blob: 6d2a99c13d3059e5cc5fcff3bda74bc7eb5396ee [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001ToDo/Notes:
2 - Find and fix bugs.
3 - Checkpoint or disable the user space journal ($UsnJrnl).
4 - In between ntfs_prepare/commit_write, need exclusion between
Anton Altaparmakov66129f82004-11-11 12:34:00 +00005 simultaneous file extensions. This is given to us by holding i_sem
6 on the inode. The only places in the kernel when a file is resized
7 are prepare/commit write and truncate for both of which i_sem is
8 held. Just have to be careful in readpage/writepage and all other
9 helpers not running under i_sem that we play nice...
10 Also need to be careful with initialized_size extention in
Linus Torvalds1da177e2005-04-16 15:20:36 -070011 ntfs_prepare_write. Basically, just be _very_ careful in this code...
Anton Altaparmakov66129f82004-11-11 12:34:00 +000012 UPDATE: The only things that need to be checked are read/writepage
13 which do not hold i_sem. Note writepage cannot change i_size but it
14 needs to cope with a concurrent i_size change, just like readpage.
15 Also both need to cope with concurrent changes to the other sizes,
16 i.e. initialized/allocated/compressed size, as well.
Linus Torvalds1da177e2005-04-16 15:20:36 -070017 - Implement mft.c::sync_mft_mirror_umount(). We currently will just
18 leave the volume dirty on umount if the final iput(vol->mft_ino)
19 causes a write of any mirrored mft records due to the mft mirror
20 inode having been discarded already. Whether this can actually ever
21 happen is unclear however so it is worth waiting until someone hits
22 the problem.
23 - Enable the code for setting the NT4 compatibility flag when we start
24 making NTFS 1.2 specific modifications.
25
262.1.23-WIP
27
28 - Add printk rate limiting for ntfs_warning() and ntfs_error() when
29 compiled without debug. This avoids a possible denial of service
30 attack. Thanks to Carl-Daniel Hailfinger from SuSE for pointing this
31 out.
Anton Altaparmakovda284382004-11-11 11:18:10 +000032 - Use i_size_read() in fs/ntfs/attrib.c::ntfs_attr_set().
Anton Altaparmakov66129f82004-11-11 12:34:00 +000033 - Use i_size_read() in fs/ntfs/logfile.c::ntfs_{check,empty}_logfile().
Anton Altaparmakovdb30d162004-11-11 12:42:00 +000034 - Use i_size_read() once and then use the cached value in
35 fs/ntfs/lcnalloc.c::ntfs_cluster_alloc().
Anton Altaparmakovd4b9ba72004-11-17 15:45:08 +000036 - Use i_size_read() in fs/ntfs/file.c::ntfs_file_open().
Linus Torvalds1da177e2005-04-16 15:20:36 -070037
382.1.22 - Many bug and race fixes and error handling improvements.
39
40 - Improve error handling in fs/ntfs/inode.c::ntfs_truncate().
41 - Change fs/ntfs/inode.c::ntfs_truncate() to return an error code
42 instead of void and provide a helper ntfs_truncate_vfs() for the
43 vfs ->truncate method.
44 - Add a new ntfs inode flag NInoTruncateFailed() and modify
45 fs/ntfs/inode.c::ntfs_truncate() to set and clear it appropriately.
46 - Fix min_size and max_size definitions in ATTR_DEF structure in
47 fs/ntfs/layout.h to be signed.
48 - Add attribute definition handling helpers to fs/ntfs/attrib.[hc]:
49 ntfs_attr_size_bounds_check(), ntfs_attr_can_be_non_resident(), and
50 ntfs_attr_can_be_resident(), which in turn use the new private helper
51 ntfs_attr_find_in_attrdef().
52 - In fs/ntfs/aops.c::mark_ntfs_record_dirty(), take the
53 mapping->private_lock around the dirtying of the buffer heads
54 analagous to the way it is done in __set_page_dirty_buffers().
55 - Ensure the mft record size does not exceed the PAGE_CACHE_SIZE at
56 mount time as this cannot work with the current implementation.
57 - Check for location of attribute name and improve error handling in
58 general in fs/ntfs/inode.c::ntfs_read_locked_inode() and friends.
59 - In fs/ntfs/aops.c::ntfs_writepage(), if the page is fully outside
60 i_size, i.e. race with truncate, invalidate the buffers on the page
61 so that they become freeable and hence the page does not leak.
62 - Remove unused function fs/ntfs/runlist.c::ntfs_rl_merge(). (Adrian
63 Bunk)
64 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_find() that resulted in
65 a NULL pointer dereference in the error code path when a corrupt
66 attribute was found. (Thanks to Domen Puncer for the bug report.)
67 - Add MODULE_VERSION() to fs/ntfs/super.c.
68 - Make several functions and variables static. (Adrian Bunk)
69 - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() so it allocates
70 buffers for the page if they are not present and then marks the
71 buffers belonging to the ntfs record dirty. This causes the buffers
72 to become busy and hence they are safe from removal until the page
73 has been written out.
74 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find() in the
75 error handling code path that resulted in a BUG() due to trying to
76 unmap an extent mft record when the mapping of it had failed and it
77 thus was not mapped. (Thanks to Ken MacFerrin for the bug report.)
78 - Drop the runlist lock after the vcn has been read in
79 fs/ntfs/lcnalloc.c::__ntfs_cluster_free().
80 - Rewrite handling of multi sector transfer errors. We now do not set
81 PageError() when such errors are detected in the async i/o handler
82 fs/ntfs/aops.c::ntfs_end_buffer_async_read(). All users of mst
83 protected attributes now check the magic of each ntfs record as they
84 use it and act appropriately. This has the effect of making errors
85 granular per ntfs record rather than per page which solves the case
86 where we cannot access any of the ntfs records in a page when a
87 single one of them had an mst error. (Thanks to Ken MacFerrin for
88 the bug report.)
89 - Fix error handling in fs/ntfs/quota.c::ntfs_mark_quotas_out_of_date()
90 where we failed to release i_sem on the $Quota/$Q attribute inode.
91 - Fix bug in handling of bad inodes in fs/ntfs/namei.c::ntfs_lookup().
92 - Add mapping of unmapped buffers to all remaining code paths, i.e.
93 fs/ntfs/aops.c::ntfs_write_mst_block(), mft.c::ntfs_sync_mft_mirror(),
94 and write_mft_record_nolock(). From now on we require that the
95 complete runlist for the mft mirror is always mapped into memory.
96 - Add creation of buffers to fs/ntfs/mft.c::ntfs_sync_mft_mirror().
97 - Improve error handling in fs/ntfs/aops.c::ntfs_{read,write}_block().
98 - Cleanup fs/ntfs/aops.c::ntfs_{read,write}page() since we know that a
99 resident attribute will be smaller than a page which makes the code
100 simpler. Also make the code more tolerant to concurrent ->truncate.
101
1022.1.21 - Fix some races and bugs, rewrite mft write code, add mft allocator.
103
104 - Implement extent mft record deallocation
105 fs/ntfs/mft.c::ntfs_extent_mft_record_free().
106 - Splitt runlist related functions off from attrib.[hc] to runlist.[hc].
107 - Add vol->mft_data_pos and initialize it at mount time.
108 - Rename init_runlist() to ntfs_init_runlist(), ntfs_vcn_to_lcn() to
109 ntfs_rl_vcn_to_lcn(), decompress_mapping_pairs() to
110 ntfs_mapping_pairs_decompress(), ntfs_merge_runlists() to
111 ntfs_runlists_merge() and adapt all callers.
112 - Add fs/ntfs/runlist.[hc]::ntfs_get_nr_significant_bytes(),
113 ntfs_get_size_for_mapping_pairs(), ntfs_write_significant_bytes(),
114 and ntfs_mapping_pairs_build(), adapted from libntfs.
115 - Make fs/ntfs/lcnalloc.c::ntfs_cluster_free_from_rl_nolock() not
116 static and add a declaration for it to lcnalloc.h.
117 - Add fs/ntfs/lcnalloc.h::ntfs_cluster_free_from_rl() which is a static
118 inline wrapper for ntfs_cluster_free_from_rl_nolock() which takes the
119 cluster bitmap lock for the duration of the call.
120 - Add fs/ntfs/attrib.[hc]::ntfs_attr_record_resize().
121 - Implement the equivalent of memset() for an ntfs attribute in
122 fs/ntfs/attrib.[hc]::ntfs_attr_set() and switch
123 fs/ntfs/logfile.c::ntfs_empty_logfile() to using it.
124 - Remove unnecessary casts from LCN_* constants.
125 - Implement fs/ntfs/runlist.c::ntfs_rl_truncate_nolock().
126 - Add MFT_RECORD_OLD as a copy of MFT_RECORD in fs/ntfs/layout.h and
127 change MFT_RECORD to contain the NTFS 3.1+ specific fields.
128 - Add a helper function fs/ntfs/aops.c::mark_ntfs_record_dirty() which
129 marks all buffers belonging to an ntfs record dirty, followed by
130 marking the page the ntfs record is in dirty and also marking the vfs
131 inode containing the ntfs record dirty (I_DIRTY_PAGES).
132 - Switch fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to using the
133 new helper fs/ntfs/aops.c::mark_ntfs_record_dirty() and remove the no
134 longer needed fs/ntfs/index.[hc]::__ntfs_index_entry_mark_dirty().
135 - Move ntfs_{un,}map_page() from ntfs.h to aops.h and fix resulting
136 include errors.
137 - Move the typedefs for runlist_element and runlist from types.h to
138 runlist.h and fix resulting include errors.
139 - Remove unused {__,}format_mft_record() from fs/ntfs/mft.c.
140 - Modify fs/ntfs/mft.c::__mark_mft_record_dirty() to use the helper
141 mark_ntfs_record_dirty() which also changes the behaviour in that we
142 now set the buffers belonging to the mft record dirty as well as the
143 page itself.
