blob: 85f797a2edad8df0e0648c0c6868948781e970dd [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001ToDo/Notes:
2 - Find and fix bugs.
Linus Torvalds1da177e2005-04-16 15:20:36 -07003 - In between ntfs_prepare/commit_write, need exclusion between
Anton Altaparmakov66129f82004-11-11 12:34:00 +00004 simultaneous file extensions. This is given to us by holding i_sem
5 on the inode. The only places in the kernel when a file is resized
6 are prepare/commit write and truncate for both of which i_sem is
7 held. Just have to be careful in readpage/writepage and all other
8 helpers not running under i_sem that we play nice...
9 Also need to be careful with initialized_size extention in
Linus Torvalds1da177e2005-04-16 15:20:36 -070010 ntfs_prepare_write. Basically, just be _very_ careful in this code...
Anton Altaparmakov66129f82004-11-11 12:34:00 +000011 UPDATE: The only things that need to be checked are read/writepage
12 which do not hold i_sem. Note writepage cannot change i_size but it
13 needs to cope with a concurrent i_size change, just like readpage.
14 Also both need to cope with concurrent changes to the other sizes,
15 i.e. initialized/allocated/compressed size, as well.
Linus Torvalds1da177e2005-04-16 15:20:36 -070016 - Implement mft.c::sync_mft_mirror_umount(). We currently will just
17 leave the volume dirty on umount if the final iput(vol->mft_ino)
18 causes a write of any mirrored mft records due to the mft mirror
19 inode having been discarded already. Whether this can actually ever
20 happen is unclear however so it is worth waiting until someone hits
21 the problem.
22 - Enable the code for setting the NT4 compatibility flag when we start
23 making NTFS 1.2 specific modifications.
24
Anton Altaparmakovfd9d6362005-10-04 13:44:48 +0100252.1.25-WIP
26
27 - Change ntfs_map_runlist_nolock() to also take an optional attribute
28 search context. This allows calling it with the mft record mapped.
29
Anton Altaparmakov7d333d62005-09-08 23:01:16 +0100302.1.24 - Lots of bug fixes and support more clean journal states.
Anton Altaparmakove7a10332005-09-08 16:12:28 +010031
32 - Support journals ($LogFile) which have been modified by chkdsk. This
33 means users can boot into Windows after we marked the volume dirty.
34 The Windows boot will run chkdsk and then reboot. The user can then
35 immediately boot into Linux rather than having to do a full Windows
36 boot first before rebooting into Linux and we will recognize such a
Anton Altaparmakov5a8c0cc2005-09-26 10:48:54 +010037 journal and empty it as it is clean by definition. Note, this only
38 works if chkdsk left the journal in an obviously clean state.
Anton Altaparmakove7a10332005-09-08 16:12:28 +010039 - Support journals ($LogFile) with only one restart page as well as
40 journals with two different restart pages. We sanity check both and
41 either use the only sane one or the more recent one of the two in the
42 case that both are valid.
Anton Altaparmakov06d0e3c2005-09-08 16:28:25 +010043 - Add fs/ntfs/malloc.h::ntfs_malloc_nofs_nofail() which is analogous to
44 ntfs_malloc_nofs() but it performs allocations with __GFP_NOFAIL and
45 hence cannot fail.
Anton Altaparmakov9529d462005-09-08 16:33:12 +010046 - Use ntfs_malloc_nofs_nofail() in the two critical regions in
47 fs/ntfs/runlist.c::ntfs_runlists_merge(). This means we no longer
48 need to panic() if the allocation fails as it now cannot fail.
Anton Altaparmakov84d6ebe2005-09-08 16:46:55 +010049 - Fix two nasty runlist merging bugs that had gone unnoticed so far.
50 Thanks to Stefano Picerno for the bug report.
Anton Altaparmakov8bb73522005-09-08 16:48:28 +010051 - Remove two bogus BUG_ON()s from fs/ntfs/mft.c.
Anton Altaparmakov2b0ada22005-09-08 16:52:31 +010052 - Fix handling of valid but empty mapping pairs array in
53 fs/ntfs/runlist.c::ntfs_mapping_pairs_decompress().
Anton Altaparmakovf94ad382005-09-08 17:04:11 +010054 - Report unrepresentable inodes during ntfs_readdir() as KERN_WARNING
55 messages and include the inode number. Thanks to Yura Pakhuchiy for
56 pointing this out.
Anton Altaparmakov3ffc5a42005-09-08 20:23:06 +010057 - Change ntfs_rl_truncate_nolock() to throw away the runlist if the new
58 length is zero.
Anton Altaparmakov6e483212005-09-08 20:26:34 +010059 - Add runlist.[hc]::ntfs_rl_punch_nolock() which punches a caller
60 specified hole into a runlist.
Anton Altaparmakov8e08cea2005-09-08 20:29:50 +010061 - Fix a bug in fs/ntfs/index.c::ntfs_index_lookup(). When the returned
62 index entry is in the index root, we forgot to set the @ir pointer in
63 the index context. Thanks to Yura Pakhuchiy for finding this bug.
Anton Altaparmakovf25dfb52005-09-08 20:35:33 +010064 - Remove bogus setting of PageError in ntfs_read_compressed_block().
Anton Altaparmakov0aaccea2005-09-08 20:40:32 +010065 - Add fs/ntfs/attrib.[hc]::ntfs_resident_attr_value_resize().
Anton Altaparmakov2983d1b2005-09-08 20:56:09 +010066 - Fix a bug in ntfs_map_runlist_nolock() where we forgot to protect
67 access to the allocated size in the ntfs inode with the size lock.
68 - Fix ntfs_attr_vcn_to_lcn_nolock() and ntfs_attr_find_vcn_nolock() to
69 return LCN_ENOENT when there is no runlist and the allocated size is
70 zero.
