
341 i_size, i.e. race with truncate, invalidate the buffers on the page
342 so that they become freeable and hence the page does not leak.
343 - Remove unused function fs/ntfs/runlist.c::ntfs_rl_merge(). (Adrian
345 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_find() that resulted in
346 a NULL pointer dereference in the error code path when a corrupt
347 attribute was found. (Thanks to Domen Puncer for the bug report.)
348 - Add MODULE_VERSION() to fs/ntfs/super.c.
349 - Make several functions and variables static. (Adrian Bunk)
350 - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() so it allocates
351 buffers for the page if they are not present and then marks the
352 buffers belonging to the ntfs record dirty. This causes the buffers
353 to become busy and hence they are safe from removal until the page
354 has been written out.
355 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find() in the
356 error handling code path that resulted in a BUG() due to trying to
357 unmap an extent mft record when the mapping of it had failed and it
358 thus was not mapped. (Thanks to Ken MacFerrin for the bug report.)
359 - Drop the runlist lock after the vcn has been read in
360 fs/ntfs/lcnalloc.c::__ntfs_cluster_free().
361 - Rewrite handling of multi sector transfer errors. We now do not set
362 PageError() when such errors are detected in the async i/o handler
363 fs/ntfs/aops.c::ntfs_end_buffer_async_read(). All users of mst
364 protected attributes now check the magic of each ntfs record as they
365 use it and act appropriately. This has the effect of making errors
366 granular per ntfs record rather than per page which solves the case
367 where we cannot access any of the ntfs records in a page when a
368 single one of them had an mst error. (Thanks to Ken MacFerrin for
370 - Fix error handling in fs/ntfs/quota.c::ntfs_mark_quotas_out_of_date()
371 where we failed to release i_sem on the $Quota/$Q attribute inode.
372 - Fix bug in handling of bad inodes in fs/ntfs/namei.c::ntfs_lookup().
373 - Add mapping of unmapped buffers to all remaining code paths, i.e.
374 fs/ntfs/aops.c::ntfs_write_mst_block(), mft.c::ntfs_sync_mft_mirror(),
375 and write_mft_record_nolock(). From now on we require that the
376 complete runlist for the mft mirror is always mapped into memory.
377 - Add creation of buffers to fs/ntfs/mft.c::ntfs_sync_mft_mirror().
378 - Improve error handling in fs/ntfs/aops.c::ntfs_{read,write}_block().
379 - Cleanup fs/ntfs/aops.c::ntfs_{read,write}page() since we know that a
380 resident attribute will be smaller than a page which makes the code
381 simpler. Also make the code more tolerant to concurrent ->truncate.
383 2.1.21 - Fix some races and bugs, rewrite mft write code, add mft allocator.
385 - Implement extent mft record deallocation
386 fs/ntfs/mft.c::ntfs_extent_mft_record_free().
387 - Splitt runlist related functions off from attrib.[hc] to runlist.[hc].
388 - Add vol->mft_data_pos and initialize it at mount time.
389 - Rename init_runlist() to ntfs_init_runlist(), ntfs_vcn_to_lcn() to
390 ntfs_rl_vcn_to_lcn(), decompress_mapping_pairs() to
391 ntfs_mapping_pairs_decompress(), ntfs_merge_runlists() to
392 ntfs_runlists_merge() and adapt all callers.
393 - Add fs/ntfs/runlist.[hc]::ntfs_get_nr_significant_bytes(),
394 ntfs_get_size_for_mapping_pairs(), ntfs_write_significant_bytes(),
395 and ntfs_mapping_pairs_build(), adapted from libntfs.
396 - Make fs/ntfs/lcnalloc.c::ntfs_cluster_free_from_rl_nolock() not
397 static and add a declaration for it to lcnalloc.h.
398 - Add fs/ntfs/lcnalloc.h::ntfs_cluster_free_from_rl() which is a static
399 inline wrapper for ntfs_cluster_free_from_rl_nolock() which takes the
400 cluster bitmap lock for the duration of the call.
401 - Add fs/ntfs/attrib.[hc]::ntfs_attr_record_resize().
402 - Implement the equivalent of memset() for an ntfs attribute in
403 fs/ntfs/attrib.[hc]::ntfs_attr_set() and switch
404 fs/ntfs/logfile.c::ntfs_empty_logfile() to using it.
405 - Remove unnecessary casts from LCN_* constants.
406 - Implement fs/ntfs/runlist.c::ntfs_rl_truncate_nolock().
407 - Add MFT_RECORD_OLD as a copy of MFT_RECORD in fs/ntfs/layout.h and
408 change MFT_RECORD to contain the NTFS 3.1+ specific fields.
409 - Add a helper function fs/ntfs/aops.c::mark_ntfs_record_dirty() which
410 marks all buffers belonging to an ntfs record dirty, followed by
411 marking the page the ntfs record is in dirty and also marking the vfs
412 inode containing the ntfs record dirty (I_DIRTY_PAGES).
413 - Switch fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to using the
414 new helper fs/ntfs/aops.c::mark_ntfs_record_dirty() and remove the no
415 longer needed fs/ntfs/index.[hc]::__ntfs_index_entry_mark_dirty().
416 - Move ntfs_{un,}map_page() from ntfs.h to aops.h and fix resulting
418 - Move the typedefs for runlist_element and runlist from types.h to
419 runlist.h and fix resulting include errors.
420 - Remove unused {__,}format_mft_record() from fs/ntfs/mft.c.
421 - Modify fs/ntfs/mft.c::__mark_mft_record_dirty() to use the helper
422 mark_ntfs_record_dirty() which also changes the behaviour in that we
423 now set the buffers belonging to the mft record dirty as well as the
425 - Update fs/ntfs/mft.c::write_mft_record_nolock() and sync_mft_mirror()
426 to cope with the fact that there now are dirty buffers in mft pages.
427 - Update fs/ntfs/inode.c::ntfs_write_inode() to also use the helper
428 mark_ntfs_record_dirty() and thus to set the buffers belonging to the
429 mft record dirty as well as the page itself.
430 - Fix compiler warnings on x86-64 in fs/ntfs/dir.c. (Randy Dunlap,
431 slightly modified by me)
432 - Add fs/ntfs/mft.c::try_map_mft_record() which fails with -EALREADY if
433 the mft record is already locked and otherwise behaves the same way
434 as fs/ntfs/mft.c::map_mft_record().
435 - Modify fs/ntfs/mft.c::write_mft_record_nolock() so that it only
436 writes the mft record if the buffers belonging to it are dirty.
437 Otherwise we assume that it was written out by other means already.
438 - Attempting to write outside initialized size is _not_ a bug so remove
439 the bug check from fs/ntfs/aops.c::ntfs_write_mst_block(). It is in
440 fact required to write outside initialized size when preparing to
441 extend the initialized size.
442 - Map the page instead of using page_address() before writing to it in
443 fs/ntfs/aops.c::ntfs_mft_writepage().
444 - Provide exclusion between opening an inode / mapping an mft record
445 and accessing the mft record in fs/ntfs/mft.c::ntfs_mft_writepage()
446 by setting the page not uptodate throughout ntfs_mft_writepage().
447 - Clear the page uptodate flag in fs/ntfs/aops.c::ntfs_write_mst_block()
448 to ensure noone can see the page whilst the mst fixups are applied.
449 - Add the helper fs/ntfs/mft.c::ntfs_may_write_mft_record() which
450 checks if an mft record may be written out safely obtaining any
451 necessary locks in the process. This is used by
452 fs/ntfs/aops.c::ntfs_write_mst_block().
453 - Modify fs/ntfs/aops.c::ntfs_write_mst_block() to also work for
454 writing mft records and improve its error handling in the process.
455 Now if any of the records in the page fail to be written out, all
456 other records will be written out instead of aborting completely.
457 - Remove ntfs_mft_aops and update all users to use ntfs_mst_aops.
458 - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to set the
459 ntfs_mst_aops for all inodes which are NInoMstProtected() and
460 ntfs_aops for all other inodes.
461 - Rename fs/ntfs/mft.c::sync_mft_mirror{,_umount}() to
462 ntfs_sync_mft_mirror{,_umount}() and change their parameters so they
463 no longer require an ntfs inode to be present. Update all callers.
464 - Cleanup the error handling in fs/ntfs/mft.c::ntfs_sync_mft_mirror().
465 - Clear the page uptodate flag in fs/ntfs/mft.c::ntfs_sync_mft_mirror()
466 to ensure noone can see the page whilst the mst fixups are applied.
467 - Remove the no longer needed fs/ntfs/mft.c::ntfs_mft_writepage() and
468 fs/ntfs/mft.c::try_map_mft_record().
469 - Fix callers of fs/ntfs/aops.c::mark_ntfs_record_dirty() to call it
470 with the ntfs inode which contains the page rather than the ntfs
471 inode the mft record of which is in the page.
472 - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by moving the
473 index inode bitmap inode release code from there to
474 fs/ntfs/inode.c::ntfs_clear_big_inode(). (Thanks to Christoph
475 Hellwig for spotting this.)
476 - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by taking the
477 inode semaphore around the code that sets ni->itype.index.bmp_ino to
478 NULL and reorganize the code to optimize it a bit. (Thanks to
479 Christoph Hellwig for spotting this.)
480 - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() to no longer take the
481 ntfs inode as a parameter as this is confusing and misleading and the
482 needed ntfs inode is available via NTFS_I(page->mapping->host).