144 - Update fs/ntfs/mft.c::write_mft_record_nolock() and sync_mft_mirror()
145 to cope with the fact that there now are dirty buffers in mft pages.
146 - Update fs/ntfs/inode.c::ntfs_write_inode() to also use the helper
147 mark_ntfs_record_dirty() and thus to set the buffers belonging to the
148 mft record dirty as well as the page itself.
149 - Fix compiler warnings on x86-64 in fs/ntfs/dir.c. (Randy Dunlap,
150 slightly modified by me)
151 - Add fs/ntfs/mft.c::try_map_mft_record() which fails with -EALREADY if
152 the mft record is already locked and otherwise behaves the same way
153 as fs/ntfs/mft.c::map_mft_record().
154 - Modify fs/ntfs/mft.c::write_mft_record_nolock() so that it only
155 writes the mft record if the buffers belonging to it are dirty.
156 Otherwise we assume that it was written out by other means already.
157 - Attempting to write outside initialized size is _not_ a bug so remove
158 the bug check from fs/ntfs/aops.c::ntfs_write_mst_block(). It is in
159 fact required to write outside initialized size when preparing to
160 extend the initialized size.
161 - Map the page instead of using page_address() before writing to it in
162 fs/ntfs/aops.c::ntfs_mft_writepage().
163 - Provide exclusion between opening an inode / mapping an mft record
164 and accessing the mft record in fs/ntfs/mft.c::ntfs_mft_writepage()
165 by setting the page not uptodate throughout ntfs_mft_writepage().
166 - Clear the page uptodate flag in fs/ntfs/aops.c::ntfs_write_mst_block()
167 to ensure noone can see the page whilst the mst fixups are applied.
168 - Add the helper fs/ntfs/mft.c::ntfs_may_write_mft_record() which
169 checks if an mft record may be written out safely obtaining any
170 necessary locks in the process. This is used by
171 fs/ntfs/aops.c::ntfs_write_mst_block().
172 - Modify fs/ntfs/aops.c::ntfs_write_mst_block() to also work for
173 writing mft records and improve its error handling in the process.
174 Now if any of the records in the page fail to be written out, all
175 other records will be written out instead of aborting completely.
176 - Remove ntfs_mft_aops and update all users to use ntfs_mst_aops.
177 - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to set the
178 ntfs_mst_aops for all inodes which are NInoMstProtected() and
179 ntfs_aops for all other inodes.
180 - Rename fs/ntfs/mft.c::sync_mft_mirror{,_umount}() to
181 ntfs_sync_mft_mirror{,_umount}() and change their parameters so they
182 no longer require an ntfs inode to be present. Update all callers.
183 - Cleanup the error handling in fs/ntfs/mft.c::ntfs_sync_mft_mirror().
184 - Clear the page uptodate flag in fs/ntfs/mft.c::ntfs_sync_mft_mirror()
185 to ensure noone can see the page whilst the mst fixups are applied.
186 - Remove the no longer needed fs/ntfs/mft.c::ntfs_mft_writepage() and
187 fs/ntfs/mft.c::try_map_mft_record().
188 - Fix callers of fs/ntfs/aops.c::mark_ntfs_record_dirty() to call it
189 with the ntfs inode which contains the page rather than the ntfs
190 inode the mft record of which is in the page.
191 - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by moving the
192 index inode bitmap inode release code from there to
193 fs/ntfs/inode.c::ntfs_clear_big_inode(). (Thanks to Christoph
194 Hellwig for spotting this.)
195 - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by taking the
196 inode semaphore around the code that sets ni->itype.index.bmp_ino to
197 NULL and reorganize the code to optimize it a bit. (Thanks to
198 Christoph Hellwig for spotting this.)
199 - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() to no longer take the
200 ntfs inode as a parameter as this is confusing and misleading and the
201 needed ntfs inode is available via NTFS_I(page->mapping->host).
202 Adapt all callers to this change.
203 - Modify fs/ntfs/mft.c::write_mft_record_nolock() and
204 fs/ntfs/aops.c::ntfs_write_mst_block() to only check the dirty state
205 of the first buffer in a record and to take this as the ntfs record
206 dirty state. We cannot look at the dirty state for subsequent
207 buffers because we might be racing with
208 fs/ntfs/aops.c::mark_ntfs_record_dirty().
209 - Move the static inline ntfs_init_big_inode() from fs/ntfs/inode.c to
210 inode.h and make fs/ntfs/inode.c::__ntfs_init_inode() non-static and
211 add a declaration for it to inode.h. Fix some compilation issues
212 that resulted due to #includes and header file interdependencies.
213 - Simplify setup of i_mode in fs/ntfs/inode.c::ntfs_read_locked_inode().
214 - Add helpers fs/ntfs/layout.h::MK_MREF() and MK_LE_MREF().
215 - Modify fs/ntfs/mft.c::map_extent_mft_record() to only verify the mft
216 record sequence number if it is specified (i.e. not zero).
217 - Add fs/ntfs/mft.[hc]::ntfs_mft_record_alloc() and various helper
218 functions used by it.
219 - Update Documentation/filesystems/ntfs.txt with instructions on how to
220 use the Device-Mapper driver with NTFS ftdisk/LDM raid. This removes
221 the linear raid problem with the Software RAID / MD driver when one
222 or more of the devices has an odd number of sectors.
223
2242.1.20 - Fix two stupid bugs introduced in 2.1.18 release.
225
226 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_reinit_search_ctx()
227 where we did not clear ctx->al_entry but it was still set due to
228 changes in ntfs_attr_lookup() and ntfs_external_attr_find() in
229 particular.
230 - Fix another stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find()
231 where we forgot to unmap the extent mft record when we had finished
232 enumerating an attribute which caused a bug check to trigger when the
233 VFS calls ->clear_inode.
234
2352.1.19 - Many cleanups, improvements, and a minor bug fix.
236
237 - Update ->setattr (fs/ntfs/inode.c::ntfs_setattr()) to refuse to
238 change the uid, gid, and mode of an inode as we do not support NTFS
239 ACLs yet.
240 - Remove BKL use from ntfs_setattr() syncing up with the rest of the
241 kernel.
242 - Get rid of the ugly transparent union in fs/ntfs/dir.c::ntfs_readdir()
243 and ntfs_filldir() as per suggestion from Al Viro.
244 - Change '\0' and L'\0' to simply 0 as per advice from Linus Torvalds.
245 - Update ->truncate (fs/ntfs/inode.c::ntfs_truncate()) to check if the
246 inode size has changed and to only output an error if so.
247 - Rename fs/ntfs/attrib.h::attribute_value_length() to ntfs_attr_size().
248 - Add le{16,32,64} as well as sle{16,32,64} data types to
249 fs/ntfs/types.h.
250 - Change ntfschar to be le16 instead of u16 in fs/ntfs/types.h.
251 - Add le versions of VCN, LCN, and LSN called leVCN, leLCN, and leLSN,
252 respectively, to fs/ntfs/types.h.
253 - Update endianness conversion macros in fs/ntfs/endian.h to use the
254 new types as appropriate.
255 - Do proper type casting when using sle64_to_cpup() in fs/ntfs/dir.c
256 and index.c.
257 - Add leMFT_REF data type to fs/ntfs/layout.h.
258 - Update all NTFS header files with the new little endian data types.
259 Affected files are fs/ntfs/layout.h, logfile.h, and time.h.
260 - Do proper type casting when using ntfs_is_*_recordp() in
261 fs/ntfs/logfile.c, mft.c, and super.c.
262 - Fix all the sparse bitwise warnings. Had to change all the typedef
263 enums storing little endian values to simple enums plus a typedef for
264 the datatype to make sparse happy.
265 - Fix a bug found by the new sparse bitwise warnings where the default
266 upcase table was defined as a pointer to wchar_t rather than ntfschar
267 in fs/ntfs/ntfs.h and super.c.
268 - Change {const_,}cpu_to_le{16,32}(0) to just 0 as suggested by Al Viro.
269
2702.1.18 - Fix scheduling latencies at mount time as well as an endianness bug.
271
272 - Remove vol->nr_mft_records as it was pretty meaningless and optimize
273 the calculation of total/free inodes as used by statfs().
274 - Fix scheduling latencies in ntfs_fill_super() by dropping the BKL
275 because the code itself is using the ntfs_lock semaphore which
276 provides safe locking. (Ingo Molnar)
277 - Fix a potential bug in fs/ntfs/mft.c::map_extent_mft_record() that
278 could occur in the future for when we start closing/freeing extent
279 inodes if we don't set base_ni->ext.extent_ntfs_inos to NULL after
280 we free it.
281 - Rename {find,lookup}_attr() to ntfs_attr_{find,lookup}() as well as
282 find_external_attr() to ntfs_external_attr_find() to cleanup the
283 namespace a bit and to be more consistent with libntfs.
284 - Rename {{re,}init,get,put}_attr_search_ctx() to
285 ntfs_attr_{{re,}init,get,put}_search_ctx() as well as the type
286 attr_search_context to ntfs_attr_search_ctx.
287 - Force use of ntfs_attr_find() in ntfs_attr_lookup() when searching
288 for the attribute list attribute itself.
289 - Fix endianness bug in ntfs_external_attr_find().
290 - Change ntfs_{external_,}attr_find() to return 0 on success, -ENOENT
291 if the attribute is not found, and -EIO on real error. In the case
292 of -ENOENT, the search context is updated to describe the attribute
293 before which the attribute being searched for would need to be
294 inserted if such an action were to be desired and in the case of
295 ntfs_external_attr_find() the search context is also updated to
296 indicate the attribute list entry before which the attribute list
297 entry of the attribute being searched for would need to be inserted
298 if such an action were to be desired. Also make ntfs_find_attr()
299 static and remove its prototype from attrib.h as it is not used
300 anywhere other than attrib.c. Update ntfs_attr_lookup() and all
301 callers of ntfs_{external,}attr_{find,lookup}() for the new return
302 values.