71 - Fix load_attribute_list() to handle the case of a NULL runlist.
Anton Altaparmakov807c4532005-09-08 21:01:17 +010072 - Fix handling of sparse attributes in ntfs_attr_make_non_resident().
73 - Add BUG() checks to ntfs_attr_make_non_resident() and ntfs_attr_set()
74 to ensure that these functions are never called for compressed or
75 encrypted attributes.
Anton Altaparmakovbbf18132005-09-08 21:09:06 +010076 - Fix cluster (de)allocators to work when the runlist is NULL and more
77 importantly to take a locked runlist rather than them locking it
78 which leads to lock reversal.
Anton Altaparmakov1c7d4692005-09-08 21:15:09 +010079 - Truncate {a,c,m}time to the ntfs supported time granularity when
80 updating the times in the inode in ntfs_setattr().
Anton Altaparmakov67bb1032005-09-08 21:19:45 +010081 - Fixup handling of sparse, compressed, and encrypted attributes in
Anton Altaparmakov311120e2005-09-08 22:04:20 +010082 fs/ntfs/inode.c::ntfs_read_locked_{,attr_,index_}inode(),
83 fs/ntfs/aops.c::ntfs_{read,write}page().
Anton Altaparmakov8dcdeba2005-09-08 21:25:48 +010084 - Make ntfs_write_block() not instantiate sparse blocks if they contain
85 only zeroes.
Anton Altaparmakov54b02eb2005-09-08 21:43:47 +010086 - Optimize fs/ntfs/aops.c::ntfs_write_block() by extending the page
87 lock protection over the buffer submission for i/o which allows the
88 removal of the get_bh()/put_bh() pairs for each buffer.
Anton Altaparmakov8273d5d2005-09-08 22:00:33 +010089 - Fix fs/ntfs/aops.c::ntfs_{read,write}_block() to handle the case
90 where a concurrent truncate has truncated the runlist under our feet.
Anton Altaparmakova01ac532005-09-08 22:08:11 +010091 - Fix page_has_buffers()/page_buffers() handling in fs/ntfs/aops.c.
Anton Altaparmakove6046352005-09-08 22:13:02 +010092 - In fs/ntfs/aops.c::ntfs_end_buffer_async_read(), use a bit spin lock
93 in the first buffer head instead of a driver global spin lock to
94 improve scalability.
Anton Altaparmakov7d333d62005-09-08 23:01:16 +010095 - Minor fix to error handling and error message display in
Anton Altaparmakov5d467702005-09-12 14:33:47 +010096 fs/ntfs/aops.c::ntfs_prepare_nonresident_write().
97 - Change the mount options {u,f,d}mask to always parse the number as
98 an octal number to conform to how chmod(1) works, too. Thanks to
99 Giuseppe Bilotta and Horst von Brand for pointing out the errors of
100 my ways.
Anton Altaparmakov5c9f6de2005-09-19 09:33:40 +0100101 - Fix various bugs in the runlist merging code. (Based on libntfs
102 changes by Richard Russon.)
Anton Altaparmakov5a8c0cc2005-09-26 10:48:54 +0100103 - Fix sparse warnings that have crept in over time.
104 - Change ntfs_cluster_free() to require a write locked runlist on entry
105 since we otherwise get into a lock reversal deadlock if a read locked
106 runlist is passed in. In the process also change it to take an ntfs
107 inode instead of a vfs inode as parameter.
108 - Fix the definition of the CHKD ntfs record magic. It had an off by
109 two error causing it to be CHKB instead of CHKD.
Anton Altaparmakov18efefa2005-10-04 13:06:00 +0100110 - Fix a stupid bug in __ntfs_bitmap_set_bits_in_run() which caused the
111 count to become negative and hence we had a wild memset() scribbling
112 all over the system's ram.
Anton Altaparmakove7a10332005-09-08 16:12:28 +0100113
Anton Altaparmakovaf859a42005-06-25 21:07:27 +01001142.1.23 - Implement extension of resident files and make writing safe as well as
115 many bug fixes, cleanups, and enhancements...
Linus Torvalds1da177e2005-04-16 15:20:36 -0700116
117 - Add printk rate limiting for ntfs_warning() and ntfs_error() when
118 compiled without debug. This avoids a possible denial of service
119 attack. Thanks to Carl-Daniel Hailfinger from SuSE for pointing this
120 out.
Randy Dunlap89075472005-03-03 11:19:53 +0000121 - Fix compilation warnings on ia64. (Randy Dunlap)
Anton Altaparmakovaf859a42005-06-25 21:07:27 +0100122 - Use i_size_{read,write}() instead of reading i_size by hand and cache
123 the value where apropriate.
Anton Altaparmakov36763672004-11-18 13:46:45 +0000124 - Add size_lock to the ntfs_inode structure. This is an rw spinlock
125 and it locks against access to the inode sizes. Note, ->size_lock
126 is also accessed from irq context so you must use the _irqsave and
Anton Altaparmakovaf859a42005-06-25 21:07:27 +0100127 _irqrestore lock and unlock functions, respectively. Protect all
128 accesses to allocated_size, initialized_size, and compressed_size.
Anton Altaparmakov218357f2004-11-18 20:34:59 +0000129 - Minor optimization to fs/ntfs/super.c::ntfs_statfs() and its helpers.
Anton Altaparmakov946929d2005-01-13 15:26:29 +0000130 - Implement extension of resident files in the regular file write code
131 paths (fs/ntfs/aops.c::ntfs_{prepare,commit}_write()). At present
132 this only works until the data attribute becomes too big for the mft
133 record after which we abort the write returning -EOPNOTSUPP from
134 ntfs_prepare_write().