483 Adapt all callers to this change.
484 - Modify fs/ntfs/mft.c::write_mft_record_nolock() and
485 fs/ntfs/aops.c::ntfs_write_mst_block() to only check the dirty state
486 of the first buffer in a record and to take this as the ntfs record
487 dirty state. We cannot look at the dirty state for subsequent
488 buffers because we might be racing with
489 fs/ntfs/aops.c::mark_ntfs_record_dirty().
490 - Move the static inline ntfs_init_big_inode() from fs/ntfs/inode.c to
491 inode.h and make fs/ntfs/inode.c::__ntfs_init_inode() non-static and
492 add a declaration for it to inode.h. Fix some compilation issues
493 that resulted due to #includes and header file interdependencies.
494 - Simplify setup of i_mode in fs/ntfs/inode.c::ntfs_read_locked_inode().
495 - Add helpers fs/ntfs/layout.h::MK_MREF() and MK_LE_MREF().
496 - Modify fs/ntfs/mft.c::map_extent_mft_record() to only verify the mft
497 record sequence number if it is specified (i.e. not zero).
498 - Add fs/ntfs/mft.[hc]::ntfs_mft_record_alloc() and various helper
499 functions used by it.
500 - Update Documentation/filesystems/ntfs.txt with instructions on how to
501 use the Device-Mapper driver with NTFS ftdisk/LDM raid. This removes
502 the linear raid problem with the Software RAID / MD driver when one
503 or more of the devices has an odd number of sectors.
505 2.1.20 - Fix two stupid bugs introduced in 2.1.18 release.
507 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_reinit_search_ctx()
508 where we did not clear ctx->al_entry but it was still set due to
509 changes in ntfs_attr_lookup() and ntfs_external_attr_find() in
511 - Fix another stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find()
512 where we forgot to unmap the extent mft record when we had finished
513 enumerating an attribute which caused a bug check to trigger when the
514 VFS calls ->clear_inode.
516 2.1.19 - Many cleanups, improvements, and a minor bug fix.
518 - Update ->setattr (fs/ntfs/inode.c::ntfs_setattr()) to refuse to
519 change the uid, gid, and mode of an inode as we do not support NTFS
521 - Remove BKL use from ntfs_setattr() syncing up with the rest of the
523 - Get rid of the ugly transparent union in fs/ntfs/dir.c::ntfs_readdir()
524 and ntfs_filldir() as per suggestion from Al Viro.
525 - Change '\0' and L'\0' to simply 0 as per advice from Linus Torvalds.
526 - Update ->truncate (fs/ntfs/inode.c::ntfs_truncate()) to check if the
527 inode size has changed and to only output an error if so.
528 - Rename fs/ntfs/attrib.h::attribute_value_length() to ntfs_attr_size().
529 - Add le{16,32,64} as well as sle{16,32,64} data types to
531 - Change ntfschar to be le16 instead of u16 in fs/ntfs/types.h.
532 - Add le versions of VCN, LCN, and LSN called leVCN, leLCN, and leLSN,
533 respectively, to fs/ntfs/types.h.
534 - Update endianness conversion macros in fs/ntfs/endian.h to use the
535 new types as appropriate.
536 - Do proper type casting when using sle64_to_cpup() in fs/ntfs/dir.c
538 - Add leMFT_REF data type to fs/ntfs/layout.h.
539 - Update all NTFS header files with the new little endian data types.
540 Affected files are fs/ntfs/layout.h, logfile.h, and time.h.
541 - Do proper type casting when using ntfs_is_*_recordp() in
542 fs/ntfs/logfile.c, mft.c, and super.c.
543 - Fix all the sparse bitwise warnings. Had to change all the typedef
544 enums storing little endian values to simple enums plus a typedef for
545 the datatype to make sparse happy.
546 - Fix a bug found by the new sparse bitwise warnings where the default
547 upcase table was defined as a pointer to wchar_t rather than ntfschar
548 in fs/ntfs/ntfs.h and super.c.
549 - Change {const_,}cpu_to_le{16,32}(0) to just 0 as suggested by Al Viro.
551 2.1.18 - Fix scheduling latencies at mount time as well as an endianness bug.
553 - Remove vol->nr_mft_records as it was pretty meaningless and optimize
554 the calculation of total/free inodes as used by statfs().
555 - Fix scheduling latencies in ntfs_fill_super() by dropping the BKL
556 because the code itself is using the ntfs_lock semaphore which
557 provides safe locking. (Ingo Molnar)
558 - Fix a potential bug in fs/ntfs/mft.c::map_extent_mft_record() that
559 could occur in the future for when we start closing/freeing extent
560 inodes if we don't set base_ni->ext.extent_ntfs_inos to NULL after
562 - Rename {find,lookup}_attr() to ntfs_attr_{find,lookup}() as well as
563 find_external_attr() to ntfs_external_attr_find() to cleanup the
564 namespace a bit and to be more consistent with libntfs.
565 - Rename {{re,}init,get,put}_attr_search_ctx() to
566 ntfs_attr_{{re,}init,get,put}_search_ctx() as well as the type
567 attr_search_context to ntfs_attr_search_ctx.
568 - Force use of ntfs_attr_find() in ntfs_attr_lookup() when searching
569 for the attribute list attribute itself.
570 - Fix endianness bug in ntfs_external_attr_find().
571 - Change ntfs_{external_,}attr_find() to return 0 on success, -ENOENT
572 if the attribute is not found, and -EIO on real error. In the case
573 of -ENOENT, the search context is updated to describe the attribute
574 before which the attribute being searched for would need to be
575 inserted if such an action were to be desired and in the case of
576 ntfs_external_attr_find() the search context is also updated to
577 indicate the attribute list entry before which the attribute list
578 entry of the attribute being searched for would need to be inserted
579 if such an action were to be desired. Also make ntfs_find_attr()
580 static and remove its prototype from attrib.h as it is not used
581 anywhere other than attrib.c. Update ntfs_attr_lookup() and all
582 callers of ntfs_{external,}attr_{find,lookup}() for the new return
584 - Minor cleanup of fs/ntfs/inode.c::ntfs_init_locked_inode().
586 2.1.17 - Fix bugs in mount time error code paths and other updates.
588 - Implement bitmap modification code (fs/ntfs/bitmap.[hc]). This
589 includes functions to set/clear a single bit or a run of bits.
590 - Add fs/ntfs/attrib.[hc]::ntfs_find_vcn() which returns the locked
591 runlist element containing a particular vcn. It also takes care of
592 mapping any needed runlist fragments.
593 - Implement cluster (de-)allocation code (fs/ntfs/lcnalloc.[hc]).
594 - Load attribute definition table from $AttrDef at mount time.
595 - Fix bugs in mount time error code paths involving (de)allocation of
596 the default and volume upcase tables.
597 - Remove ntfs_nr_mounts as it is no longer used.
599 2.1.16 - Implement access time updates, file sync, async io, and read/writev.
601 - Add support for readv/writev and aio_read/aio_write (fs/ntfs/file.c).
602 This is done by setting the appropriate file operations pointers to
603 the generic helper functions provided by mm/filemap.c.
604 - Implement fsync, fdatasync, and msync both for files (fs/ntfs/file.c)
605 and directories (fs/ntfs/dir.c).
606 - Add support for {a,m,c}time updates to inode.c::ntfs_write_inode().
607 Note, except for the root directory and any other system files opened
608 by the user, the system files will not have their access times
609 updated as they are only accessed at the inode level an hence the
610 file level functions which cause the times to be updated are never
613 2.1.15 - Invalidate quotas when (re)mounting read-write.
615 - Add new element itype.index.collation_rule to the ntfs inode
616 structure and set it appropriately in ntfs_read_locked_inode().
617 - Implement a new inode type "index" to allow efficient access to the
618 indices found in various system files and adapt inode handling
619 accordingly (fs/ntfs/inode.[hc]). An index inode is essentially an
620 attribute inode (NInoAttr() is true) with an attribute type of
621 AT_INDEX_ALLOCATION. As such, it is no longer allowed to call
622 ntfs_attr_iget() with an attribute type of AT_INDEX_ALLOCATION as
623 there would be no way to distinguish between normal attribute inodes
624 and index inodes. The function to obtain an index inode is
625 ntfs_index_iget() and it uses the helper function
626 ntfs_read_locked_index_inode(). Note, we do not overload
627 ntfs_attr_iget() as indices consist of multiple attributes so using
628 ntfs_attr_iget() to obtain an index inode would be confusing.
629 - Ensure that there is no overflow when doing page->index <<
630 PAGE_CACHE_SHIFT by casting page->index to s64 in fs/ntfs/aops.c.
631 - Use atomic kmap instead of kmap() in fs/ntfs/aops.c::ntfs_read_page()
632 and ntfs_read_block().
633 - Use case sensitive attribute lookups instead of case insensitive ones.
634 - Lock all page cache pages belonging to mst protected attributes while
635 accessing them to ensure we never see corrupt data while the page is
637 - Add framework for generic ntfs collation (fs/ntfs/collation.[hc]).