303 - Minor cleanup of fs/ntfs/inode.c::ntfs_init_locked_inode().
304
3052.1.17 - Fix bugs in mount time error code paths and other updates.
306
307 - Implement bitmap modification code (fs/ntfs/bitmap.[hc]). This
308 includes functions to set/clear a single bit or a run of bits.
309 - Add fs/ntfs/attrib.[hc]::ntfs_find_vcn() which returns the locked
310 runlist element containing a particular vcn. It also takes care of
311 mapping any needed runlist fragments.
312 - Implement cluster (de-)allocation code (fs/ntfs/lcnalloc.[hc]).
313 - Load attribute definition table from $AttrDef at mount time.
314 - Fix bugs in mount time error code paths involving (de)allocation of
315 the default and volume upcase tables.
316 - Remove ntfs_nr_mounts as it is no longer used.
317
3182.1.16 - Implement access time updates, file sync, async io, and read/writev.
319
320 - Add support for readv/writev and aio_read/aio_write (fs/ntfs/file.c).
321 This is done by setting the appropriate file operations pointers to
322 the generic helper functions provided by mm/filemap.c.
323 - Implement fsync, fdatasync, and msync both for files (fs/ntfs/file.c)
324 and directories (fs/ntfs/dir.c).
325 - Add support for {a,m,c}time updates to inode.c::ntfs_write_inode().
326 Note, except for the root directory and any other system files opened
327 by the user, the system files will not have their access times
328 updated as they are only accessed at the inode level an hence the
329 file level functions which cause the times to be updated are never
330 invoked.
331
3322.1.15 - Invalidate quotas when (re)mounting read-write.
333
334 - Add new element itype.index.collation_rule to the ntfs inode
335 structure and set it appropriately in ntfs_read_locked_inode().
336 - Implement a new inode type "index" to allow efficient access to the
337 indices found in various system files and adapt inode handling
338 accordingly (fs/ntfs/inode.[hc]). An index inode is essentially an
339 attribute inode (NInoAttr() is true) with an attribute type of
340 AT_INDEX_ALLOCATION. As such, it is no longer allowed to call
341 ntfs_attr_iget() with an attribute type of AT_INDEX_ALLOCATION as
342 there would be no way to distinguish between normal attribute inodes
343 and index inodes. The function to obtain an index inode is
344 ntfs_index_iget() and it uses the helper function
345 ntfs_read_locked_index_inode(). Note, we do not overload
346 ntfs_attr_iget() as indices consist of multiple attributes so using
347 ntfs_attr_iget() to obtain an index inode would be confusing.
348 - Ensure that there is no overflow when doing page->index <<
349 PAGE_CACHE_SHIFT by casting page->index to s64 in fs/ntfs/aops.c.
350 - Use atomic kmap instead of kmap() in fs/ntfs/aops.c::ntfs_read_page()
351 and ntfs_read_block().
352 - Use case sensitive attribute lookups instead of case insensitive ones.
353 - Lock all page cache pages belonging to mst protected attributes while
354 accessing them to ensure we never see corrupt data while the page is
355 under writeout.
356 - Add framework for generic ntfs collation (fs/ntfs/collation.[hc]).
357 We have ntfs_is_collation_rule_supported() to check if the collation
358 rule you want to use is supported and ntfs_collation() which actually
359 collates two data items. We currently only support COLLATION_BINARY
360 and COLLATION_NTOFS_ULONG but support for other collation rules will
361 be added as the need arises.
362 - Add a new type, ntfs_index_context, to allow retrieval of an index
363 entry using the corresponding index key. To get an index context,
364 use ntfs_index_ctx_get() and to release it, use ntfs_index_ctx_put().
365 This also adds a new slab cache for the index contexts. To lookup a
366 key in an index inode, use ntfs_index_lookup(). After modifying an
367 index entry, call ntfs_index_entry_flush_dcache_page() followed by
368 ntfs_index_entry_mark_dirty() to ensure the changes are written out
369 to disk. For details see fs/ntfs/index.[hc]. Note, at present, if
370 an index entry is in the index allocation attribute rather than the
371 index root attribute it will not be written out (you will get a
372 warning message about discarded changes instead).
373 - Load the quota file ($Quota) and check if quota tracking is enabled
374 and if so, mark the quotas out of date. This causes windows to
375 rescan the volume on boot and update all quota entries.
376 - Add a set_page_dirty address space operation for ntfs_m[fs]t_aops.
377 It is simply set to __set_page_dirty_nobuffers() to make sure that
378 running set_page_dirty() on a page containing mft/ntfs records will
379 not affect the dirty state of the page buffers.
380 - Add fs/ntfs/index.c::__ntfs_index_entry_mark_dirty() which sets all
381 buffers that are inside the ntfs record in the page dirty after which
382 it sets the page dirty. This allows ->writepage to only write the
383 dirty index records rather than having to write all the records in
384 the page. Modify fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to
385 use this rather than __set_page_dirty_nobuffers().
386 - Implement fs/ntfs/aops.c::ntfs_write_mst_block() which enables the
387 writing of page cache pages belonging to mst protected attributes
388 like the index allocation attribute in directory indices and other
389 indices like $Quota/$Q, etc. This means that the quota is now marked
390 out of date on all volumes rather than only on ones where the quota
391 defaults entry is in the index root attribute of the $Quota/$Q index.
392
3932.1.14 - Fix an NFSd caused deadlock reported by several users.
394
395 - Modify fs/ntfs/ntfs_readdir() to copy the index root attribute value
396 to a buffer so that we can put the search context and unmap the mft
397 record before calling the filldir() callback. We need to do this
398 because of NFSd which calls ->lookup() from its filldir callback()
399 and this causes NTFS to deadlock as ntfs_lookup() maps the mft record
400 of the directory and since ntfs_readdir() has got it mapped already
401 ntfs_lookup() deadlocks.
402
4032.1.13 - Enable overwriting of resident files and housekeeping of system files.
404
405 - Implement writing of mft records (fs/ntfs/mft.[hc]), which includes
406 keeping the mft mirror in sync with the mft when mirrored mft records
407 are written. The functions are write_mft_record{,_nolock}(). The
408 implementation is quite rudimentary for now with lots of things not
409 implemented yet but I am not sure any of them can actually occur so
410 I will wait for people to hit each one and only then implement it.
411 - Commit open system inodes at umount time. This should make it
412 virtually impossible for sync_mft_mirror_umount() to ever be needed.
413 - Implement ->write_inode (fs/ntfs/inode.c::ntfs_write_inode()) for the
414 ntfs super operations. This gives us inode writing via the VFS inode
415 dirty code paths. Note: Access time updates are not implemented yet.
416 - Implement fs/ntfs/mft.[hc]::{,__}mark_mft_record_dirty() and make
417 fs/ntfs/aops.c::ntfs_writepage() and ntfs_commit_write() use it, thus
418 finally enabling resident file overwrite! (-8 This also includes a
419 placeholder for ->writepage (ntfs_mft_writepage()), which for now
420 just redirties the page and returns. Also, at umount time, we for
421 now throw away all mft data page cache pages after the last call to
422 ntfs_commit_inode() in the hope that all inodes will have been
423 written out by then and hence no dirty (meta)data will be lost. We
424 also check for this case and emit an error message telling the user
425 to run chkdsk.
426 - Use set_page_writeback() and end_page_writeback() in the resident
427 attribute code path of fs/ntfs/aops.c::ntfs_writepage() otherwise
428 the radix-tree tag PAGECACHE_TAG_DIRTY remains set even though the
429 page is clean.
430 - Implement ntfs_mft_writepage() so it now checks if any of the mft
431 records in the page are dirty and if so redirties the page and
432 returns. Otherwise it just returns (after doing set_page_writeback(),
433 unlock_page(), end_page_writeback() or the radix-tree tag
434 PAGECACHE_TAG_DIRTY remains set even though the page is clean), thus
435 alowing the VM to do with the page as it pleases. Also, at umount
436 time, now only throw away dirty mft (meta)data pages if dirty inodes
437 are present and ask the user to email us if they see this happening.
438 - Add functions ntfs_{clear,set}_volume_flags(), to modify the volume
439 information flags (fs/ntfs/super.c).
440 - Mark the volume dirty when (re)mounting read-write and mark it clean
441 when unmounting or remounting read-only. If any volume errors are
442 found, the volume is left marked dirty to force chkdsk to run.
443 - Add code to set the NT4 compatibility flag when (re)mounting
444 read-write for newer NTFS versions but leave it commented out for now
445 since we do not make any modifications that are NTFS 1.2 specific yet
446 and since setting this flag breaks Captive-NTFS which is not nice.
447 This code must be enabled once we start writing NTFS 1.2 specific
448 changes otherwise Windows NTFS driver might crash / cause corruption.
449
4502.1.12 - Fix the second fix to the decompression engine and some cleanups.
451
452 - Add a new address space operations struct, ntfs_mst_aops, for mst
453 protected attributes. This is because the default ntfs_aops do not
454 make sense with mst protected data and were they to write anything to
455 such an attribute they would cause data corruption so we provide
456 ntfs_mst_aops which does not have any write related operations set.
457 - Cleanup dirty ntfs inode handling (fs/ntfs/inode.[hc]) which also
458 includes an adapted ntfs_commit_inode() and an implementation of
459 ntfs_write_inode() which for now just cleans dirty inodes without
460 writing them (it does emit a warning that this is happening).
461 - Undo the second decompression engine fix (see 2.1.9 release ChangeLog
462 entry) as it was only fixing a theoretical bug but at the same time
463 it badly broke the handling of sparse and uncompressed compression
464 blocks.
465
4662.1.11 - Driver internal cleanups.
467
468 - Only build logfile.o if building the driver with read-write support.
469 - Really final white space cleanups.