Anton Altaparmakovc002f422005-02-03 12:02:56 +0000135 - Add disable_sparse mount option together with a per volume sparse
136 enable bit which is set appropriately and a per inode sparse disable
137 bit which is preset on some system file inodes as appropriate.
138 - Enforce that sparse support is disabled on NTFS volumes pre 3.0.
Anton Altaparmakov1a0df152005-02-03 12:04:36 +0000139 - Fix a bug in fs/ntfs/runlist.c::ntfs_mapping_pairs_decompress() in
140 the creation of the unmapped runlist element for the base attribute
141 extent.
Anton Altaparmakovb6ad6c52005-02-15 10:08:43 +0000142 - Split ntfs_map_runlist() into ntfs_map_runlist() and a non-locking
143 helper ntfs_map_runlist_nolock() which is used by ntfs_map_runlist().
144 This allows us to map runlist fragments with the runlist lock already
145 held without having to drop and reacquire it around the call. Adapt
146 all callers.
147 - Change ntfs_find_vcn() to ntfs_find_vcn_nolock() which takes a locked
148 runlist. This allows us to find runlist elements with the runlist
149 lock already held without having to drop and reacquire it around the
150 call. Adapt all callers.
Anton Altaparmakov37e4c132005-02-18 10:03:13 +0000151 - Change time to u64 in time.h::ntfs2utc() as it otherwise generates a
Anton Altaparmakovd8ec7852005-02-18 09:23:39 +0000152 warning in the do_div() call on sparc32. Thanks to Meelis Roos for
153 the report and analysis of the warning.
Anton Altaparmakov37e4c132005-02-18 10:03:13 +0000154 - Fix a nasty runlist merge bug when merging two holes.
Anton Altaparmakov5ae9fcf2005-03-02 17:03:24 +0000155 - Set the ntfs_inode->allocated_size to the real allocated size in the
156 mft record for resident attributes (fs/ntfs/inode.c).
157 - Small readability cleanup to use "a" instead of "ctx->attr"
158 everywhere (fs/ntfs/inode.c).
Anton Altaparmakov41382682005-03-03 13:44:15 +0000159 - Make fs/ntfs/namei.c::ntfs_get_{parent,dentry} static and move the
160 definition of ntfs_export_ops from fs/ntfs/super.c to namei.c. Also,
161 declare ntfs_export_ops in fs/ntfs/ntfs.h.
Anton Altaparmakov9451f852005-03-03 14:43:43 +0000162 - Correct sparse file handling. The compressed values need to be
163 checked and set in the ntfs inode as done for compressed files and
164 the compressed size needs to be used for vfs inode->i_blocks instead
165 of the allocated size, again, as done for compressed files.
Anton Altaparmakov271849a2005-03-07 21:36:18 +0000166 - Add AT_EA in addition to AT_DATA to whitelist for being allowed to be
167 non-resident in fs/ntfs/attrib.c::ntfs_attr_can_be_non_resident().
168 - Add fs/ntfs/attrib.c::ntfs_attr_vcn_to_lcn_nolock() used by the new
169 write code.
Anton Altaparmakovc0c1cc02005-03-07 21:43:38 +0000170 - Fix bug in fs/ntfs/attrib.c::ntfs_find_vcn_nolock() where after
171 dropping the read lock and taking the write lock we were not checking
172 whether someone else did not already do the work we wanted to do.
173 - Rename fs/ntfs/attrib.c::ntfs_find_vcn_nolock() to
174 ntfs_attr_find_vcn_nolock() and update all callers.
Anton Altaparmakov2bfb4ff2005-03-09 15:15:06 +0000175 - Add fs/ntfs/attrib.[hc]::ntfs_attr_make_non_resident().
Anton Altaparmakov43b01fd2005-03-09 15:18:43 +0000176 - Fix sign of various error return values to be negative in
177 fs/ntfs/lcnalloc.c.
Anton Altaparmakov905685f2005-03-10 11:06:19 +0000178 - Modify ->readpage and ->writepage (fs/ntfs/aops.c) so they detect and
179 handle the case where an attribute is converted from resident to
180 non-resident by a concurrent file write.
Jesper Juhl251c8422005-04-04 14:59:56 +0100181 - Remove checks for NULL before calling kfree() since kfree() does the
182 checking itself. (Jesper Juhl)
Anton Altaparmakovb0d23742005-04-04 16:20:14 +0100183 - Some utilities modify the boot sector but do not update the checksum.
184 Thus, relax the checking in fs/ntfs/super.c::is_boot_sector_ntfs() to
185 only emit a warning when the checksum is incorrect rather than
186 refusing the mount. Thanks to Bernd Casimir for pointing this
187 problem out.
Anton Altaparmakovbb3cf332005-04-06 13:34:31 +0100188 - Update attribute definition handling.
Anton Altaparmakov7fafb8b2005-04-06 16:09:21 +0100189 - Add NTFS_MAX_CLUSTER_SIZE and NTFS_MAX_PAGES_PER_CLUSTER constants.
190 - Use NTFS_MAX_CLUSTER_SIZE in super.c instead of hard coding 0x10000.
Anton Altaparmakovd53ee322005-04-06 16:11:20 +0100191 - Use MAX_BUF_PER_PAGE instead of variable sized array allocation for
192 better code generation and one less sparse warning in fs/ntfs/aops.c.
Pekka Enberg2fb21db2005-05-25 21:15:34 +0300193 - Remove spurious void pointer casts from fs/ntfs/. (Pekka Enberg)
Anton Altaparmakov442d2072005-05-27 16:42:56 +0100194 - Use C99 style structure initialization after memory allocation where
195 possible (fs/ntfs/{attrib.c,index.c,super.c}). Thanks to Al Viro and
196 Pekka Enberg.
Anton Altaparmakov3f2faef2005-06-25 15:28:56 +0100197 - Stamp the transaction log ($UsnJrnl), aka user space journal, if it
198 is active on the volume and we are mounting read-write or remounting
199 from read-only to read-write.