638 We have ntfs_is_collation_rule_supported() to check if the collation
639 rule you want to use is supported and ntfs_collation() which actually
640 collates two data items. We currently only support COLLATION_BINARY
641 and COLLATION_NTOFS_ULONG but support for other collation rules will
642 be added as the need arises.
643 - Add a new type, ntfs_index_context, to allow retrieval of an index
644 entry using the corresponding index key. To get an index context,
645 use ntfs_index_ctx_get() and to release it, use ntfs_index_ctx_put().
646 This also adds a new slab cache for the index contexts. To lookup a
647 key in an index inode, use ntfs_index_lookup(). After modifying an
648 index entry, call ntfs_index_entry_flush_dcache_page() followed by
649 ntfs_index_entry_mark_dirty() to ensure the changes are written out
650 to disk. For details see fs/ntfs/index.[hc]. Note, at present, if
651 an index entry is in the index allocation attribute rather than the
652 index root attribute it will not be written out (you will get a
653 warning message about discarded changes instead).
654 - Load the quota file ($Quota) and check if quota tracking is enabled
655 and if so, mark the quotas out of date. This causes windows to
656 rescan the volume on boot and update all quota entries.
657 - Add a set_page_dirty address space operation for ntfs_m[fs]t_aops.
658 It is simply set to __set_page_dirty_nobuffers() to make sure that
659 running set_page_dirty() on a page containing mft/ntfs records will
660 not affect the dirty state of the page buffers.
661 - Add fs/ntfs/index.c::__ntfs_index_entry_mark_dirty() which sets all
662 buffers that are inside the ntfs record in the page dirty after which
663 it sets the page dirty. This allows ->writepage to only write the
664 dirty index records rather than having to write all the records in
665 the page. Modify fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to
666 use this rather than __set_page_dirty_nobuffers().
667 - Implement fs/ntfs/aops.c::ntfs_write_mst_block() which enables the
668 writing of page cache pages belonging to mst protected attributes
669 like the index allocation attribute in directory indices and other
670 indices like $Quota/$Q, etc. This means that the quota is now marked
671 out of date on all volumes rather than only on ones where the quota
672 defaults entry is in the index root attribute of the $Quota/$Q index.
674 2.1.14 - Fix an NFSd caused deadlock reported by several users.
676 - Modify fs/ntfs/ntfs_readdir() to copy the index root attribute value
677 to a buffer so that we can put the search context and unmap the mft
678 record before calling the filldir() callback. We need to do this
679 because of NFSd which calls ->lookup() from its filldir callback()
680 and this causes NTFS to deadlock as ntfs_lookup() maps the mft record
681 of the directory and since ntfs_readdir() has got it mapped already
682 ntfs_lookup() deadlocks.
684 2.1.13 - Enable overwriting of resident files and housekeeping of system files.
686 - Implement writing of mft records (fs/ntfs/mft.[hc]), which includes
687 keeping the mft mirror in sync with the mft when mirrored mft records
688 are written. The functions are write_mft_record{,_nolock}(). The
689 implementation is quite rudimentary for now with lots of things not
690 implemented yet but I am not sure any of them can actually occur so
691 I will wait for people to hit each one and only then implement it.
692 - Commit open system inodes at umount time. This should make it
693 virtually impossible for sync_mft_mirror_umount() to ever be needed.
694 - Implement ->write_inode (fs/ntfs/inode.c::ntfs_write_inode()) for the
695 ntfs super operations. This gives us inode writing via the VFS inode
696 dirty code paths. Note: Access time updates are not implemented yet.
697 - Implement fs/ntfs/mft.[hc]::{,__}mark_mft_record_dirty() and make
698 fs/ntfs/aops.c::ntfs_writepage() and ntfs_commit_write() use it, thus
699 finally enabling resident file overwrite! (-8 This also includes a
700 placeholder for ->writepage (ntfs_mft_writepage()), which for now
701 just redirties the page and returns. Also, at umount time, we for
702 now throw away all mft data page cache pages after the last call to
703 ntfs_commit_inode() in the hope that all inodes will have been
704 written out by then and hence no dirty (meta)data will be lost. We
705 also check for this case and emit an error message telling the user
707 - Use set_page_writeback() and end_page_writeback() in the resident
708 attribute code path of fs/ntfs/aops.c::ntfs_writepage() otherwise
709 the radix-tree tag PAGECACHE_TAG_DIRTY remains set even though the
711 - Implement ntfs_mft_writepage() so it now checks if any of the mft
712 records in the page are dirty and if so redirties the page and
713 returns. Otherwise it just returns (after doing set_page_writeback(),
714 unlock_page(), end_page_writeback() or the radix-tree tag
715 PAGECACHE_TAG_DIRTY remains set even though the page is clean), thus
716 alowing the VM to do with the page as it pleases. Also, at umount
717 time, now only throw away dirty mft (meta)data pages if dirty inodes
718 are present and ask the user to email us if they see this happening.
719 - Add functions ntfs_{clear,set}_volume_flags(), to modify the volume
720 information flags (fs/ntfs/super.c).
721 - Mark the volume dirty when (re)mounting read-write and mark it clean
722 when unmounting or remounting read-only. If any volume errors are
723 found, the volume is left marked dirty to force chkdsk to run.
724 - Add code to set the NT4 compatibility flag when (re)mounting
725 read-write for newer NTFS versions but leave it commented out for now
726 since we do not make any modifications that are NTFS 1.2 specific yet
727 and since setting this flag breaks Captive-NTFS which is not nice.
728 This code must be enabled once we start writing NTFS 1.2 specific
729 changes otherwise Windows NTFS driver might crash / cause corruption.
731 2.1.12 - Fix the second fix to the decompression engine and some cleanups.
733 - Add a new address space operations struct, ntfs_mst_aops, for mst
734 protected attributes. This is because the default ntfs_aops do not
735 make sense with mst protected data and were they to write anything to
736 such an attribute they would cause data corruption so we provide
737 ntfs_mst_aops which does not have any write related operations set.
738 - Cleanup dirty ntfs inode handling (fs/ntfs/inode.[hc]) which also
739 includes an adapted ntfs_commit_inode() and an implementation of
740 ntfs_write_inode() which for now just cleans dirty inodes without
741 writing them (it does emit a warning that this is happening).
742 - Undo the second decompression engine fix (see 2.1.9 release ChangeLog
743 entry) as it was only fixing a theoretical bug but at the same time
744 it badly broke the handling of sparse and uncompressed compression
747 2.1.11 - Driver internal cleanups.
749 - Only build logfile.o if building the driver with read-write support.
750 - Really final white space cleanups.
751 - Use generic_ffs() instead of ffs() in logfile.c which allows the
752 log_page_size variable to be optimized by gcc into a constant.
753 - Rename uchar_t to ntfschar everywhere as uchar_t is unsigned 1-byte
754 char as defined by POSIX and as found on some systems.
756 2.1.10 - Force read-only (re)mounting of volumes with unsupported volume flags.
758 - Finish off the white space cleanups (remove trailing spaces, etc).
759 - Clean up ntfs_fill_super() and ntfs_read_inode_mount() by removing
760 the kludges around the first iget(). Instead of (re)setting ->s_op
761 we have the $MFT inode set up by explicit new_inode() / set ->i_ino /
762 insert_inode_hash() / call ntfs_read_inode_mount() directly. This
763 kills the need for second super_operations and allows to return error
764 from ntfs_read_inode_mount() without resorting to ugly "poisoning"
766 - Force read-only (re)mounting if any of the following bits are set in
767 the volume information flags:
768 VOLUME_IS_DIRTY, VOLUME_RESIZE_LOG_FILE,
769 VOLUME_UPGRADE_ON_MOUNT, VOLUME_DELETE_USN_UNDERWAY,
770 VOLUME_REPAIR_OBJECT_ID, VOLUME_MODIFIED_BY_CHKDSK
771 To make this easier we define VOLUME_MUST_MOUNT_RO_MASK with all the
772 above bits set so the test is made easy.
774 2.1.9 - Fix two bugs in decompression engine.
776 - Fix a bug where we would not always detect that we have reached the
777 end of a compression block because we were ending at minus one byte
778 which is effectively the same as being at the end. The fix is to
779 check whether the uncompressed buffer has been fully filled and if so
780 we assume we have reached the end of the compression block. A big
781 thank you to Marcin Gibuła for the bug report, the assistance in
782 tracking down the bug and testing the fix.
783 - Fix a possible bug where when a compressed read is truncated to the
784 end of the file, the offset inside the last page was not truncated.
786 2.1.8 - Handle $MFT mirror and $LogFile, improve time handling, and cleanups.
788 - Use get_bh() instead of manual atomic_inc() in fs/ntfs/compress.c.
789 - Modify fs/ntfs/time.c::ntfs2utc(), get_current_ntfs_time(), and
790 utc2ntfs() to work with struct timespec instead of time_t on the
791 Linux UTC time side thus preserving the full precision of the NTFS
792 time and only loosing up to 99 nano-seconds in the Linux UTC time.