470 - Use generic_ffs() instead of ffs() in logfile.c which allows the
471 log_page_size variable to be optimized by gcc into a constant.
472 - Rename uchar_t to ntfschar everywhere as uchar_t is unsigned 1-byte
473 char as defined by POSIX and as found on some systems.
474
4752.1.10 - Force read-only (re)mounting of volumes with unsupported volume flags.
476
477 - Finish off the white space cleanups (remove trailing spaces, etc).
478 - Clean up ntfs_fill_super() and ntfs_read_inode_mount() by removing
479 the kludges around the first iget(). Instead of (re)setting ->s_op
480 we have the $MFT inode set up by explicit new_inode() / set ->i_ino /
481 insert_inode_hash() / call ntfs_read_inode_mount() directly. This
482 kills the need for second super_operations and allows to return error
483 from ntfs_read_inode_mount() without resorting to ugly "poisoning"
484 tricks. (Al Viro)
485 - Force read-only (re)mounting if any of the following bits are set in
486 the volume information flags:
487 VOLUME_IS_DIRTY, VOLUME_RESIZE_LOG_FILE,
488 VOLUME_UPGRADE_ON_MOUNT, VOLUME_DELETE_USN_UNDERWAY,
489 VOLUME_REPAIR_OBJECT_ID, VOLUME_MODIFIED_BY_CHKDSK
490 To make this easier we define VOLUME_MUST_MOUNT_RO_MASK with all the
491 above bits set so the test is made easy.
492
4932.1.9 - Fix two bugs in decompression engine.
494
495 - Fix a bug where we would not always detect that we have reached the
496 end of a compression block because we were ending at minus one byte
497 which is effectively the same as being at the end. The fix is to
498 check whether the uncompressed buffer has been fully filled and if so
499 we assume we have reached the end of the compression block. A big
500 thank you to Marcin Gibuła for the bug report, the assistance in
501 tracking down the bug and testing the fix.
502 - Fix a possible bug where when a compressed read is truncated to the
503 end of the file, the offset inside the last page was not truncated.
504
5052.1.8 - Handle $MFT mirror and $LogFile, improve time handling, and cleanups.
506
507 - Use get_bh() instead of manual atomic_inc() in fs/ntfs/compress.c.
508 - Modify fs/ntfs/time.c::ntfs2utc(), get_current_ntfs_time(), and
509 utc2ntfs() to work with struct timespec instead of time_t on the
510 Linux UTC time side thus preserving the full precision of the NTFS
511 time and only loosing up to 99 nano-seconds in the Linux UTC time.
512 - Move fs/ntfs/time.c to fs/ntfs/time.h and make the time functions
513 static inline.
514 - Remove unused ntfs_dirty_inode().
515 - Cleanup super operations declaration in fs/ntfs/super.c.
516 - Wrap flush_dcache_mft_record_page() in #ifdef NTFS_RW.
517 - Add NInoTestSetFoo() and NInoTestClearFoo() macro magic to
518 fs/ntfs/inode.h and use it to declare NInoTest{Set,Clear}Dirty.
519 - Move typedefs for ntfs_attr and test_t from fs/ntfs/inode.c to
520 fs/ntfs/inode.h so they can be used elsewhere.
521 - Determine the mft mirror size as the number of mirrored mft records
522 and store it in ntfs_volume->mftmirr_size (fs/ntfs/super.c).
523 - Load the mft mirror at mount time and compare the mft records stored
524 in it to the ones in the mft. Force a read-only mount if the two do
525 not match (fs/ntfs/super.c).
526 - Fix type casting related warnings on 64-bit architectures. Thanks
527 to Meelis Roos for reporting them.
528 - Move %L to %ll as %L is floating point and %ll is integer which is
529 what we want.
530 - Read the journal ($LogFile) and determine if the volume has been
531 shutdown cleanly and force a read-only mount if not (fs/ntfs/super.c
532 and fs/ntfs/logfile.c). This is a little bit of a crude check in
533 that we only look at the restart areas and not at the actual log
534 records so that there will be a very small number of cases where we
535 think that a volume is dirty when in fact it is clean. This should
536 only affect volumes that have not been shutdown cleanly and did not
537 have any pending, non-check-pointed i/o.
538 - If the $LogFile indicates a clean shutdown and a read-write (re)mount
539 is requested, empty $LogFile by overwriting it with 0xff bytes to
540 ensure that Windows cannot cause data corruption by replaying a stale
541 journal after Linux has written to the volume.
542
5432.1.7 - Enable NFS exporting of mounted NTFS volumes.
544
545 - Set i_generation in the VFS inode from the seq_no of the NTFS inode.
546 - Make ntfs_lookup() NFS export safe, i.e. use d_splice_alias(), etc.
547 - Implement ->get_dentry() in fs/ntfs/namei.c::ntfs_get_dentry() as the
548 default doesn't allow inode number 0 which is a valid inode on NTFS
549 and even if it did allow that it uses iget() instead of ntfs_iget()
550 which makes it useless for us.
551 - Implement ->get_parent() in fs/ntfs/namei.c::ntfs_get_parent() as the
552 default just returns -EACCES which is not very useful.
553 - Define export operations (->s_export_op) for NTFS (ntfs_export_ops)
554 and set them up in the super block at mount time (super.c) this
555 allows mounted NTFS volumes to be exported via NFS.
556 - Add missing return -EOPNOTSUPP; in
557 fs/ntfs/aops.c::ntfs_commit_nonresident_write().
558 - Enforce no atime and no dir atime updates at mount/remount time as
559 they are not implemented yet anyway.
560 - Move a few assignments in fs/ntfs/attrib.c::load_attribute_list() to
561 after a NULL check. Thanks to Dave Jones for pointing this out.
562
5632.1.6 - Fix minor bug in handling of compressed directories.
564
565 - Fix bug in handling of compressed directories. A compressed
566 directory is not really compressed so when we set the ->i_blocks
567 field of a compressed directory inode we were setting it from the
568 non-existing field ni->itype.compressed.size which gave random
569 results... For directories we now always use ni->allocated_size.
570
5712.1.5 - Fix minor bug in attribute list attribute handling.
572
573 - Fix bug in attribute list handling. Actually it is not as much a bug
574 as too much protection in that we were not allowing attribute lists
575 which waste space on disk while Windows XP clearly allows it and in
576 fact creates such attribute lists so our driver was failing.
577 - Update NTFS documentation ready for 2.6 kernel release.
578
5792.1.4 - Reduce compiler requirements.
580
581 - Remove all uses of unnamed structs and unions in the driver to make
582 old and newer gcc versions happy. Makes it a bit uglier IMO but at
583 least people will stop hassling me about it.
584
5852.1.3 - Important bug fixes in corner cases.
586
587 - super.c::parse_ntfs_boot_sector(): Correct the check for 64-bit
588 clusters. (Philipp Thomas)
589 - attrib.c::load_attribute_list(): Fix bug when initialized_size is a
590 multiple of the block_size but not the cluster size. (Szabolcs
591 Szakacsits <szaka@sienet.hu>)
592
5932.1.2 - Important bug fixes aleviating the hangs in statfs.
594
595 - Fix buggy free cluster and free inode determination logic.
596
5972.1.1 - Minor updates.
598
599 - Add handling for initialized_size != data_size in compressed files.
600 - Reduce function local stack usage from 0x3d4 bytes to just noise in
601 fs/ntfs/upcase.c. (Randy Dunlap <rddunlap@osdl.ord>)
602 - Remove compiler warnings for newer gcc.
603 - Pages are no longer kmapped by mm/filemap.c::generic_file_write()
604 around calls to ->{prepare,commit}_write. Adapt NTFS appropriately
605 in fs/ntfs/aops.c::ntfs_prepare_nonresident_write() by using
606 kmap_atomic(KM_USER0).
607
6082.1.0 - First steps towards write support: implement file overwrite.
609
610 - Add configuration option for developmental write support with an
611 appropriately scary configuration help text.
612 - Initial implementation of fs/ntfs/aops.c::ntfs_writepage() and its
613 helper fs/ntfs/aops.c::ntfs_write_block(). This enables mmap(2) based
614 overwriting of existing files on ntfs. Note: Resident files are
615 only written into memory, and not written out to disk at present, so
616 avoid writing to files smaller than about 1kiB.
617 - Initial implementation of fs/ntfs/aops.c::ntfs_prepare_write(), its
618 helper fs/ntfs/aops.c::ntfs_prepare_nonresident_write() and their
619 counterparts, fs/ntfs/aops.c::ntfs_commit_write(), and
620 fs/ntfs/aops.c::ntfs_commit_nonresident_write(), respectively. Also,
621 add generic_file_write() to the ntfs file operations (fs/ntfs/file.c).
622 This enables write(2) based overwriting of existing files on ntfs.
623 Note: As with mmap(2) based overwriting, resident files are only
624 written into memory, and not written out to disk at present, so avoid
625 writing to files smaller than about 1kiB.
626 - Implement ->truncate (fs/ntfs/inode.c::ntfs_truncate()) and
627 ->setattr() (fs/ntfs/inode.c::ntfs_setattr()) inode operations for
628 files with the purpose of intercepting and aborting all i_size
629 changes which we do not support yet. ntfs_truncate() actually only
630 emits a warning message but AFAICS our interception of i_size changes
631 elsewhere means ntfs_truncate() never gets called for i_size changes.
632 It is only called from generic_file_write() when we fail in
633 ntfs_prepare_{,nonresident_}write() in order to discard any
634 instantiated buffers beyond i_size. Thus i_size is not actually
635 changed so our warning message is enough. Unfortunately it is not
636 possible to easily determine if i_size is being changed or not hence
637 we just emit an appropriately worded error message.
638
6392.0.25 - Small bug fixes and cleanups.
640
641 - Unlock the page in an out of memory error code path in
642 fs/ntfs/aops.c::ntfs_read_block().