Anton Altaparmakov9f993fe2005-06-25 16:15:36 +0100200 - Fix a bug in address space operations error recovery code paths where
201 if the runlist was not mapped at all and a mapping error occured we
202 would leave the runlist locked on exit to the function so that the
203 next access to the same file would try to take the lock and deadlock.
Anton Altaparmakovca8fd7a2005-06-25 16:31:27 +0100204 - Detect the case when Windows has been suspended to disk on the volume
205 to be mounted and if this is the case do not allow (re)mounting
206 read-write. This is done by parsing hiberfil.sys if present.
Anton Altaparmakov3bd1f4a2005-06-25 16:51:58 +0100207 - Fix several occurences of a bug where we would perform 'var & ~const'
208 with a 64-bit variable and a int, i.e. 32-bit, constant. This causes
209 the higher order 32-bits of the 64-bit variable to be zeroed. To fix
210 this cast the 'const' to the same 64-bit type as 'var'.
Anton Altaparmakov1d58b272005-06-25 17:04:55 +0100211 - Change the runlist terminator of the newly allocated cluster(s) to
212 LCN_ENOENT in ntfs_attr_make_non_resident(). Otherwise the runlist
213 code gets confused.
Anton Altaparmakovfa3be922005-06-25 17:15:36 +0100214 - Add an extra parameter @last_vcn to ntfs_get_size_for_mapping_pairs()
215 and ntfs_mapping_pairs_build() to allow the runlist encoding to be
216 partial which is desirable when filling holes in sparse attributes.
217 Update all callers.
Anton Altaparmakov4757d7df2005-06-25 17:24:08 +0100218 - Change ntfs_map_runlist_nolock() to only decompress the mapping pairs
219 if the requested vcn is inside it. Otherwise we get into problems
220 when we try to map an out of bounds vcn because we then try to map
221 the already mapped runlist fragment which causes
222 ntfs_mapping_pairs_decompress() to fail and return error. Update
223 ntfs_attr_find_vcn_nolock() accordingly.
Anton Altaparmakovba6d2372005-06-26 22:12:02 +0100224 - Fix a nasty deadlock that appeared in recent kernels.
225 The situation: VFS inode X on a mounted ntfs volume is dirty. For
226 same inode X, the ntfs_inode is dirty and thus corresponding on-disk
227 inode, i.e. mft record, which is in a dirty PAGE_CACHE_PAGE belonging
228 to the table of inodes, i.e. $MFT, inode 0.
229 What happens:
230 Process 1: sys_sync()/umount()/whatever... calls
231 __sync_single_inode() for $MFT -> do_writepages() -> write_page for
232 the dirty page containing the on-disk inode X, the page is now locked
233 -> ntfs_write_mst_block() which clears PageUptodate() on the page to
234 prevent anyone else getting hold of it whilst it does the write out.
235 This is necessary as the on-disk inode needs "fixups" applied before
236 the write to disk which are removed again after the write and
237 PageUptodate is then set again. It then analyses the page looking
238 for dirty on-disk inodes and when it finds one it calls
239 ntfs_may_write_mft_record() to see if it is safe to write this
240 on-disk inode. This then calls ilookup5() to check if the
241 corresponding VFS inode is in icache(). This in turn calls ifind()
242 which waits on the inode lock via wait_on_inode whilst holding the
243 global inode_lock.
244 Process 2: pdflush results in a call to __sync_single_inode for the
245 same VFS inode X on the ntfs volume. This locks the inode (I_LOCK)
246 then calls write-inode -> ntfs_write_inode -> map_mft_record() ->
247 read_cache_page() for the page (in page cache of table of inodes
248 $MFT, inode 0) containing the on-disk inode. This page has
249 PageUptodate() clear because of Process 1 (see above) so
250 read_cache_page() blocks when it tries to take the page lock for the
251 page so it can call ntfs_read_page().
252 Thus Process 1 is holding the page lock on the page containing the
253 on-disk inode X and it is waiting on the inode X to be unlocked in
254 ifind() so it can write the page out and then unlock the page.
255 And Process 2 is holding the inode lock on inode X and is waiting for
256 the page to be unlocked so it can call ntfs_readpage() or discover
257 that Process 1 set PageUptodate() again and use the page.
258 Thus we have a deadlock due to ifind() waiting on the inode lock.
259 The solution: The fix is to use the newly introduced
260 ilookup5_nowait() which does not wait on the inode's lock and hence
261 avoids the deadlock. This is safe as we do not care about the VFS
262 inode and only use the fact that it is in the VFS inode cache and the
263 fact that the vfs and ntfs inodes are one struct in memory to find
264 the ntfs inode in memory if present. Also, the ntfs inode has its
265 own locking so it does not matter if the vfs inode is locked.
Anton Altaparmakove74589a2005-08-16 16:38:28 +0100266 - Fix bug in mft record writing where we forgot to set the device in
Anton Altaparmakov481d0372005-08-16 19:42:56 +0100267 the buffers when mapping them after the VM had discarded them.
Anton Altaparmakove74589a2005-08-16 16:38:28 +0100268 Thanks to Martin MOKREJÅ  for the bug report.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700269
2702.1.22 - Many bug and race fixes and error handling improvements.
271
272 - Improve error handling in fs/ntfs/inode.c::ntfs_truncate().
273 - Change fs/ntfs/inode.c::ntfs_truncate() to return an error code
274 instead of void and provide a helper ntfs_truncate_vfs() for the
275 vfs ->truncate method.