793 - Move fs/ntfs/time.c to fs/ntfs/time.h and make the time functions
795 - Remove unused ntfs_dirty_inode().
796 - Cleanup super operations declaration in fs/ntfs/super.c.
797 - Wrap flush_dcache_mft_record_page() in #ifdef NTFS_RW.
798 - Add NInoTestSetFoo() and NInoTestClearFoo() macro magic to
799 fs/ntfs/inode.h and use it to declare NInoTest{Set,Clear}Dirty.
800 - Move typedefs for ntfs_attr and test_t from fs/ntfs/inode.c to
801 fs/ntfs/inode.h so they can be used elsewhere.
802 - Determine the mft mirror size as the number of mirrored mft records
803 and store it in ntfs_volume->mftmirr_size (fs/ntfs/super.c).
804 - Load the mft mirror at mount time and compare the mft records stored
805 in it to the ones in the mft. Force a read-only mount if the two do
806 not match (fs/ntfs/super.c).
807 - Fix type casting related warnings on 64-bit architectures. Thanks
808 to Meelis Roos for reporting them.
809 - Move %L to %ll as %L is floating point and %ll is integer which is
811 - Read the journal ($LogFile) and determine if the volume has been
812 shutdown cleanly and force a read-only mount if not (fs/ntfs/super.c
813 and fs/ntfs/logfile.c). This is a little bit of a crude check in
814 that we only look at the restart areas and not at the actual log
815 records so that there will be a very small number of cases where we
816 think that a volume is dirty when in fact it is clean. This should
817 only affect volumes that have not been shutdown cleanly and did not
818 have any pending, non-check-pointed i/o.
819 - If the $LogFile indicates a clean shutdown and a read-write (re)mount
820 is requested, empty $LogFile by overwriting it with 0xff bytes to
821 ensure that Windows cannot cause data corruption by replaying a stale
822 journal after Linux has written to the volume.
824 2.1.7 - Enable NFS exporting of mounted NTFS volumes.
826 - Set i_generation in the VFS inode from the seq_no of the NTFS inode.
827 - Make ntfs_lookup() NFS export safe, i.e. use d_splice_alias(), etc.
828 - Implement ->get_dentry() in fs/ntfs/namei.c::ntfs_get_dentry() as the
829 default doesn't allow inode number 0 which is a valid inode on NTFS
830 and even if it did allow that it uses iget() instead of ntfs_iget()
831 which makes it useless for us.
832 - Implement ->get_parent() in fs/ntfs/namei.c::ntfs_get_parent() as the
833 default just returns -EACCES which is not very useful.
834 - Define export operations (->s_export_op) for NTFS (ntfs_export_ops)
835 and set them up in the super block at mount time (super.c) this
836 allows mounted NTFS volumes to be exported via NFS.
837 - Add missing return -EOPNOTSUPP; in
838 fs/ntfs/aops.c::ntfs_commit_nonresident_write().
839 - Enforce no atime and no dir atime updates at mount/remount time as
840 they are not implemented yet anyway.
841 - Move a few assignments in fs/ntfs/attrib.c::load_attribute_list() to
842 after a NULL check. Thanks to Dave Jones for pointing this out.
844 2.1.6 - Fix minor bug in handling of compressed directories.
846 - Fix bug in handling of compressed directories. A compressed
847 directory is not really compressed so when we set the ->i_blocks
848 field of a compressed directory inode we were setting it from the
849 non-existing field ni->itype.compressed.size which gave random
850 results... For directories we now always use ni->allocated_size.
852 2.1.5 - Fix minor bug in attribute list attribute handling.
854 - Fix bug in attribute list handling. Actually it is not as much a bug
855 as too much protection in that we were not allowing attribute lists
856 which waste space on disk while Windows XP clearly allows it and in
857 fact creates such attribute lists so our driver was failing.
858 - Update NTFS documentation ready for 2.6 kernel release.
860 2.1.4 - Reduce compiler requirements.
862 - Remove all uses of unnamed structs and unions in the driver to make
863 old and newer gcc versions happy. Makes it a bit uglier IMO but at
864 least people will stop hassling me about it.
866 2.1.3 - Important bug fixes in corner cases.
868 - super.c::parse_ntfs_boot_sector(): Correct the check for 64-bit
869 clusters. (Philipp Thomas)
870 - attrib.c::load_attribute_list(): Fix bug when initialized_size is a
871 multiple of the block_size but not the cluster size. (Szabolcs
872 Szakacsits <szaka@sienet.hu>)
874 2.1.2 - Important bug fixes aleviating the hangs in statfs.
876 - Fix buggy free cluster and free inode determination logic.
878 2.1.1 - Minor updates.
880 - Add handling for initialized_size != data_size in compressed files.
881 - Reduce function local stack usage from 0x3d4 bytes to just noise in
882 fs/ntfs/upcase.c. (Randy Dunlap <rddunlap@osdl.ord>)
883 - Remove compiler warnings for newer gcc.
884 - Pages are no longer kmapped by mm/filemap.c::generic_file_write()
885 around calls to ->{prepare,commit}_write. Adapt NTFS appropriately
886 in fs/ntfs/aops.c::ntfs_prepare_nonresident_write() by using
887 kmap_atomic(KM_USER0).
889 2.1.0 - First steps towards write support: implement file overwrite.
891 - Add configuration option for developmental write support with an
892 appropriately scary configuration help text.
893 - Initial implementation of fs/ntfs/aops.c::ntfs_writepage() and its
894 helper fs/ntfs/aops.c::ntfs_write_block(). This enables mmap(2) based
895 overwriting of existing files on ntfs. Note: Resident files are
896 only written into memory, and not written out to disk at present, so
897 avoid writing to files smaller than about 1kiB.
898 - Initial implementation of fs/ntfs/aops.c::ntfs_prepare_write(), its
899 helper fs/ntfs/aops.c::ntfs_prepare_nonresident_write() and their
900 counterparts, fs/ntfs/aops.c::ntfs_commit_write(), and
901 fs/ntfs/aops.c::ntfs_commit_nonresident_write(), respectively. Also,
902 add generic_file_write() to the ntfs file operations (fs/ntfs/file.c).
903 This enables write(2) based overwriting of existing files on ntfs.
904 Note: As with mmap(2) based overwriting, resident files are only
905 written into memory, and not written out to disk at present, so avoid
906 writing to files smaller than about 1kiB.
907 - Implement ->truncate (fs/ntfs/inode.c::ntfs_truncate()) and
908 ->setattr() (fs/ntfs/inode.c::ntfs_setattr()) inode operations for
909 files with the purpose of intercepting and aborting all i_size
910 changes which we do not support yet. ntfs_truncate() actually only
911 emits a warning message but AFAICS our interception of i_size changes
912 elsewhere means ntfs_truncate() never gets called for i_size changes.
913 It is only called from generic_file_write() when we fail in
914 ntfs_prepare_{,nonresident_}write() in order to discard any
915 instantiated buffers beyond i_size. Thus i_size is not actually
916 changed so our warning message is enough. Unfortunately it is not
917 possible to easily determine if i_size is being changed or not hence
918 we just emit an appropriately worded error message.
920 2.0.25 - Small bug fixes and cleanups.
922 - Unlock the page in an out of memory error code path in
923 fs/ntfs/aops.c::ntfs_read_block().
924 - If fs/ntfs/aops.c::ntfs_read_page() is called on an uptodate page,
925 just unlock the page and return. (This can happen due to ->writepage
926 clearing PageUptodate() during write out of MstProtected()
928 - Remove leaked write code again.
932 - Treat BUG_ON() as ASSERT() not VERIFY(), i.e. do not use side effects
933 inside BUG_ON(). (Adam J. Richter)
934 - Split logical OR expressions inside BUG_ON() into individual BUG_ON()
935 calls for improved debugging. (Adam J. Richter)
936 - Add errors flag to the ntfs volume state, accessed via
937 NVol{,Set,Clear}Errors(vol).
938 - Do not allow read-write remounts of read-only volumes with errors.
939 - Clarify comment for ntfs file operation sendfile which was added by
940 Christoph Hellwig a while ago (just using generic_file_sendfile())
941 to say that ntfs ->sendfile is only used for the case where the
942 source data is on the ntfs partition and the destination is
943 somewhere else, i.e. nothing we need to concern ourselves with.
944 - Add generic_file_write() as our ntfs file write operation.
946 2.0.23 - Major bug fixes (races, deadlocks, non-i386 architectures).
948 - Massive internal locking changes to mft record locking. Fixes lock
949 recursion and replaces the mrec_lock read/write semaphore with a
950 mutex. Also removes the now superfluous mft_count. This fixes several
951 race conditions and deadlocks, especially in the future write code.
952 - Fix ntfs over loopback for compressed files by adding an
953 optimization barrier. (gcc was screwing up otherwise ?)
954 - Miscellaneous cleanups all over the code and a fix or two in error
956 Thanks go to Christoph Hellwig for pointing out the following two:
957 - Remove now unused function fs/ntfs/malloc.h::vmalloc_nofs().
958 - Fix ntfs_free() for ia64 and parisc by checking for VMALLOC_END, too.
960 2.0.22 - Cleanups, mainly to ntfs_readdir(), and use C99 initializers.