643 - If fs/ntfs/aops.c::ntfs_read_page() is called on an uptodate page,
644 just unlock the page and return. (This can happen due to ->writepage
645 clearing PageUptodate() during write out of MstProtected()
646 attributes.
647 - Remove leaked write code again.
648
6492.0.24 - Cleanups.
650
651 - Treat BUG_ON() as ASSERT() not VERIFY(), i.e. do not use side effects
652 inside BUG_ON(). (Adam J. Richter)
653 - Split logical OR expressions inside BUG_ON() into individual BUG_ON()
654 calls for improved debugging. (Adam J. Richter)
655 - Add errors flag to the ntfs volume state, accessed via
656 NVol{,Set,Clear}Errors(vol).
657 - Do not allow read-write remounts of read-only volumes with errors.
658 - Clarify comment for ntfs file operation sendfile which was added by
659 Christoph Hellwig a while ago (just using generic_file_sendfile())
660 to say that ntfs ->sendfile is only used for the case where the
661 source data is on the ntfs partition and the destination is
662 somewhere else, i.e. nothing we need to concern ourselves with.
663 - Add generic_file_write() as our ntfs file write operation.
664
6652.0.23 - Major bug fixes (races, deadlocks, non-i386 architectures).
666
667 - Massive internal locking changes to mft record locking. Fixes lock
668 recursion and replaces the mrec_lock read/write semaphore with a
669 mutex. Also removes the now superfluous mft_count. This fixes several
670 race conditions and deadlocks, especially in the future write code.
671 - Fix ntfs over loopback for compressed files by adding an
672 optimization barrier. (gcc was screwing up otherwise ?)
673 - Miscellaneous cleanups all over the code and a fix or two in error
674 handling code paths.
675 Thanks go to Christoph Hellwig for pointing out the following two:
676 - Remove now unused function fs/ntfs/malloc.h::vmalloc_nofs().
677 - Fix ntfs_free() for ia64 and parisc by checking for VMALLOC_END, too.
678
6792.0.22 - Cleanups, mainly to ntfs_readdir(), and use C99 initializers.
680
681 - Change fs/ntfs/dir.c::ntfs_reddir() to only read/write ->f_pos once
682 at entry/exit respectively.
683 - Use C99 initializers for structures.
684 - Remove unused variable blocks from fs/ntfs/aops.c::ntfs_read_block().
685
6862.0.21 - Check for, and refuse to work with too large files/directories/volumes.
687
688 - Limit volume size at mount time to 2TiB on architectures where
689 unsigned long is 32-bits (fs/ntfs/super.c::parse_ntfs_boot_sector()).
690 This is the most we can do without overflowing the 32-bit limit of
691 the block device size imposed on us by sb_bread() and sb_getblk()
692 for the time being.
693 - Limit file/directory size at open() time to 16TiB on architectures
694 where unsigned long is 32-bits (fs/ntfs/file.c::ntfs_file_open() and
695 fs/ntfs/dir.c::ntfs_dir_open()). This is the most we can do without
696 overflowing the page cache page index.
697
6982.0.20 - Support non-resident directory index bitmaps, fix page leak in readdir.
699
700 - Move the directory index bitmap to use an attribute inode instead of
701 having special fields for it inside the ntfs inode structure. This
702 means that the index bitmaps now use the page cache for i/o, too,
703 and also as a side effect we get support for non-resident index
704 bitmaps for free.
705 - Simplify/cleanup error handling in fs/ntfs/dir.c::ntfs_readdir() and
706 fix a page leak that manifested itself in some cases.
707 - Add fs/ntfs/inode.c::ntfs_put_inode(), which we need to release the
708 index bitmap inode on the final iput().
709
7102.0.19 - Fix race condition, improvements, and optimizations in i/o interface.
711
712 - Apply block optimization added to fs/ntfs/aops.c::ntfs_read_block()
713 to fs/ntfs/compress.c::ntfs_file_read_compressed_block() as well.
714 - Drop the "file" from ntfs_file_read_compressed_block().
715 - Rename fs/ntfs/aops.c::ntfs_enb_buffer_read_async() to
716 ntfs_end_buffer_async_read() (more like the fs/buffer.c counterpart).
717 - Update ntfs_end_buffer_async_read() with the improved logic from
718 its updated counterpart fs/buffer.c::end_buffer_async_read(). Apply
719 further logic improvements to better determine when we set PageError.
720 - Update submission of buffers in fs/ntfs/aops.c::ntfs_read_block() to
721 check for the buffers being uptodate first in line with the updated
722 fs/buffer.c::block_read_full_page(). This plugs a small race
723 condition.
724
7252.0.18 - Fix race condition in reading of compressed files.
726
727 - There was a narrow window between checking a buffer head for being
728 uptodate and locking it in ntfs_file_read_compressed_block(). We now
729 lock the buffer and then check whether it is uptodate or not.
730
7312.0.17 - Cleanups and optimizations - shrinking the ToDo list.
732
733 - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to return an error
734 code and update callers, i.e. ntfs_iget(), to pass that error code
735 up instead of just using -EIO.
736 - Modifications to super.c to ensure that both mount and remount
737 cannot set any write related options when the driver is compiled
738 read-only.
739 - Optimize block resolution in fs/ntfs/aops.c::ntfs_read_block() to
740 cache the current runlist element. This should improve performance
741 when reading very large and/or very fragmented data.
742
7432.0.16 - Convert access to $MFT/$BITMAP to attribute inode API.
744
745 - Fix a stupid bug introduced in 2.0.15 where we were unmapping the
746 wrong inode in fs/ntfs/inode.c::ntfs_attr_iget().
747 - Fix debugging check in fs/ntfs/aops.c::ntfs_read_block().
748 - Convert $MFT/$BITMAP access to attribute inode API and remove all
749 remnants of the ugly mftbmp address space and operations hack. This
750 means we finally have only one readpage function as well as only one
751 async io completion handler. Yey! The mft bitmap is now just an
752 attribute inode and is accessed from vol->mftbmp_ino just as if it
753 were a normal file. Fake inodes rule. (-:
754
7552.0.15 - Fake inodes based attribute i/o via the pagecache, fixes and cleanups.
756
757 - Fix silly bug in fs/ntfs/super.c::parse_options() which was causing
758 remounts to fail when the partition had an entry in /etc/fstab and
759 the entry specified the nls= option.
760 - Apply same macro magic used in fs/ntfs/inode.h to fs/ntfs/volume.h to
761 expand all the helper functions NVolFoo(), NVolSetFoo(), and
762 NVolClearFoo().
763 - Move copyright statement from driver initialisation message to
764 module description (fs/super.c). This makes the initialisation
765 message fit on one line and fits in better with rest of kernel.
766 - Update fs/ntfs/attrib.c::map_run_list() to work on both real and
767 attribute inodes, and both for files and directories.
768 - Implement fake attribute inodes allowing all attribute i/o to go via
769 the page cache and to use all the normal vfs/mm functionality:
770 - Add ntfs_attr_iget() and its helper ntfs_read_locked_attr_inode()
771 to fs/ntfs/inode.c.
772 - Add needed cleanup code to ntfs_clear_big_inode().
773 - Merge address space operations for files and directories (aops.c),
774 now just have ntfs_aops:
775 - Rename:
776 end_buffer_read_attr_async() -> ntfs_end_buffer_read_async(),
777 ntfs_attr_read_block() -> ntfs_read_block(),
778 ntfs_file_read_page() -> ntfs_readpage().
779 - Rewrite fs/ntfs/aops.c::ntfs_readpage() to work on both real and
780 attribute inodes, and both for files and directories.
781 - Remove obsolete fs/ntfs/aops.c::ntfs_mst_readpage().
782
7832.0.14 - Run list merging code cleanup, minor locking changes, typo fixes.
784
785 - Change fs/ntfs/super.c::ntfs_statfs() to not rely on BKL by moving
786 the locking out of super.c::get_nr_free_mft_records() and taking and
787 dropping the mftbmp_lock rw_semaphore in ntfs_statfs() itself.
788 - Bring attribute runlist merging code (fs/ntfs/attrib.c) in sync with
789 current userspace ntfs library code. This means that if a merge
790 fails the original runlists are always left unmodified instead of
791 being silently corrupted.
792 - Misc typo fixes.
793
7942.0.13 - Use iget5_locked() in preparation for fake inodes and small cleanups.
795
796 - Remove nr_mft_bits and the now superfluous union with nr_mft_records
797 from ntfs_volume structure.
798 - Remove nr_lcn_bits and the now superfluous union with nr_clusters
799 from ntfs_volume structure.
800 - Use iget5_locked() and friends instead of conventional iget(). Wrap
801 the call in fs/ntfs/inode.c::ntfs_iget() and update callers of iget()
802 to use ntfs_iget(). Leave only one iget() call at mount time so we
803 don't need an ntfs_iget_mount().
804 - Change fs/ntfs/inode.c::ntfs_new_extent_inode() to take mft_no as an
805 additional argument.
806
8072.0.12 - Initial cleanup of address space operations following 2.0.11 changes.
808
809 - Merge fs/ntfs/aops.c::end_buffer_read_mst_async() and
810 fs/ntfs/aops.c::end_buffer_read_file_async() into one function
811 fs/ntfs/aops.c::end_buffer_read_attr_async() using NInoMstProtected()
812 to determine whether to apply mst fixups or not.
813 - Above change allows merging fs/ntfs/aops.c::ntfs_file_read_block()
814 and fs/ntfs/aops.c::ntfs_mst_readpage() into one function
815 fs/ntfs/aops.c::ntfs_attr_read_block(). Also, create a tiny wrapper
816 fs/ntfs/aops.c::ntfs_mst_readpage() to transform the parameters from
817 the VFS readpage function prototype to the ntfs_attr_read_block()
818 function prototype.
819
8202.0.11 - Initial preparations for fake inode based attribute i/o.