276 - Add a new ntfs inode flag NInoTruncateFailed() and modify
277 fs/ntfs/inode.c::ntfs_truncate() to set and clear it appropriately.
278 - Fix min_size and max_size definitions in ATTR_DEF structure in
279 fs/ntfs/layout.h to be signed.
280 - Add attribute definition handling helpers to fs/ntfs/attrib.[hc]:
281 ntfs_attr_size_bounds_check(), ntfs_attr_can_be_non_resident(), and
282 ntfs_attr_can_be_resident(), which in turn use the new private helper
283 ntfs_attr_find_in_attrdef().
284 - In fs/ntfs/aops.c::mark_ntfs_record_dirty(), take the
285 mapping->private_lock around the dirtying of the buffer heads
286 analagous to the way it is done in __set_page_dirty_buffers().
287 - Ensure the mft record size does not exceed the PAGE_CACHE_SIZE at
288 mount time as this cannot work with the current implementation.
289 - Check for location of attribute name and improve error handling in
290 general in fs/ntfs/inode.c::ntfs_read_locked_inode() and friends.
291 - In fs/ntfs/aops.c::ntfs_writepage(), if the page is fully outside
292 i_size, i.e. race with truncate, invalidate the buffers on the page
293 so that they become freeable and hence the page does not leak.
294 - Remove unused function fs/ntfs/runlist.c::ntfs_rl_merge(). (Adrian
295 Bunk)
296 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_find() that resulted in
297 a NULL pointer dereference in the error code path when a corrupt
298 attribute was found. (Thanks to Domen Puncer for the bug report.)
299 - Add MODULE_VERSION() to fs/ntfs/super.c.
300 - Make several functions and variables static. (Adrian Bunk)
301 - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() so it allocates
302 buffers for the page if they are not present and then marks the
303 buffers belonging to the ntfs record dirty. This causes the buffers
304 to become busy and hence they are safe from removal until the page
305 has been written out.
306 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find() in the
307 error handling code path that resulted in a BUG() due to trying to
308 unmap an extent mft record when the mapping of it had failed and it
309 thus was not mapped. (Thanks to Ken MacFerrin for the bug report.)
310 - Drop the runlist lock after the vcn has been read in
311 fs/ntfs/lcnalloc.c::__ntfs_cluster_free().
312 - Rewrite handling of multi sector transfer errors. We now do not set
313 PageError() when such errors are detected in the async i/o handler
314 fs/ntfs/aops.c::ntfs_end_buffer_async_read(). All users of mst
315 protected attributes now check the magic of each ntfs record as they
316 use it and act appropriately. This has the effect of making errors
317 granular per ntfs record rather than per page which solves the case
318 where we cannot access any of the ntfs records in a page when a
319 single one of them had an mst error. (Thanks to Ken MacFerrin for
320 the bug report.)
321 - Fix error handling in fs/ntfs/quota.c::ntfs_mark_quotas_out_of_date()
322 where we failed to release i_sem on the $Quota/$Q attribute inode.
323 - Fix bug in handling of bad inodes in fs/ntfs/namei.c::ntfs_lookup().
324 - Add mapping of unmapped buffers to all remaining code paths, i.e.
325 fs/ntfs/aops.c::ntfs_write_mst_block(), mft.c::ntfs_sync_mft_mirror(),
326 and write_mft_record_nolock(). From now on we require that the
327 complete runlist for the mft mirror is always mapped into memory.
328 - Add creation of buffers to fs/ntfs/mft.c::ntfs_sync_mft_mirror().
329 - Improve error handling in fs/ntfs/aops.c::ntfs_{read,write}_block().
330 - Cleanup fs/ntfs/aops.c::ntfs_{read,write}page() since we know that a
331 resident attribute will be smaller than a page which makes the code
332 simpler. Also make the code more tolerant to concurrent ->truncate.
333
3342.1.21 - Fix some races and bugs, rewrite mft write code, add mft allocator.
335
336 - Implement extent mft record deallocation
337 fs/ntfs/mft.c::ntfs_extent_mft_record_free().
338 - Splitt runlist related functions off from attrib.[hc] to runlist.[hc].
339 - Add vol->mft_data_pos and initialize it at mount time.
340 - Rename init_runlist() to ntfs_init_runlist(), ntfs_vcn_to_lcn() to
341 ntfs_rl_vcn_to_lcn(), decompress_mapping_pairs() to
342 ntfs_mapping_pairs_decompress(), ntfs_merge_runlists() to
343 ntfs_runlists_merge() and adapt all callers.
344 - Add fs/ntfs/runlist.[hc]::ntfs_get_nr_significant_bytes(),
345 ntfs_get_size_for_mapping_pairs(), ntfs_write_significant_bytes(),
346 and ntfs_mapping_pairs_build(), adapted from libntfs.
347 - Make fs/ntfs/lcnalloc.c::ntfs_cluster_free_from_rl_nolock() not
348 static and add a declaration for it to lcnalloc.h.
349 - Add fs/ntfs/lcnalloc.h::ntfs_cluster_free_from_rl() which is a static
350 inline wrapper for ntfs_cluster_free_from_rl_nolock() which takes the
351 cluster bitmap lock for the duration of the call.
352 - Add fs/ntfs/attrib.[hc]::ntfs_attr_record_resize().
353 - Implement the equivalent of memset() for an ntfs attribute in
354 fs/ntfs/attrib.[hc]::ntfs_attr_set() and switch
355 fs/ntfs/logfile.c::ntfs_empty_logfile() to using it.
356 - Remove unnecessary casts from LCN_* constants.
357 - Implement fs/ntfs/runlist.c::ntfs_rl_truncate_nolock().