962 - Change fs/ntfs/dir.c::ntfs_reddir() to only read/write ->f_pos once
963 at entry/exit respectively.
964 - Use C99 initializers for structures.
965 - Remove unused variable blocks from fs/ntfs/aops.c::ntfs_read_block().
967 2.0.21 - Check for, and refuse to work with too large files/directories/volumes.
969 - Limit volume size at mount time to 2TiB on architectures where
970 unsigned long is 32-bits (fs/ntfs/super.c::parse_ntfs_boot_sector()).
971 This is the most we can do without overflowing the 32-bit limit of
972 the block device size imposed on us by sb_bread() and sb_getblk()
974 - Limit file/directory size at open() time to 16TiB on architectures
975 where unsigned long is 32-bits (fs/ntfs/file.c::ntfs_file_open() and
976 fs/ntfs/dir.c::ntfs_dir_open()). This is the most we can do without
977 overflowing the page cache page index.
979 2.0.20 - Support non-resident directory index bitmaps, fix page leak in readdir.
981 - Move the directory index bitmap to use an attribute inode instead of
982 having special fields for it inside the ntfs inode structure. This
983 means that the index bitmaps now use the page cache for i/o, too,
984 and also as a side effect we get support for non-resident index
986 - Simplify/cleanup error handling in fs/ntfs/dir.c::ntfs_readdir() and
987 fix a page leak that manifested itself in some cases.
988 - Add fs/ntfs/inode.c::ntfs_put_inode(), which we need to release the
989 index bitmap inode on the final iput().
991 2.0.19 - Fix race condition, improvements, and optimizations in i/o interface.
993 - Apply block optimization added to fs/ntfs/aops.c::ntfs_read_block()
994 to fs/ntfs/compress.c::ntfs_file_read_compressed_block() as well.
995 - Drop the "file" from ntfs_file_read_compressed_block().
996 - Rename fs/ntfs/aops.c::ntfs_enb_buffer_read_async() to
997 ntfs_end_buffer_async_read() (more like the fs/buffer.c counterpart).
998 - Update ntfs_end_buffer_async_read() with the improved logic from
999 its updated counterpart fs/buffer.c::end_buffer_async_read(). Apply
1000 further logic improvements to better determine when we set PageError.
1001 - Update submission of buffers in fs/ntfs/aops.c::ntfs_read_block() to
1002 check for the buffers being uptodate first in line with the updated
1003 fs/buffer.c::block_read_full_page(). This plugs a small race
1006 2.0.18 - Fix race condition in reading of compressed files.
1008 - There was a narrow window between checking a buffer head for being
1009 uptodate and locking it in ntfs_file_read_compressed_block(). We now
1010 lock the buffer and then check whether it is uptodate or not.
1012 2.0.17 - Cleanups and optimizations - shrinking the ToDo list.
1014 - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to return an error
1015 code and update callers, i.e. ntfs_iget(), to pass that error code
1016 up instead of just using -EIO.
1017 - Modifications to super.c to ensure that both mount and remount
1018 cannot set any write related options when the driver is compiled
1020 - Optimize block resolution in fs/ntfs/aops.c::ntfs_read_block() to
1021 cache the current runlist element. This should improve performance
1022 when reading very large and/or very fragmented data.
1024 2.0.16 - Convert access to $MFT/$BITMAP to attribute inode API.
1026 - Fix a stupid bug introduced in 2.0.15 where we were unmapping the
1027 wrong inode in fs/ntfs/inode.c::ntfs_attr_iget().
1028 - Fix debugging check in fs/ntfs/aops.c::ntfs_read_block().
1029 - Convert $MFT/$BITMAP access to attribute inode API and remove all
1030 remnants of the ugly mftbmp address space and operations hack. This
1031 means we finally have only one readpage function as well as only one
1032 async io completion handler. Yey! The mft bitmap is now just an
1033 attribute inode and is accessed from vol->mftbmp_ino just as if it
1034 were a normal file. Fake inodes rule. (-:
1036 2.0.15 - Fake inodes based attribute i/o via the pagecache, fixes and cleanups.
1038 - Fix silly bug in fs/ntfs/super.c::parse_options() which was causing
1039 remounts to fail when the partition had an entry in /etc/fstab and
1040 the entry specified the nls= option.
1041 - Apply same macro magic used in fs/ntfs/inode.h to fs/ntfs/volume.h to
1042 expand all the helper functions NVolFoo(), NVolSetFoo(), and
1044 - Move copyright statement from driver initialisation message to
1045 module description (fs/super.c). This makes the initialisation
1046 message fit on one line and fits in better with rest of kernel.
1047 - Update fs/ntfs/attrib.c::map_run_list() to work on both real and
1048 attribute inodes, and both for files and directories.
1049 - Implement fake attribute inodes allowing all attribute i/o to go via
1050 the page cache and to use all the normal vfs/mm functionality:
1051 - Add ntfs_attr_iget() and its helper ntfs_read_locked_attr_inode()
1053 - Add needed cleanup code to ntfs_clear_big_inode().
1054 - Merge address space operations for files and directories (aops.c),
1055 now just have ntfs_aops:
1057 end_buffer_read_attr_async() -> ntfs_end_buffer_read_async(),
1058 ntfs_attr_read_block() -> ntfs_read_block(),
1059 ntfs_file_read_page() -> ntfs_readpage().
1060 - Rewrite fs/ntfs/aops.c::ntfs_readpage() to work on both real and
1061 attribute inodes, and both for files and directories.
1062 - Remove obsolete fs/ntfs/aops.c::ntfs_mst_readpage().
1064 2.0.14 - Run list merging code cleanup, minor locking changes, typo fixes.
1066 - Change fs/ntfs/super.c::ntfs_statfs() to not rely on BKL by moving
1067 the locking out of super.c::get_nr_free_mft_records() and taking and
1068 dropping the mftbmp_lock rw_semaphore in ntfs_statfs() itself.
1069 - Bring attribute runlist merging code (fs/ntfs/attrib.c) in sync with
1070 current userspace ntfs library code. This means that if a merge
1071 fails the original runlists are always left unmodified instead of
1072 being silently corrupted.
1075 2.0.13 - Use iget5_locked() in preparation for fake inodes and small cleanups.
1077 - Remove nr_mft_bits and the now superfluous union with nr_mft_records
1078 from ntfs_volume structure.
1079 - Remove nr_lcn_bits and the now superfluous union with nr_clusters
1080 from ntfs_volume structure.
1081 - Use iget5_locked() and friends instead of conventional iget(). Wrap
1082 the call in fs/ntfs/inode.c::ntfs_iget() and update callers of iget()
1083 to use ntfs_iget(). Leave only one iget() call at mount time so we
1084 don't need an ntfs_iget_mount().
1085 - Change fs/ntfs/inode.c::ntfs_new_extent_inode() to take mft_no as an
1086 additional argument.
1088 2.0.12 - Initial cleanup of address space operations following 2.0.11 changes.
1090 - Merge fs/ntfs/aops.c::end_buffer_read_mst_async() and
1091 fs/ntfs/aops.c::end_buffer_read_file_async() into one function
1092 fs/ntfs/aops.c::end_buffer_read_attr_async() using NInoMstProtected()
1093 to determine whether to apply mst fixups or not.
1094 - Above change allows merging fs/ntfs/aops.c::ntfs_file_read_block()
1095 and fs/ntfs/aops.c::ntfs_mst_readpage() into one function
1096 fs/ntfs/aops.c::ntfs_attr_read_block(). Also, create a tiny wrapper
1097 fs/ntfs/aops.c::ntfs_mst_readpage() to transform the parameters from
1098 the VFS readpage function prototype to the ntfs_attr_read_block()
1101 2.0.11 - Initial preparations for fake inode based attribute i/o.
1103 - Move definition of ntfs_inode_state_bits to fs/ntfs/inode.h and
1104 do some macro magic (adapted from include/linux/buffer_head.h) to
1105 expand all the helper functions NInoFoo(), NInoSetFoo(), and
1107 - Add new flag to ntfs_inode_state_bits: NI_Sparse.
1108 - Add new fields to ntfs_inode structure to allow use of fake inodes
1109 for attribute i/o: type, name, name_len. Also add new state bits:
1110 NI_Attr, which, if set, indicates the inode is a fake inode, and
1111 NI_MstProtected, which, if set, indicates the attribute uses multi
1112 sector transfer protection, i.e. fixups need to be applied after
1113 reads and before/after writes.
1114 - Rename fs/ntfs/inode.c::ntfs_{new,clear,destroy}_inode() to
1115 ntfs_{new,clear,destroy}_extent_inode() and update callers.
1116 - Use ntfs_clear_extent_inode() in fs/ntfs/inode.c::__ntfs_clear_inode()
1117 instead of ntfs_destroy_extent_inode().
1118 - Cleanup memory deallocations in {__,}ntfs_clear_{,big_}inode().
1119 - Make all operations on ntfs inode state bits use the NIno* functions.
1120 - Set up the new ntfs inode fields and state bits in
1121 fs/ntfs/inode.c::ntfs_read_inode() and add appropriate cleanup of
1122 allocated memory to __ntfs_clear_inode().