821
822 - Move definition of ntfs_inode_state_bits to fs/ntfs/inode.h and
823 do some macro magic (adapted from include/linux/buffer_head.h) to
824 expand all the helper functions NInoFoo(), NInoSetFoo(), and
825 NInoClearFoo().
826 - Add new flag to ntfs_inode_state_bits: NI_Sparse.
827 - Add new fields to ntfs_inode structure to allow use of fake inodes
828 for attribute i/o: type, name, name_len. Also add new state bits:
829 NI_Attr, which, if set, indicates the inode is a fake inode, and
830 NI_MstProtected, which, if set, indicates the attribute uses multi
831 sector transfer protection, i.e. fixups need to be applied after
832 reads and before/after writes.
833 - Rename fs/ntfs/inode.c::ntfs_{new,clear,destroy}_inode() to
834 ntfs_{new,clear,destroy}_extent_inode() and update callers.
835 - Use ntfs_clear_extent_inode() in fs/ntfs/inode.c::__ntfs_clear_inode()
836 instead of ntfs_destroy_extent_inode().
837 - Cleanup memory deallocations in {__,}ntfs_clear_{,big_}inode().
838 - Make all operations on ntfs inode state bits use the NIno* functions.
839 - Set up the new ntfs inode fields and state bits in
840 fs/ntfs/inode.c::ntfs_read_inode() and add appropriate cleanup of
841 allocated memory to __ntfs_clear_inode().
842 - Cleanup ntfs_inode structure a bit for better ordering of elements
843 w.r.t. their size to allow better packing of the structure in memory.
844
8452.0.10 - There can only be 2^32 - 1 inodes on an NTFS volume.
846
847 - Add check at mount time to verify that the number of inodes on the
848 volume does not exceed 2^32 - 1, which is the maximum allowed for
849 NTFS according to Microsoft.
850 - Change mft_no member of ntfs_inode structure to be unsigned long.
851 Update all users. This makes ntfs_inode->mft_no just a copy of struct
852 inode->i_ino. But we can't just always use struct inode->i_ino and
853 remove mft_no because extent inodes do not have an attached struct
854 inode.
855
8562.0.9 - Decompression engine now uses a single buffer and other cleanups.
857
858 - Change decompression engine to use a single buffer protected by a
859 spin lock instead of per-CPU buffers. (Rusty Russell)
860 - Do not update cb_pos when handling a partial final page during
861 decompression of a sparse compression block, as the value is later
862 reset without being read/used. (Rusty Russell)
863 - Switch to using the new KM_BIO_SRC_IRQ for atomic kmap()s. (Andrew
864 Morton)
865 - Change buffer size in ntfs_readdir()/ntfs_filldir() to use
866 NLS_MAX_CHARSET_SIZE which makes the buffers almost 1kiB each but
867 it also makes everything safer so it is a good thing.
868 - Miscellaneous minor cleanups to comments.
869
8702.0.8 - Major updates for handling of case sensitivity and dcache aliasing.
871
872 Big thanks go to Al Viro and other inhabitants of #kernel for investing
873 their time to discuss the case sensitivity and dcache aliasing issues.
874
875 - Remove unused source file fs/ntfs/attraops.c.
876 - Remove show_inodes mount option(s), thus dropping support for
877 displaying of short file names.
878 - Remove deprecated mount option posix.
879 - Restore show_sys_files mount option.
880 - Add new mount option case_sensitive, to determine if the driver
881 treats file names as case sensitive or not. If case sensitive, create
882 file names in the POSIX namespace. Otherwise create file names in the
883 LONG/WIN32 namespace. Note, files remain accessible via their short
884 file name, if it exists.
885 - Remove really dumb logic bug in boot sector recovery code.
886 - Fix dcache aliasing issues wrt short/long file names via changes
887 to fs/ntfs/dir.c::ntfs_lookup_inode_by_name() and
888 fs/ntfs/namei.c::ntfs_lookup():
889 - Add additional argument to ntfs_lookup_inode_by_name() in which we
890 return information about the matching file name if the case is not
891 matching or the match is a short file name. See comments above the
892 function definition for details.
893 - Change ntfs_lookup() to only create dcache entries for the correctly
894 cased file name and only for the WIN32 namespace counterpart of DOS
895 namespace file names. This ensures we have only one dentry per
896 directory and also removes all dcache aliasing issues between short
897 and long file names once we add write support. See comments above
898 function for details.
899 - Fix potential 1 byte overflow in fs/ntfs/unistr.c::ntfs_ucstonls().
900
9012.0.7 - Minor cleanups and updates for changes in core kernel code.
902
903 - Remove much of the NULL struct element initializers.
904 - Various updates to make compatible with recent kernels.
905 - Remove defines of MAX_BUF_PER_PAGE and include linux/buffer_head.h
906 in fs/ntfs/ntfs.h instead.
907 - Remove no longer needed KERNEL_VERSION checks. We are now in the
908 kernel proper so they are no longer needed.
909
9102.0.6 - Major bugfix to make compatible with other kernel changes.
911
912 - Initialize the mftbmp address space properly now that there are more
913 fields in the struct address_space. This was leading to hangs and
914 oopses on umount since 2.5.12 because of changes to other parts of
915 the kernel. We probably want a kernel generic init_address_space()
916 function...
917 - Drop BKL from ntfs_readdir() after consultation with Al Viro. The
918 only caller of ->readdir() is vfs_readdir() which holds i_sem during
919 the call, and i_sem is sufficient protection against changes in the
920 directory inode (including ->i_size).
921 - Use generic_file_llseek() for directories (as opposed to
922 default_llseek()) as this downs i_sem instead of the BKL which is
923 what we now need for exclusion against ->f_pos changes considering we
924 no longer take the BKL in ntfs_readdir().
925
9262.0.5 - Major bugfix. Buffer overflow in extent inode handling.
927
928 - No need to set old blocksize in super.c::ntfs_fill_super() as the
929 VFS does so via invocation of deactivate_super() calling
930 fs->fill_super() calling block_kill_super() which does it.
931 - BKL moved from VFS into dir.c::ntfs_readdir(). (Linus Torvalds)
932 -> Do we really need it? I don't think so as we have exclusion on
933 the directory ntfs_inode rw_semaphore mrec_lock. We mmight have to
934 move the ->f_pos accesses under the mrec_lock though. Check this...
935 - Fix really, really, really stupid buffer overflow in extent inode
936 handling in mft.c::map_extent_mft_record().
937
9382.0.4 - Cleanups and updates for kernel 2.5.11.
939
940 - Add documentation on how to use the MD driver to be able to use NTFS
941 stripe and volume sets in Linux and generally cleanup documentation
942 a bit.
943 Remove all uses of kdev_t in favour of struct block_device *:
944 - Change compress.c::ntfs_file_read_compressed_block() to use
945 sb_getblk() instead of getblk().
946 - Change super.c::ntfs_fill_super() to use bdev_hardsect_size() instead
947 of get_hardsect_size().
948 - No need to get old blocksize in super.c::ntfs_fill_super() as
949 fs/super.c::get_sb_bdev() already does this.
950 - Set bh->b_bdev instead of bh->b_dev throughout aops.c.
951
9522.0.3 - Small bug fixes, cleanups, and performance improvements.
953
954 - Remove some dead code from mft.c.
955 - Optimize readpage and read_block functions throughout aops.c so that
956 only initialized blocks are read. Non-initialized ones have their
957 buffer head mapped, zeroed, and set up to date, without scheduling
958 any i/o. Thanks to Al Viro for advice on how to avoid the device i/o.
959 Thanks go to Andrew Morton for spotting the below:
960 - Fix buglet in allocate_compression_buffers() error code path.
961 - Call flush_dcache_page() after modifying page cache page contents in
962 ntfs_file_readpage().
963 - Check for existence of page buffers throughout aops.c before calling
964 create_empty_buffers(). This happens when an I/O error occurs and the
965 read is retried. (It also happens once writing is implemented so that
966 needed doing anyway but I had left it for later...)
967 - Don't BUG_ON() uptodate and/or mapped buffers throughout aops.c in
968 readpage and read_block functions. Reasoning same as above (i.e. I/O
969 error retries and future write code paths.)
970
9712.0.2 - Minor updates and cleanups.
972
973 - Cleanup: rename mst.c::__post_read_mst_fixup to post_write_mst_fixup
974 and cleanup the code a bit, removing the unused size parameter.
975 - Change default fmask to 0177 and update documentation.
976 - Change attrib.c::get_attr_search_ctx() to return the search context
977 directly instead of taking the address of a pointer. A return value
978 of NULL means the allocation failed. Updated all callers
979 appropriately.
980 - Update to 2.5.9 kernel (preserving backwards compatibility) by
981 replacing all occurences of page->buffers with page_buffers(page).
982 - Fix minor bugs in runlist merging, also minor cleanup.
983 - Updates to bootsector layout and mft mirror contents descriptions.
984 - Small bug fix in error detection in unistr.c and some cleanups.
985 - Grow name buffer allocations in unistr.c in aligned mutlipled of 64
986 bytes.
987
9882.0.1 - Minor updates.
989
990 - Make default umask correspond to documentation.
991 - Improve documentation.
992 - Set default mode to include execute bit. The {u,f,d}mask can be used
993 to take it away if desired. This allows binaries to be executed from
994 a mounted ntfs partition.
995
9962.0.0 - New version number. Remove TNG from the name. Now in the kernel.
997
998 - Add kill_super, just keeping up with the vfs changes in the kernel.
999 - Repeat some changes from tng-0.0.8 that somehow got lost on the way
1000 from the CVS import into BitKeeper.
1001 - Begin to implement proper handling of allocated_size vs
1002 initialized_size vs data_size (i.e. i_size). Done are
1003 mft.c::ntfs_mft_readpage(), aops.c::end_buffer_read_index_async(),
1004 and attrib.c::load_attribute_list().