358 - Add MFT_RECORD_OLD as a copy of MFT_RECORD in fs/ntfs/layout.h and
359 change MFT_RECORD to contain the NTFS 3.1+ specific fields.
360 - Add a helper function fs/ntfs/aops.c::mark_ntfs_record_dirty() which
361 marks all buffers belonging to an ntfs record dirty, followed by
362 marking the page the ntfs record is in dirty and also marking the vfs
363 inode containing the ntfs record dirty (I_DIRTY_PAGES).
364 - Switch fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to using the
365 new helper fs/ntfs/aops.c::mark_ntfs_record_dirty() and remove the no
366 longer needed fs/ntfs/index.[hc]::__ntfs_index_entry_mark_dirty().
367 - Move ntfs_{un,}map_page() from ntfs.h to aops.h and fix resulting
368 include errors.
369 - Move the typedefs for runlist_element and runlist from types.h to
370 runlist.h and fix resulting include errors.
371 - Remove unused {__,}format_mft_record() from fs/ntfs/mft.c.
372 - Modify fs/ntfs/mft.c::__mark_mft_record_dirty() to use the helper
373 mark_ntfs_record_dirty() which also changes the behaviour in that we
374 now set the buffers belonging to the mft record dirty as well as the
375 page itself.
376 - Update fs/ntfs/mft.c::write_mft_record_nolock() and sync_mft_mirror()
377 to cope with the fact that there now are dirty buffers in mft pages.
378 - Update fs/ntfs/inode.c::ntfs_write_inode() to also use the helper
379 mark_ntfs_record_dirty() and thus to set the buffers belonging to the
380 mft record dirty as well as the page itself.
381 - Fix compiler warnings on x86-64 in fs/ntfs/dir.c. (Randy Dunlap,
382 slightly modified by me)
383 - Add fs/ntfs/mft.c::try_map_mft_record() which fails with -EALREADY if
384 the mft record is already locked and otherwise behaves the same way
385 as fs/ntfs/mft.c::map_mft_record().
386 - Modify fs/ntfs/mft.c::write_mft_record_nolock() so that it only
387 writes the mft record if the buffers belonging to it are dirty.
388 Otherwise we assume that it was written out by other means already.
389 - Attempting to write outside initialized size is _not_ a bug so remove
390 the bug check from fs/ntfs/aops.c::ntfs_write_mst_block(). It is in
391 fact required to write outside initialized size when preparing to
392 extend the initialized size.
393 - Map the page instead of using page_address() before writing to it in
394 fs/ntfs/aops.c::ntfs_mft_writepage().
395 - Provide exclusion between opening an inode / mapping an mft record
396 and accessing the mft record in fs/ntfs/mft.c::ntfs_mft_writepage()
397 by setting the page not uptodate throughout ntfs_mft_writepage().
398 - Clear the page uptodate flag in fs/ntfs/aops.c::ntfs_write_mst_block()
399 to ensure noone can see the page whilst the mst fixups are applied.
400 - Add the helper fs/ntfs/mft.c::ntfs_may_write_mft_record() which
401 checks if an mft record may be written out safely obtaining any
402 necessary locks in the process. This is used by
403 fs/ntfs/aops.c::ntfs_write_mst_block().
404 - Modify fs/ntfs/aops.c::ntfs_write_mst_block() to also work for
405 writing mft records and improve its error handling in the process.
406 Now if any of the records in the page fail to be written out, all
407 other records will be written out instead of aborting completely.
408 - Remove ntfs_mft_aops and update all users to use ntfs_mst_aops.
409 - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to set the
410 ntfs_mst_aops for all inodes which are NInoMstProtected() and
411 ntfs_aops for all other inodes.
412 - Rename fs/ntfs/mft.c::sync_mft_mirror{,_umount}() to
413 ntfs_sync_mft_mirror{,_umount}() and change their parameters so they
414 no longer require an ntfs inode to be present. Update all callers.
415 - Cleanup the error handling in fs/ntfs/mft.c::ntfs_sync_mft_mirror().
416 - Clear the page uptodate flag in fs/ntfs/mft.c::ntfs_sync_mft_mirror()
417 to ensure noone can see the page whilst the mst fixups are applied.
418 - Remove the no longer needed fs/ntfs/mft.c::ntfs_mft_writepage() and
419 fs/ntfs/mft.c::try_map_mft_record().
420 - Fix callers of fs/ntfs/aops.c::mark_ntfs_record_dirty() to call it
421 with the ntfs inode which contains the page rather than the ntfs
422 inode the mft record of which is in the page.
423 - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by moving the
424 index inode bitmap inode release code from there to
425 fs/ntfs/inode.c::ntfs_clear_big_inode(). (Thanks to Christoph
426 Hellwig for spotting this.)
427 - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by taking the
428 inode semaphore around the code that sets ni->itype.index.bmp_ino to
429 NULL and reorganize the code to optimize it a bit. (Thanks to
430 Christoph Hellwig for spotting this.)
431 - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() to no longer take the
432 ntfs inode as a parameter as this is confusing and misleading and the
433 needed ntfs inode is available via NTFS_I(page->mapping->host).
434 Adapt all callers to this change.
435 - Modify fs/ntfs/mft.c::write_mft_record_nolock() and
436 fs/ntfs/aops.c::ntfs_write_mst_block() to only check the dirty state
437 of the first buffer in a record and to take this as the ntfs record
438 dirty state. We cannot look at the dirty state for subsequent
439 buffers because we might be racing with
440 fs/ntfs/aops.c::mark_ntfs_record_dirty().
441 - Move the static inline ntfs_init_big_inode() from fs/ntfs/inode.c to
442 inode.h and make fs/ntfs/inode.c::__ntfs_init_inode() non-static and
443 add a declaration for it to inode.h. Fix some compilation issues
444 that resulted due to #includes and header file interdependencies.