1123 - Cleanup ntfs_inode structure a bit for better ordering of elements
1124 w.r.t. their size to allow better packing of the structure in memory.
1126 2.0.10 - There can only be 2^32 - 1 inodes on an NTFS volume.
1128 - Add check at mount time to verify that the number of inodes on the
1129 volume does not exceed 2^32 - 1, which is the maximum allowed for
1130 NTFS according to Microsoft.
1131 - Change mft_no member of ntfs_inode structure to be unsigned long.
1132 Update all users. This makes ntfs_inode->mft_no just a copy of struct
1133 inode->i_ino. But we can't just always use struct inode->i_ino and
1134 remove mft_no because extent inodes do not have an attached struct
1137 2.0.9 - Decompression engine now uses a single buffer and other cleanups.
1139 - Change decompression engine to use a single buffer protected by a
1140 spin lock instead of per-CPU buffers. (Rusty Russell)
1141 - Do not update cb_pos when handling a partial final page during
1142 decompression of a sparse compression block, as the value is later
1143 reset without being read/used. (Rusty Russell)
1144 - Switch to using the new KM_BIO_SRC_IRQ for atomic kmap()s. (Andrew
1146 - Change buffer size in ntfs_readdir()/ntfs_filldir() to use
1147 NLS_MAX_CHARSET_SIZE which makes the buffers almost 1kiB each but
1148 it also makes everything safer so it is a good thing.
1149 - Miscellaneous minor cleanups to comments.
1151 2.0.8 - Major updates for handling of case sensitivity and dcache aliasing.
1153 Big thanks go to Al Viro and other inhabitants of #kernel for investing
1154 their time to discuss the case sensitivity and dcache aliasing issues.
1156 - Remove unused source file fs/ntfs/attraops.c.
1157 - Remove show_inodes mount option(s), thus dropping support for
1158 displaying of short file names.
1159 - Remove deprecated mount option posix.
1160 - Restore show_sys_files mount option.
1161 - Add new mount option case_sensitive, to determine if the driver
1162 treats file names as case sensitive or not. If case sensitive, create
1163 file names in the POSIX namespace. Otherwise create file names in the
1164 LONG/WIN32 namespace. Note, files remain accessible via their short
1165 file name, if it exists.
1166 - Remove really dumb logic bug in boot sector recovery code.
1167 - Fix dcache aliasing issues wrt short/long file names via changes
1168 to fs/ntfs/dir.c::ntfs_lookup_inode_by_name() and
1169 fs/ntfs/namei.c::ntfs_lookup():
1170 - Add additional argument to ntfs_lookup_inode_by_name() in which we
1171 return information about the matching file name if the case is not
1172 matching or the match is a short file name. See comments above the
1173 function definition for details.
1174 - Change ntfs_lookup() to only create dcache entries for the correctly
1175 cased file name and only for the WIN32 namespace counterpart of DOS
1176 namespace file names. This ensures we have only one dentry per
1177 directory and also removes all dcache aliasing issues between short
1178 and long file names once we add write support. See comments above
1179 function for details.
1180 - Fix potential 1 byte overflow in fs/ntfs/unistr.c::ntfs_ucstonls().
1182 2.0.7 - Minor cleanups and updates for changes in core kernel code.
1184 - Remove much of the NULL struct element initializers.
1185 - Various updates to make compatible with recent kernels.
1186 - Remove defines of MAX_BUF_PER_PAGE and include linux/buffer_head.h
1187 in fs/ntfs/ntfs.h instead.
1188 - Remove no longer needed KERNEL_VERSION checks. We are now in the
1189 kernel proper so they are no longer needed.
1191 2.0.6 - Major bugfix to make compatible with other kernel changes.
1193 - Initialize the mftbmp address space properly now that there are more
1194 fields in the struct address_space. This was leading to hangs and
1195 oopses on umount since 2.5.12 because of changes to other parts of
1196 the kernel. We probably want a kernel generic init_address_space()
1198 - Drop BKL from ntfs_readdir() after consultation with Al Viro. The
1199 only caller of ->readdir() is vfs_readdir() which holds i_sem during
1200 the call, and i_sem is sufficient protection against changes in the
1201 directory inode (including ->i_size).
1202 - Use generic_file_llseek() for directories (as opposed to
1203 default_llseek()) as this downs i_sem instead of the BKL which is
1204 what we now need for exclusion against ->f_pos changes considering we
1205 no longer take the BKL in ntfs_readdir().
1207 2.0.5 - Major bugfix. Buffer overflow in extent inode handling.
1209 - No need to set old blocksize in super.c::ntfs_fill_super() as the
1210 VFS does so via invocation of deactivate_super() calling
1211 fs->fill_super() calling block_kill_super() which does it.
1212 - BKL moved from VFS into dir.c::ntfs_readdir(). (Linus Torvalds)
1213 -> Do we really need it? I don't think so as we have exclusion on
1214 the directory ntfs_inode rw_semaphore mrec_lock. We mmight have to
1215 move the ->f_pos accesses under the mrec_lock though. Check this...
1216 - Fix really, really, really stupid buffer overflow in extent inode
1217 handling in mft.c::map_extent_mft_record().
1219 2.0.4 - Cleanups and updates for kernel 2.5.11.
1221 - Add documentation on how to use the MD driver to be able to use NTFS
1222 stripe and volume sets in Linux and generally cleanup documentation
1224 Remove all uses of kdev_t in favour of struct block_device *:
1225 - Change compress.c::ntfs_file_read_compressed_block() to use
1226 sb_getblk() instead of getblk().
1227 - Change super.c::ntfs_fill_super() to use bdev_hardsect_size() instead
1228 of get_hardsect_size().
1229 - No need to get old blocksize in super.c::ntfs_fill_super() as
1230 fs/super.c::get_sb_bdev() already does this.
1231 - Set bh->b_bdev instead of bh->b_dev throughout aops.c.
1233 2.0.3 - Small bug fixes, cleanups, and performance improvements.
1235 - Remove some dead code from mft.c.
1236 - Optimize readpage and read_block functions throughout aops.c so that
1237 only initialized blocks are read. Non-initialized ones have their
1238 buffer head mapped, zeroed, and set up to date, without scheduling
1239 any i/o. Thanks to Al Viro for advice on how to avoid the device i/o.
1240 Thanks go to Andrew Morton for spotting the below:
1241 - Fix buglet in allocate_compression_buffers() error code path.
1242 - Call flush_dcache_page() after modifying page cache page contents in
1243 ntfs_file_readpage().
1244 - Check for existence of page buffers throughout aops.c before calling
1245 create_empty_buffers(). This happens when an I/O error occurs and the
1246 read is retried. (It also happens once writing is implemented so that
1247 needed doing anyway but I had left it for later...)
1248 - Don't BUG_ON() uptodate and/or mapped buffers throughout aops.c in
1249 readpage and read_block functions. Reasoning same as above (i.e. I/O
1250 error retries and future write code paths.)
1252 2.0.2 - Minor updates and cleanups.
1254 - Cleanup: rename mst.c::__post_read_mst_fixup to post_write_mst_fixup
1255 and cleanup the code a bit, removing the unused size parameter.
1256 - Change default fmask to 0177 and update documentation.
1257 - Change attrib.c::get_attr_search_ctx() to return the search context
1258 directly instead of taking the address of a pointer. A return value
1259 of NULL means the allocation failed. Updated all callers
1261 - Update to 2.5.9 kernel (preserving backwards compatibility) by
1262 replacing all occurences of page->buffers with page_buffers(page).
1263 - Fix minor bugs in runlist merging, also minor cleanup.
1264 - Updates to bootsector layout and mft mirror contents descriptions.
1265 - Small bug fix in error detection in unistr.c and some cleanups.
1266 - Grow name buffer allocations in unistr.c in aligned mutlipled of 64
1269 2.0.1 - Minor updates.
1271 - Make default umask correspond to documentation.
1272 - Improve documentation.
1273 - Set default mode to include execute bit. The {u,f,d}mask can be used
1274 to take it away if desired. This allows binaries to be executed from
1275 a mounted ntfs partition.
1277 2.0.0 - New version number. Remove TNG from the name. Now in the kernel.
1279 - Add kill_super, just keeping up with the vfs changes in the kernel.
1280 - Repeat some changes from tng-0.0.8 that somehow got lost on the way
1281 from the CVS import into BitKeeper.
1282 - Begin to implement proper handling of allocated_size vs
1283 initialized_size vs data_size (i.e. i_size). Done are
1284 mft.c::ntfs_mft_readpage(), aops.c::end_buffer_read_index_async(),
1285 and attrib.c::load_attribute_list().
1286 - Lock the runlist in attrib.c::load_attribute_list() while using it.
1287 - Fix memory leak in ntfs_file_read_compressed_block() and generally
1288 clean up compress.c a little, removing some uncommented/unused debug
1290 - Tidy up dir.c a little bit.
1291 - Don't bother getting the runlist in inode.c::ntfs_read_inode().
1292 - Merge mft.c::ntfs_mft_readpage() and aops.c::ntfs_index_readpage()
1293 creating aops.c::ntfs_mst_readpage(), improving the handling of
1294 holes and overflow in the process and implementing the correct
1295 equivalent of ntfs_file_get_block() in ntfs_mst_readpage() itself.