1005 - Lock the runlist in attrib.c::load_attribute_list() while using it.
1006 - Fix memory leak in ntfs_file_read_compressed_block() and generally
1007 clean up compress.c a little, removing some uncommented/unused debug
1008 code.
1009 - Tidy up dir.c a little bit.
1010 - Don't bother getting the runlist in inode.c::ntfs_read_inode().
1011 - Merge mft.c::ntfs_mft_readpage() and aops.c::ntfs_index_readpage()
1012 creating aops.c::ntfs_mst_readpage(), improving the handling of
1013 holes and overflow in the process and implementing the correct
1014 equivalent of ntfs_file_get_block() in ntfs_mst_readpage() itself.
1015 I am aiming for correctness at the moment. Modularisation can come
1016 later.
1017 - Rename aops.c::end_buffer_read_index_async() to
1018 end_buffer_read_mst_async() and optimize the overflow checking and
1019 handling.
1020 - Use the host of the mftbmp address space mapping to hold the ntfs
1021 volume. This is needed so the async i/o completion handler can
1022 retrieve a pointer to the volume. Hopefully this will not cause
1023 problems elsewhere in the kernel... Otherwise will need to use a
1024 fake inode.
1025 - Complete implementation of proper handling of allocated_size vs
1026 initialized_size vs data_size (i.e. i_size) in whole driver.
1027 Basically aops.c is now completely rewritten.
1028 - Change NTFS driver name to just NTFS and set version number to 2.0.0
1029 to make a clear distinction from the old driver which is still on
1030 version 1.1.22.
1031
1032tng-0.0.8 - 08/03/2002 - Now using BitKeeper, http://linux-ntfs.bkbits.net/
1033
1034 - Replace bdevname(sb->s_dev) with sb->s_id.
1035 - Remove now superfluous new-line characters in all callers of
1036 ntfs_debug().
1037 - Apply kludge in ntfs_read_inode(), setting i_nlink to 1 for
1038 directories. Without this the "find" utility gets very upset which is
1039 fair enough as Linux/Unix do not support directory hard links.
1040 - Further runlist merging work. (Richard Russon)
1041 - Backwards compatibility for gcc-2.95. (Richard Russon)
1042 - Update to kernel 2.5.5-pre1 and rediff the now tiny patch.
1043 - Convert to new file system declaration using ->ntfs_get_sb() and
1044 replacing ntfs_read_super() with ntfs_fill_super().
1045 - Set s_maxbytes to MAX_LFS_FILESIZE to avoid page cache page index
1046 overflow on 32-bit architectures.
1047 - Cleanup upcase loading code to use ntfs_(un)map_page().
1048 - Disable/reenable preemtion in critical sections of compession engine.
1049 - Replace device size determination in ntfs_fill_super() with
1050 sb->s_bdev->bd_inode->i_size (in bytes) and remove now superfluous
1051 function super.c::get_nr_blocks().
1052 - Implement a mount time option (show_inodes) allowing choice of which
1053 types of inode names readdir() returns and modify ntfs_filldir()
1054 accordingly. There are several parameters to show_inodes:
1055 system: system files
1056 win32: long file names (including POSIX file names) [DEFAULT]
1057 long: same as win32
1058 dos: short file names only (excluding POSIX file names)
1059 short: same as dos
1060 posix: same as both win32 and dos
1061 all: all file names
1062 Note that the options are additive, i.e. specifying:
1063 -o show_inodes=system,show_inodes=win32,show_inodes=dos
1064 is the same as specifying:
1065 -o show_inodes=all
1066 Note that the "posix" and "all" options will show all directory
1067 names, BUT the link count on each directory inode entry is set to 1,
1068 due to Linux not supporting directory hard links. This may well
1069 confuse some userspace applications, since the directory names will
1070 have the same inode numbers. Thus it is NOT advisable to use the
1071 "posix" or "all" options. We provide them only for completeness sake.
1072 - Add copies of allocated_size, initialized_size, and compressed_size to
1073 the ntfs inode structure and set them up in
1074 inode.c::ntfs_read_inode(). These reflect the unnamed data attribute
1075 for files and the index allocation attribute for directories.
1076 - Add copies of allocated_size and initialized_size to ntfs inode for
1077 $BITMAP attribute of large directories and set them up in
1078 inode.c::ntfs_read_inode().
1079 - Add copies of allocated_size and initialized_size to ntfs volume for
1080 $BITMAP attribute of $MFT and set them up in
1081 super.c::load_system_files().
1082 - Parse deprecated ntfs driver options (iocharset, show_sys_files,
1083 posix, and utf8) and tell user what the new options to use are. Note
1084 we still do support them but they will be removed with kernel 2.7.x.
1085 - Change all occurences of integer long long printf formatting to hex
1086 as printk() will not support long long integer format if/when the
1087 div64 patch goes into the kernel.
1088 - Make slab caches have stable names and change the names to what they
1089 were intended to be. These changes are required/made possible by the
1090 new slab cache name handling which removes the length limitation by
1091 requiring the caller of kmem_cache_create() to supply a stable name
1092 which is then referenced but not copied.
1093 - Rename run_list structure to run_list_element and create a new
1094 run_list structure containing a pointer to a run_list_element
1095 structure and a read/write semaphore. Adapt all users of runlists
1096 to new scheme and take and release the lock as needed. This fixes a
1097 nasty race as the run_list changes even when inodes are locked for
1098 reading and even when the inode isn't locked at all, so we really
1099 needed the serialization. We use a semaphore rather than a spinlock
1100 as memory allocations can sleep and doing everything GFP_ATOMIC
1101 would be silly.
1102 - Cleanup read_inode() removing all code checking for lowest_vcn != 0.
1103 This can never happen due to the nature of lookup_attr() and how we
1104 support attribute lists. If it did happen it would imply the inode
1105 being corrupt.
1106 - Check for lowest_vcn != 0 in ntfs_read_inode() and mark the inode as
1107 bad if found.
1108 - Update to 2.5.6-pre2 changes in struct address_space.
1109 - Use parent_ino() when accessing d_parent inode number in dir.c.
1110 - Import Sourceforge CVS repository into BitKeeper repository:
1111 http://linux-ntfs.bkbits.net/ntfs-tng-2.5
1112 - Update fs/Makefile, fs/Config.help, fs/Config.in, and
1113 Documentation/filesystems/ntfs.txt for NTFS TNG.
1114 - Create kernel configuration option controlling whether debugging
1115 is enabled or not.
1116 - Add the required export of end_buffer_io_sync() from the patches
1117 directory to the kernel code.
1118 - Update inode.c::ntfs_show_options() with show_inodes mount option.
1119 - Update errors mount option.
1120
1121tng-0.0.7 - 13/02/2002 - The driver is now feature complete for read-only!
1122
1123 - Cleanup mft.c and it's debug/error output in particular. Fix a minor
1124 bug in mapping of extent inodes. Update all the comments to fit all
1125 the recent code changes.
1126 - Modify vcn_to_lcn() to cope with entirely unmapped runlists.
1127 - Cleanups in compress.c, mostly comments and folding help.
1128 - Implement attrib.c::map_run_list() as a generic helper.
1129 - Make compress.c::ntfs_file_read_compressed_block() use map_run_list()
1130 thus making code shorter and enabling attribute list support.
1131 - Cleanup incorrect use of [su]64 with %L printf format specifier in
1132 all source files. Type casts to [unsigned] long long added to correct
1133 the mismatches (important for architectures which have long long not
1134 being 64 bits).
1135 - Merge async io completion handlers for directory indexes and $MFT
1136 data into one by setting the index_block_size{_bits} of the ntfs
1137 inode for $MFT to the mft_record_size{_bits} of the ntfs_volume.
1138 - Cleanup aops.c, update comments.
1139 - Make ntfs_file_get_block() use map_run_list() so all files now
1140 support attribute lists.
1141 - Make ntfs_dir_readpage() almost verbatim copy of
1142 block_read_full_page() by using ntfs_file_get_block() with only real
1143 difference being the use of our own async io completion handler
1144 rather than the default one, thus reducing the amount of code and
1145 automatically enabling attribute list support for directory indices.
1146 - Fix bug in load_attribute_list() - forgot to call brelse in error
1147 code path.
1148 - Change parameters to find_attr() and lookup_attr(). We no longer
1149 pass in the upcase table and its length. These can be gotten from
1150 ctx->ntfs_ino->vol->upcase{_len}. Update all callers.
1151 - Cleanups in attrib.c.
1152 - Implement merging of runlists, attrib.c::merge_run_lists() and its
1153 helpers. (Richard Russon)
1154 - Attribute lists part 2, attribute extents and multi part runlists:
1155 enable proper support for LCN_RL_NOT_MAPPED and automatic mapping of
1156 further runlist parts via attrib.c::map_run_list().
1157 - Tiny endianness bug fix in decompress_mapping_pairs().
1158
1159tng-0.0.6 - Encrypted directories, bug fixes, cleanups, debugging enhancements.
1160
1161 - Enable encrypted directories. (Their index root is marked encrypted
1162 to indicate that new files in that directory should be created
1163 encrypted.)
1164 - Fix bug in NInoBmpNonResident() macro. (Cut and paste error.)
1165 - Enable $Extend system directory. Most (if not all) extended system
1166 files do not have unnamed data attributes so ntfs_read_inode() had to
1167 special case them but that is ok, as the special casing recovery
1168 happens inside an error code path so there is zero slow down in the
1169 normal fast path. The special casing is done by introducing a new
1170 function inode.c::ntfs_is_extended_system_file() which checks if any
1171 of the hard links in the inode point to $Extend as being their parent
1172 directory and if they do we assume this is an extended system file.
1173 - Create a sysctl/proc interface to allow {dis,en}abling of debug output
1174 when compiled with -DDEBUG. Default is debug messages to be disabled.