445 - Simplify setup of i_mode in fs/ntfs/inode.c::ntfs_read_locked_inode().
446 - Add helpers fs/ntfs/layout.h::MK_MREF() and MK_LE_MREF().
447 - Modify fs/ntfs/mft.c::map_extent_mft_record() to only verify the mft
448 record sequence number if it is specified (i.e. not zero).
449 - Add fs/ntfs/mft.[hc]::ntfs_mft_record_alloc() and various helper
450 functions used by it.
451 - Update Documentation/filesystems/ntfs.txt with instructions on how to
452 use the Device-Mapper driver with NTFS ftdisk/LDM raid. This removes
453 the linear raid problem with the Software RAID / MD driver when one
454 or more of the devices has an odd number of sectors.
455
4562.1.20 - Fix two stupid bugs introduced in 2.1.18 release.
457
458 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_reinit_search_ctx()
459 where we did not clear ctx->al_entry but it was still set due to
460 changes in ntfs_attr_lookup() and ntfs_external_attr_find() in
461 particular.
462 - Fix another stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find()
463 where we forgot to unmap the extent mft record when we had finished
464 enumerating an attribute which caused a bug check to trigger when the
465 VFS calls ->clear_inode.
466
4672.1.19 - Many cleanups, improvements, and a minor bug fix.
468
469 - Update ->setattr (fs/ntfs/inode.c::ntfs_setattr()) to refuse to
470 change the uid, gid, and mode of an inode as we do not support NTFS
471 ACLs yet.
472 - Remove BKL use from ntfs_setattr() syncing up with the rest of the
473 kernel.
474 - Get rid of the ugly transparent union in fs/ntfs/dir.c::ntfs_readdir()
475 and ntfs_filldir() as per suggestion from Al Viro.
476 - Change '\0' and L'\0' to simply 0 as per advice from Linus Torvalds.
477 - Update ->truncate (fs/ntfs/inode.c::ntfs_truncate()) to check if the
478 inode size has changed and to only output an error if so.
479 - Rename fs/ntfs/attrib.h::attribute_value_length() to ntfs_attr_size().
480 - Add le{16,32,64} as well as sle{16,32,64} data types to
481 fs/ntfs/types.h.
482 - Change ntfschar to be le16 instead of u16 in fs/ntfs/types.h.
483 - Add le versions of VCN, LCN, and LSN called leVCN, leLCN, and leLSN,
484 respectively, to fs/ntfs/types.h.
485 - Update endianness conversion macros in fs/ntfs/endian.h to use the
486 new types as appropriate.
487 - Do proper type casting when using sle64_to_cpup() in fs/ntfs/dir.c
488 and index.c.
489 - Add leMFT_REF data type to fs/ntfs/layout.h.
490 - Update all NTFS header files with the new little endian data types.
491 Affected files are fs/ntfs/layout.h, logfile.h, and time.h.
492 - Do proper type casting when using ntfs_is_*_recordp() in
493 fs/ntfs/logfile.c, mft.c, and super.c.
494 - Fix all the sparse bitwise warnings. Had to change all the typedef
495 enums storing little endian values to simple enums plus a typedef for
496 the datatype to make sparse happy.
497 - Fix a bug found by the new sparse bitwise warnings where the default
498 upcase table was defined as a pointer to wchar_t rather than ntfschar
499 in fs/ntfs/ntfs.h and super.c.
500 - Change {const_,}cpu_to_le{16,32}(0) to just 0 as suggested by Al Viro.
501
5022.1.18 - Fix scheduling latencies at mount time as well as an endianness bug.
503
504 - Remove vol->nr_mft_records as it was pretty meaningless and optimize
505 the calculation of total/free inodes as used by statfs().
506 - Fix scheduling latencies in ntfs_fill_super() by dropping the BKL
507 because the code itself is using the ntfs_lock semaphore which
508 provides safe locking. (Ingo Molnar)
509 - Fix a potential bug in fs/ntfs/mft.c::map_extent_mft_record() that
510 could occur in the future for when we start closing/freeing extent
511 inodes if we don't set base_ni->ext.extent_ntfs_inos to NULL after
512 we free it.
513 - Rename {find,lookup}_attr() to ntfs_attr_{find,lookup}() as well as
514 find_external_attr() to ntfs_external_attr_find() to cleanup the
515 namespace a bit and to be more consistent with libntfs.
516 - Rename {{re,}init,get,put}_attr_search_ctx() to
517 ntfs_attr_{{re,}init,get,put}_search_ctx() as well as the type
518 attr_search_context to ntfs_attr_search_ctx.
519 - Force use of ntfs_attr_find() in ntfs_attr_lookup() when searching
520 for the attribute list attribute itself.
521 - Fix endianness bug in ntfs_external_attr_find().
522 - Change ntfs_{external_,}attr_find() to return 0 on success, -ENOENT
523 if the attribute is not found, and -EIO on real error. In the case
524 of -ENOENT, the search context is updated to describe the attribute
525 before which the attribute being searched for would need to be
526 inserted if such an action were to be desired and in the case of
527 ntfs_external_attr_find() the search context is also updated to
528 indicate the attribute list entry before which the attribute list
529 entry of the attribute being searched for would need to be inserted
530 if such an action were to be desired. Also make ntfs_find_attr()
531 static and remove its prototype from attrib.h as it is not used
532 anywhere other than attrib.c. Update ntfs_attr_lookup() and all
533 callers of ntfs_{external,}attr_{find,lookup}() for the new return
534 values.
535 - Minor cleanup of fs/ntfs/inode.c::ntfs_init_locked_inode().
536
5372.1.17 - Fix bugs in mount time error code paths and other updates.