1296 I am aiming for correctness at the moment. Modularisation can come
1298 - Rename aops.c::end_buffer_read_index_async() to
1299 end_buffer_read_mst_async() and optimize the overflow checking and
1301 - Use the host of the mftbmp address space mapping to hold the ntfs
1302 volume. This is needed so the async i/o completion handler can
1303 retrieve a pointer to the volume. Hopefully this will not cause
1304 problems elsewhere in the kernel... Otherwise will need to use a
1306 - Complete implementation of proper handling of allocated_size vs
1307 initialized_size vs data_size (i.e. i_size) in whole driver.
1308 Basically aops.c is now completely rewritten.
1309 - Change NTFS driver name to just NTFS and set version number to 2.0.0
1310 to make a clear distinction from the old driver which is still on
1313 tng-0.0.8 - 08/03/2002 - Now using BitKeeper, http://linux-ntfs.bkbits.net/
1315 - Replace bdevname(sb->s_dev) with sb->s_id.
1316 - Remove now superfluous new-line characters in all callers of
1318 - Apply kludge in ntfs_read_inode(), setting i_nlink to 1 for
1319 directories. Without this the "find" utility gets very upset which is
1320 fair enough as Linux/Unix do not support directory hard links.
1321 - Further runlist merging work. (Richard Russon)
1322 - Backwards compatibility for gcc-2.95. (Richard Russon)
1323 - Update to kernel 2.5.5-pre1 and rediff the now tiny patch.
1324 - Convert to new filesystem declaration using ->ntfs_get_sb() and
1325 replacing ntfs_read_super() with ntfs_fill_super().
1326 - Set s_maxbytes to MAX_LFS_FILESIZE to avoid page cache page index
1327 overflow on 32-bit architectures.
1328 - Cleanup upcase loading code to use ntfs_(un)map_page().
1329 - Disable/reenable preemtion in critical sections of compession engine.
1330 - Replace device size determination in ntfs_fill_super() with
1331 sb->s_bdev->bd_inode->i_size (in bytes) and remove now superfluous
1332 function super.c::get_nr_blocks().
1333 - Implement a mount time option (show_inodes) allowing choice of which
1334 types of inode names readdir() returns and modify ntfs_filldir()
1335 accordingly. There are several parameters to show_inodes:
1336 system: system files
1337 win32: long file names (including POSIX file names) [DEFAULT]
1339 dos: short file names only (excluding POSIX file names)
1341 posix: same as both win32 and dos
1343 Note that the options are additive, i.e. specifying:
1344 -o show_inodes=system,show_inodes=win32,show_inodes=dos
1345 is the same as specifying:
1347 Note that the "posix" and "all" options will show all directory
1348 names, BUT the link count on each directory inode entry is set to 1,
1349 due to Linux not supporting directory hard links. This may well
1350 confuse some userspace applications, since the directory names will
1351 have the same inode numbers. Thus it is NOT advisable to use the
1352 "posix" or "all" options. We provide them only for completeness sake.
1353 - Add copies of allocated_size, initialized_size, and compressed_size to
1354 the ntfs inode structure and set them up in
1355 inode.c::ntfs_read_inode(). These reflect the unnamed data attribute
1356 for files and the index allocation attribute for directories.
1357 - Add copies of allocated_size and initialized_size to ntfs inode for
1358 $BITMAP attribute of large directories and set them up in
1359 inode.c::ntfs_read_inode().
1360 - Add copies of allocated_size and initialized_size to ntfs volume for
1361 $BITMAP attribute of $MFT and set them up in
1362 super.c::load_system_files().
1363 - Parse deprecated ntfs driver options (iocharset, show_sys_files,
1364 posix, and utf8) and tell user what the new options to use are. Note
1365 we still do support them but they will be removed with kernel 2.7.x.
1366 - Change all occurences of integer long long printf formatting to hex
1367 as printk() will not support long long integer format if/when the
1368 div64 patch goes into the kernel.
1369 - Make slab caches have stable names and change the names to what they
1370 were intended to be. These changes are required/made possible by the
1371 new slab cache name handling which removes the length limitation by
1372 requiring the caller of kmem_cache_create() to supply a stable name
1373 which is then referenced but not copied.
1374 - Rename run_list structure to run_list_element and create a new
1375 run_list structure containing a pointer to a run_list_element
1376 structure and a read/write semaphore. Adapt all users of runlists
1377 to new scheme and take and release the lock as needed. This fixes a
1378 nasty race as the run_list changes even when inodes are locked for
1379 reading and even when the inode isn't locked at all, so we really
1380 needed the serialization. We use a semaphore rather than a spinlock
1381 as memory allocations can sleep and doing everything GFP_ATOMIC
1383 - Cleanup read_inode() removing all code checking for lowest_vcn != 0.
1384 This can never happen due to the nature of lookup_attr() and how we
1385 support attribute lists. If it did happen it would imply the inode
1387 - Check for lowest_vcn != 0 in ntfs_read_inode() and mark the inode as
1389 - Update to 2.5.6-pre2 changes in struct address_space.
1390 - Use parent_ino() when accessing d_parent inode number in dir.c.
1391 - Import Sourceforge CVS repository into BitKeeper repository:
1392 http://linux-ntfs.bkbits.net/ntfs-tng-2.5
1393 - Update fs/Makefile, fs/Config.help, fs/Config.in, and
1394 Documentation/filesystems/ntfs.txt for NTFS TNG.
1395 - Create kernel configuration option controlling whether debugging
1397 - Add the required export of end_buffer_io_sync() from the patches
1398 directory to the kernel code.
1399 - Update inode.c::ntfs_show_options() with show_inodes mount option.
1400 - Update errors mount option.
1402 tng-0.0.7 - 13/02/2002 - The driver is now feature complete for read-only!
1404 - Cleanup mft.c and it's debug/error output in particular. Fix a minor
1405 bug in mapping of extent inodes. Update all the comments to fit all
1406 the recent code changes.
1407 - Modify vcn_to_lcn() to cope with entirely unmapped runlists.
1408 - Cleanups in compress.c, mostly comments and folding help.
1409 - Implement attrib.c::map_run_list() as a generic helper.
1410 - Make compress.c::ntfs_file_read_compressed_block() use map_run_list()
1411 thus making code shorter and enabling attribute list support.
1412 - Cleanup incorrect use of [su]64 with %L printf format specifier in
1413 all source files. Type casts to [unsigned] long long added to correct
1414 the mismatches (important for architectures which have long long not
1416 - Merge async io completion handlers for directory indexes and $MFT
1417 data into one by setting the index_block_size{_bits} of the ntfs
1418 inode for $MFT to the mft_record_size{_bits} of the ntfs_volume.
1419 - Cleanup aops.c, update comments.
1420 - Make ntfs_file_get_block() use map_run_list() so all files now
1421 support attribute lists.
1422 - Make ntfs_dir_readpage() almost verbatim copy of
1423 block_read_full_page() by using ntfs_file_get_block() with only real
1424 difference being the use of our own async io completion handler
1425 rather than the default one, thus reducing the amount of code and
1426 automatically enabling attribute list support for directory indices.
1427 - Fix bug in load_attribute_list() - forgot to call brelse in error
1429 - Change parameters to find_attr() and lookup_attr(). We no longer
1430 pass in the upcase table and its length. These can be gotten from
1431 ctx->ntfs_ino->vol->upcase{_len}. Update all callers.
1432 - Cleanups in attrib.c.
1433 - Implement merging of runlists, attrib.c::merge_run_lists() and its
1434 helpers. (Richard Russon)
1435 - Attribute lists part 2, attribute extents and multi part runlists:
1436 enable proper support for LCN_RL_NOT_MAPPED and automatic mapping of
1437 further runlist parts via attrib.c::map_run_list().
1438 - Tiny endianness bug fix in decompress_mapping_pairs().
1440 tng-0.0.6 - Encrypted directories, bug fixes, cleanups, debugging enhancements.
1442 - Enable encrypted directories. (Their index root is marked encrypted
1443 to indicate that new files in that directory should be created
1445 - Fix bug in NInoBmpNonResident() macro. (Cut and paste error.)
1446 - Enable $Extend system directory. Most (if not all) extended system
1447 files do not have unnamed data attributes so ntfs_read_inode() had to
1448 special case them but that is ok, as the special casing recovery
1449 happens inside an error code path so there is zero slow down in the
1450 normal fast path. The special casing is done by introducing a new
1451 function inode.c::ntfs_is_extended_system_file() which checks if any
1452 of the hard links in the inode point to $Extend as being their parent
1453 directory and if they do we assume this is an extended system file.
1454 - Create a sysctl/proc interface to allow {dis,en}abling of debug output
1455 when compiled with -DDEBUG. Default is debug messages to be disabled.
1456 To enable them, one writes a non-zero value to /proc/sys/fs/ntfs-debug
1457 (if /proc is enabled) or uses sysctl(2) to effect the same (if sysctl
1458 interface is enabled). Inspired by old ntfs driver.
1459 - Add debug_msgs insmod/kernel boot parameter to set whether debug
1460 messages are {dis,en}abled. This is useful to enable debug messages
1461 during ntfs initialization and is the only way to activate debugging
1462 when the sysctl interface is not enabled.