1175 To enable them, one writes a non-zero value to /proc/sys/fs/ntfs-debug
1176 (if /proc is enabled) or uses sysctl(2) to effect the same (if sysctl
1177 interface is enabled). Inspired by old ntfs driver.
1178 - Add debug_msgs insmod/kernel boot parameter to set whether debug
1179 messages are {dis,en}abled. This is useful to enable debug messages
1180 during ntfs initialization and is the only way to activate debugging
1181 when the sysctl interface is not enabled.
1182 - Cleanup debug output in various places.
1183 - Remove all dollar signs ($) from the source (except comments) to
1184 enable compilation on architectures whose gcc compiler does not
1185 support dollar signs in the names of variables/constants. Attribute
1186 types now start with AT_ instead of $ and $I30 is now just I30.
1187 - Cleanup ntfs_lookup() and add consistency check of sequence numbers.
1188 - Load complete runlist for $MFT/$BITMAP during mount and cleanup
1189 access functions. This means we now cope with $MFT/$BITMAP being
1190 spread accross several mft records.
1191 - Disable modification of mft_zone_multiplier on remount. We can always
1192 reenable this later on if we really want to, but we will need to make
1193 sure we readjust the mft_zone size / layout accordingly.
1194
1195tng-0.0.5 - Modernize for 2.5.x and further in line-ing with Al Viro's comments.
1196
1197 - Use sb_set_blocksize() instead of set_blocksize() and verify the
1198 return value.
1199 - Use sb_bread() instead of bread() throughout.
1200 - Add index_vcn_size{_bits} to ntfs_inode structure to store the size
1201 of a directory index block vcn. Apply resulting simplifications in
1202 dir.c everywhere.
1203 - Fix a small bug somewhere (but forgot what it was).
1204 - Change ntfs_{debug,error,warning} to enable gcc to do type checking
1205 on the printf-format parameter list and fix bugs reported by gcc
1206 as a result. (Richard Russon)
1207 - Move inode allocation strategy to Al's new stuff but maintain the
1208 divorce of ntfs_inode from struct inode. To achieve this we have two
1209 separate slab caches, one for big ntfs inodes containing a struct
1210 inode and pure ntfs inodes and at the same time fix some faulty
1211 error code paths in ntfs_read_inode().
1212 - Show mount options in proc (inode.c::ntfs_show_options()).
1213
1214tng-0.0.4 - Big changes, getting in line with Al Viro's comments.
1215
1216 - Modified (un)map_mft_record functions to be common for read and write
1217 case. To specify which is which, added extra parameter at front of
1218 parameter list. Pass either READ or WRITE to this, each has the
1219 obvious meaning.
1220 - General cleanups to allow for easier folding in vi.
1221 - attrib.c::decompress_mapping_pairs() now accepts the old runlist
1222 argument, and invokes attrib.c::merge_run_lists() to merge the old
1223 and the new runlists.
1224 - Removed attrib.c::find_first_attr().
1225 - Implemented loading of attribute list and complete runlist for $MFT.
1226 This means we now cope with $MFT being spread across several mft
1227 records.
1228 - Adapt to 2.5.2-pre9 and the changed create_empty_buffers() syntax.
1229 - Adapt major/minor/kdev_t/[bk]devname stuff to new 2.5.x kernels.
1230 - Make ntfs_volume be allocated via kmalloc() instead of using a slab
1231 cache. There are too little ntfs_volume structures at any one time
1232 to justify a private slab cache.
1233 - Fix bogus kmap() use in async io completion. Now use kmap_atomic().
1234 Use KM_BIO_IRQ on advice from IRC/kernel...
1235 - Use ntfs_map_page() in map_mft_record() and create ->readpage method
1236 for reading $MFT (ntfs_mft_readpage). In the process create dedicated
1237 address space operations (ntfs_mft_aops) for $MFT inode mapping. Also
1238 removed the now superfluous exports from the kernel core patch.
1239 - Fix a bug where kfree() was used insted of ntfs_free().
1240 - Change map_mft_record() to take ntfs_inode as argument instead of
1241 vfs inode. Dito for unmap_mft_record(). Adapt all callers.
1242 - Add pointer to ntfs_volume to ntfs_inode.
1243 - Add mft record number and sequence number to ntfs_inode. Stop using
1244 i_ino and i_generation for in-driver purposes.
1245 - Implement attrib.c::merge_run_lists(). (Richard Russon)
1246 - Remove use of proper inodes by extent inodes. Move i_ino and
1247 i_generation to ntfs_inode to do this. Apply simplifications that
1248 result and remove iget_no_wait(), etc.
1249 - Pass ntfs_inode everywhere in the driver (used to be struct inode).
1250 - Add reference counting in ntfs_inode for the ntfs inode itself and
1251 for the mapped mft record.
1252 - Extend mft record mapping so we can (un)map extent mft records (new
1253 functions (un)map_extent_mft_record), and so mappings are reference
1254 counted and don't have to happen twice if already mapped - just ref
1255 count increases.
1256 - Add -o iocharset as alias to -o nls for backwards compatibility.
1257 - The latest core patch is now tiny. In fact just a single additional
1258 export is necessary over the base kernel.
1259
1260tng-0.0.3 - Cleanups, enhancements, bug fixes.
1261
1262 - Work on attrib.c::decompress_mapping_pairs() to detect base extents
1263 and setup the runlist appropriately using knowledge provided by the
1264 sizes in the base attribute record.
1265 - Balance the get_/put_attr_search_ctx() calls so we don't leak memory
1266 any more.
1267 - Introduce ntfs_malloc_nofs() and ntfs_free() to allocate/free a single
1268 page or use vmalloc depending on the amount of memory requested.
1269 - Cleanup error output. The __FUNCTION__ "(): " is now added
1270 automatically. Introduced a new header file debug.h to support this
1271 and also moved ntfs_debug() function into it.
1272 - Make reading of compressed files more intelligent and especially get
1273 rid of the vmalloc_nofs() from readpage(). This now uses per CPU
1274 buffers (allocated at first mount with cluster size <= 4kiB and
1275 deallocated on last umount with cluster size <= 4kiB), and
1276 asynchronous io for the compressed data using a list of buffer heads.
1277 Er, we use synchronous io as async io only works on whole pages
1278 covered by buffers and not on individual buffer heads...
1279 - Bug fix for reading compressed files with sparse compression blocks.
1280
1281tng-0.0.2 - Now handles larger/fragmented/compressed volumes/files/dirs.
1282
1283 - Fixed handling of directories when cluster size exceeds index block
1284 size.
1285 - Hide DOS only name space directory entries from readdir() but allow
1286 them in lookup(). This should fix the problem that Linux doesn't
1287 support directory hard links, while still allowing access to entries
1288 via their short file name. This also has the benefit of mimicking
1289 what Windows users are used to, so it is the ideal solution.
1290 - Implemented sync_page everywhere so no more hangs in D state when
1291 waiting for a page.
1292 - Stop using bforget() in favour of brelse().
1293 - Stop locking buffers unnecessarily.
1294 - Implemented compressed files (inode->mapping contains uncompressed
1295 data, raw compressed data is currently bread() into a vmalloc()ed
1296 memory buffer).
1297 - Enable compressed directories. (Their index root is marked compressed
1298 to indicate that new files in that directory should be created
1299 compressed.)
1300 - Use vsnprintf rather than vsprintf in the ntfs_error and ntfs_warning
1301 functions. (Thanks to Will Dyson for pointing this out.)
1302 - Moved the ntfs_inode and ntfs_volume (the former ntfs_inode_info and
1303 ntfs_sb_info) out of the common inode and super_block structures and
1304 started using the generic_ip and generic_sbp pointers instead. This
1305 makes ntfs entirely private with respect to the kernel tree.
1306 - Detect compiler version and abort with error message if gcc less than
1307 2.96 is used.
1308 - Fix bug in name comparison function in unistr.c.
1309 - Implement attribute lists part 1, the infrastructure: search contexts
1310 and operations, find_external_attr(), lookup_attr()) and make the
1311 code use the infrastructure.
1312 - Fix stupid buffer overflow bug that became apparent on larger run
1313 list containing attributes.
1314 - Fix bugs in readdir() that became apparent on larger directories.
1315
1316 The driver is now really useful and survives the test
1317 find . -type f -exec md5sum "{}" \;
1318 without any error messages on a over 1GiB sized partition with >16k
1319 files on it, including compressed files and directories and many files
1320 and directories with attribute lists.
1321
1322tng-0.0.1 - The first useful version.
1323
1324 - Added ntfs_lookup().
1325 - Added default upcase generation and handling.
1326 - Added compile options to be shown on module init.
1327 - Many bug fixes that were "hidden" before.
1328 - Update to latest kernel.
1329 - Added ntfs_readdir().
1330 - Added file operations for mmap(), read(), open() and llseek(). We just
1331 use the generic ones. The whole point of going through implementing
1332 readpage() methods and where possible get_block() call backs is that
1333 this allows us to make use of the generic high level methods provided
1334 by the kernel.
1335
1336 The driver is now actually useful! Yey. (-: It undoubtedly has got bugs
1337 though and it doesn't implement accesssing compressed files yet. Also,
1338 accessing files with attribute list attributes is not implemented yet
1339 either. But for small or simple file systems it should work and allow
1340 you to list directories, use stat on directory entries and the file
1341 system, open, read, mmap and llseek around in files. A big mile stone
1342 has been reached!
1343
1344tng-0.0.0 - Initial version tag.
1345
1346 Initial driver implementation. The driver can mount and umount simple
1347 NTFS file systems (i.e. ones without attribute lists in the system
1348 files). If the mount fails there might be problems in the error handling
1349 code paths, so be warned. Otherwise it seems to be loading the system
1350 files nicely and the mft record read mapping/unmapping seems to be
1351 working nicely, too. Proof of inode metadata in the page cache and non-
1352 resident file unnamed stream data in the page cache concepts is thus
1353 complete.