1463 - Cleanup debug output in various places.
1464 - Remove all dollar signs ($) from the source (except comments) to
1465 enable compilation on architectures whose gcc compiler does not
1466 support dollar signs in the names of variables/constants. Attribute
1467 types now start with AT_ instead of $ and $I30 is now just I30.
1468 - Cleanup ntfs_lookup() and add consistency check of sequence numbers.
1469 - Load complete runlist for $MFT/$BITMAP during mount and cleanup
1470 access functions. This means we now cope with $MFT/$BITMAP being
1471 spread accross several mft records.
1472 - Disable modification of mft_zone_multiplier on remount. We can always
1473 reenable this later on if we really want to, but we will need to make
1474 sure we readjust the mft_zone size / layout accordingly.
1476 tng-0.0.5 - Modernize for 2.5.x and further in line-ing with Al Viro's comments.
1478 - Use sb_set_blocksize() instead of set_blocksize() and verify the
1480 - Use sb_bread() instead of bread() throughout.
1481 - Add index_vcn_size{_bits} to ntfs_inode structure to store the size
1482 of a directory index block vcn. Apply resulting simplifications in
1484 - Fix a small bug somewhere (but forgot what it was).
1485 - Change ntfs_{debug,error,warning} to enable gcc to do type checking
1486 on the printf-format parameter list and fix bugs reported by gcc
1487 as a result. (Richard Russon)
1488 - Move inode allocation strategy to Al's new stuff but maintain the
1489 divorce of ntfs_inode from struct inode. To achieve this we have two
1490 separate slab caches, one for big ntfs inodes containing a struct
1491 inode and pure ntfs inodes and at the same time fix some faulty
1492 error code paths in ntfs_read_inode().
1493 - Show mount options in proc (inode.c::ntfs_show_options()).
1495 tng-0.0.4 - Big changes, getting in line with Al Viro's comments.
1497 - Modified (un)map_mft_record functions to be common for read and write
1498 case. To specify which is which, added extra parameter at front of
1499 parameter list. Pass either READ or WRITE to this, each has the
1501 - General cleanups to allow for easier folding in vi.
1502 - attrib.c::decompress_mapping_pairs() now accepts the old runlist
1503 argument, and invokes attrib.c::merge_run_lists() to merge the old
1504 and the new runlists.
1505 - Removed attrib.c::find_first_attr().
1506 - Implemented loading of attribute list and complete runlist for $MFT.
1507 This means we now cope with $MFT being spread across several mft
1509 - Adapt to 2.5.2-pre9 and the changed create_empty_buffers() syntax.
1510 - Adapt major/minor/kdev_t/[bk]devname stuff to new 2.5.x kernels.
1511 - Make ntfs_volume be allocated via kmalloc() instead of using a slab
1512 cache. There are too little ntfs_volume structures at any one time
1513 to justify a private slab cache.
1514 - Fix bogus kmap() use in async io completion. Now use kmap_atomic().
1515 Use KM_BIO_IRQ on advice from IRC/kernel...
1516 - Use ntfs_map_page() in map_mft_record() and create ->readpage method
1517 for reading $MFT (ntfs_mft_readpage). In the process create dedicated
1518 address space operations (ntfs_mft_aops) for $MFT inode mapping. Also
1519 removed the now superfluous exports from the kernel core patch.
1520 - Fix a bug where kfree() was used insted of ntfs_free().
1521 - Change map_mft_record() to take ntfs_inode as argument instead of
1522 vfs inode. Dito for unmap_mft_record(). Adapt all callers.
1523 - Add pointer to ntfs_volume to ntfs_inode.
1524 - Add mft record number and sequence number to ntfs_inode. Stop using
1525 i_ino and i_generation for in-driver purposes.
1526 - Implement attrib.c::merge_run_lists(). (Richard Russon)
1527 - Remove use of proper inodes by extent inodes. Move i_ino and
1528 i_generation to ntfs_inode to do this. Apply simplifications that
1529 result and remove iget_no_wait(), etc.
1530 - Pass ntfs_inode everywhere in the driver (used to be struct inode).
1531 - Add reference counting in ntfs_inode for the ntfs inode itself and
1532 for the mapped mft record.
1533 - Extend mft record mapping so we can (un)map extent mft records (new
1534 functions (un)map_extent_mft_record), and so mappings are reference
1535 counted and don't have to happen twice if already mapped - just ref
1537 - Add -o iocharset as alias to -o nls for backwards compatibility.
1538 - The latest core patch is now tiny. In fact just a single additional
1539 export is necessary over the base kernel.
1541 tng-0.0.3 - Cleanups, enhancements, bug fixes.
1543 - Work on attrib.c::decompress_mapping_pairs() to detect base extents
1544 and setup the runlist appropriately using knowledge provided by the
1545 sizes in the base attribute record.
1546 - Balance the get_/put_attr_search_ctx() calls so we don't leak memory
1548 - Introduce ntfs_malloc_nofs() and ntfs_free() to allocate/free a single
1549 page or use vmalloc depending on the amount of memory requested.
1550 - Cleanup error output. The __FUNCTION__ "(): " is now added
1551 automatically. Introduced a new header file debug.h to support this
1552 and also moved ntfs_debug() function into it.
1553 - Make reading of compressed files more intelligent and especially get
1554 rid of the vmalloc_nofs() from readpage(). This now uses per CPU
1555 buffers (allocated at first mount with cluster size <= 4kiB and
1556 deallocated on last umount with cluster size <= 4kiB), and
1557 asynchronous io for the compressed data using a list of buffer heads.
1558 Er, we use synchronous io as async io only works on whole pages
1559 covered by buffers and not on individual buffer heads...
1560 - Bug fix for reading compressed files with sparse compression blocks.
1562 tng-0.0.2 - Now handles larger/fragmented/compressed volumes/files/dirs.
1564 - Fixed handling of directories when cluster size exceeds index block
1566 - Hide DOS only name space directory entries from readdir() but allow
1567 them in lookup(). This should fix the problem that Linux doesn't
1568 support directory hard links, while still allowing access to entries
1569 via their short file name. This also has the benefit of mimicking
1570 what Windows users are used to, so it is the ideal solution.
1571 - Implemented sync_page everywhere so no more hangs in D state when
1573 - Stop using bforget() in favour of brelse().
1574 - Stop locking buffers unnecessarily.
1575 - Implemented compressed files (inode->mapping contains uncompressed
1576 data, raw compressed data is currently bread() into a vmalloc()ed
1578 - Enable compressed directories. (Their index root is marked compressed
1579 to indicate that new files in that directory should be created
1581 - Use vsnprintf rather than vsprintf in the ntfs_error and ntfs_warning
1582 functions. (Thanks to Will Dyson for pointing this out.)
1583 - Moved the ntfs_inode and ntfs_volume (the former ntfs_inode_info and
1584 ntfs_sb_info) out of the common inode and super_block structures and
1585 started using the generic_ip and generic_sbp pointers instead. This
1586 makes ntfs entirely private with respect to the kernel tree.
1587 - Detect compiler version and abort with error message if gcc less than
1589 - Fix bug in name comparison function in unistr.c.
1590 - Implement attribute lists part 1, the infrastructure: search contexts
1591 and operations, find_external_attr(), lookup_attr()) and make the
1592 code use the infrastructure.
1593 - Fix stupid buffer overflow bug that became apparent on larger run
1594 list containing attributes.
1595 - Fix bugs in readdir() that became apparent on larger directories.
1597 The driver is now really useful and survives the test
1598 find . -type f -exec md5sum "{}" \;
1599 without any error messages on a over 1GiB sized partition with >16k
1600 files on it, including compressed files and directories and many files
1601 and directories with attribute lists.
1603 tng-0.0.1 - The first useful version.
1605 - Added ntfs_lookup().
1606 - Added default upcase generation and handling.
1607 - Added compile options to be shown on module init.
1608 - Many bug fixes that were "hidden" before.
1609 - Update to latest kernel.
1610 - Added ntfs_readdir().
1611 - Added file operations for mmap(), read(), open() and llseek(). We just
1612 use the generic ones. The whole point of going through implementing
1613 readpage() methods and where possible get_block() call backs is that
1614 this allows us to make use of the generic high level methods provided
1617 The driver is now actually useful! Yey. (-: It undoubtedly has got bugs
1618 though and it doesn't implement accesssing compressed files yet. Also,
1619 accessing files with attribute list attributes is not implemented yet
1620 either. But for small or simple filesystems it should work and allow
1621 you to list directories, use stat on directory entries and the file
1622 system, open, read, mmap and llseek around in files. A big mile stone
1625 tng-0.0.0 - Initial version tag.
1627 Initial driver implementation. The driver can mount and umount simple
1628 NTFS filesystems (i.e. ones without attribute lists in the system
1629 files). If the mount fails there might be problems in the error handling
1630 code paths, so be warned. Otherwise it seems to be loading the system
1631 files nicely and the mft record read mapping/unmapping seems to be
1632 working nicely, too. Proof of inode metadata in the page cache and non-
1633 resident file unnamed stream data in the page cache concepts is thus
projects / linux-2.6 / blob