538
539 - Implement bitmap modification code (fs/ntfs/bitmap.[hc]). This
540 includes functions to set/clear a single bit or a run of bits.
541 - Add fs/ntfs/attrib.[hc]::ntfs_find_vcn() which returns the locked
542 runlist element containing a particular vcn. It also takes care of
543 mapping any needed runlist fragments.
544 - Implement cluster (de-)allocation code (fs/ntfs/lcnalloc.[hc]).
545 - Load attribute definition table from $AttrDef at mount time.
546 - Fix bugs in mount time error code paths involving (de)allocation of
547 the default and volume upcase tables.
548 - Remove ntfs_nr_mounts as it is no longer used.
549
5502.1.16 - Implement access time updates, file sync, async io, and read/writev.
551
552 - Add support for readv/writev and aio_read/aio_write (fs/ntfs/file.c).
553 This is done by setting the appropriate file operations pointers to
554 the generic helper functions provided by mm/filemap.c.
555 - Implement fsync, fdatasync, and msync both for files (fs/ntfs/file.c)
556 and directories (fs/ntfs/dir.c).
557 - Add support for {a,m,c}time updates to inode.c::ntfs_write_inode().
558 Note, except for the root directory and any other system files opened
559 by the user, the system files will not have their access times
560 updated as they are only accessed at the inode level an hence the
561 file level functions which cause the times to be updated are never
562 invoked.
563
5642.1.15 - Invalidate quotas when (re)mounting read-write.
565
566 - Add new element itype.index.collation_rule to the ntfs inode
567 structure and set it appropriately in ntfs_read_locked_inode().
568 - Implement a new inode type "index" to allow efficient access to the
569 indices found in various system files and adapt inode handling
570 accordingly (fs/ntfs/inode.[hc]). An index inode is essentially an
571 attribute inode (NInoAttr() is true) with an attribute type of
572 AT_INDEX_ALLOCATION. As such, it is no longer allowed to call
573 ntfs_attr_iget() with an attribute type of AT_INDEX_ALLOCATION as
574 there would be no way to distinguish between normal attribute inodes
575 and index inodes. The function to obtain an index inode is
576 ntfs_index_iget() and it uses the helper function
577 ntfs_read_locked_index_inode(). Note, we do not overload
578 ntfs_attr_iget() as indices consist of multiple attributes so using
579 ntfs_attr_iget() to obtain an index inode would be confusing.
580 - Ensure that there is no overflow when doing page->index <<
581 PAGE_CACHE_SHIFT by casting page->index to s64 in fs/ntfs/aops.c.
582 - Use atomic kmap instead of kmap() in fs/ntfs/aops.c::ntfs_read_page()
583 and ntfs_read_block().
584 - Use case sensitive attribute lookups instead of case insensitive ones.
585 - Lock all page cache pages belonging to mst protected attributes while
586 accessing them to ensure we never see corrupt data while the page is
587 under writeout.
588 - Add framework for generic ntfs collation (fs/ntfs/collation.[hc]).
589 We have ntfs_is_collation_rule_supported() to check if the collation
590 rule you want to use is supported and ntfs_collation() which actually
591 collates two data items. We currently only support COLLATION_BINARY
592 and COLLATION_NTOFS_ULONG but support for other collation rules will
593 be added as the need arises.
594 - Add a new type, ntfs_index_context, to allow retrieval of an index
595 entry using the corresponding index key. To get an index context,
596 use ntfs_index_ctx_get() and to release it, use ntfs_index_ctx_put().
597 This also adds a new slab cache for the index contexts. To lookup a
598 key in an index inode, use ntfs_index_lookup(). After modifying an
599 index entry, call ntfs_index_entry_flush_dcache_page() followed by
600 ntfs_index_entry_mark_dirty() to ensure the changes are written out
601 to disk. For details see fs/ntfs/index.[hc]. Note, at present, if
602 an index entry is in the index allocation attribute rather than the
603 index root attribute it will not be written out (you will get a
604 warning message about discarded changes instead).
605 - Load the quota file ($Quota) and check if quota tracking is enabled
606 and if so, mark the quotas out of date. This causes windows to
607 rescan the volume on boot and update all quota entries.
608 - Add a set_page_dirty address space operation for ntfs_m[fs]t_aops.
609 It is simply set to __set_page_dirty_nobuffers() to make sure that
610 running set_page_dirty() on a page containing mft/ntfs records will
611 not affect the dirty state of the page buffers.
612 - Add fs/ntfs/index.c::__ntfs_index_entry_mark_dirty() which sets all
613 buffers that are inside the ntfs record in the page dirty after which
614 it sets the page dirty. This allows ->writepage to only write the
615 dirty index records rather than having to write all the records in
616 the page. Modify fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to
617 use this rather than __set_page_dirty_nobuffers().
618 - Implement fs/ntfs/aops.c::ntfs_write_mst_block() which enables the
619 writing of page cache pages belonging to mst protected attributes
620 like the index allocation attribute in directory indices and other
621 indices like $Quota/$Q, etc. This means that the quota is now marked
622 out of date on all volumes rather than only on ones where the quota
623 defaults entry is in the index root attribute of the $Quota/$Q index.
624
6252.1.14 - Fix an NFSd caused deadlock reported by several users.
626
627 - Modify fs/ntfs/ntfs_readdir() to copy the index root attribute value
628 to a buffer so that we can put the search context and unmap the mft
629 record before calling the filldir() callback. We need to do this
630 because of NFSd which calls ->lookup() from its filldir callback()
631 and this causes NTFS to deadlock as ntfs_lookup() maps the mft record
632 of the directory and since ntfs_readdir() has got it mapped already
633 ntfs_lookup() deadlocks.
634
6352.1.13 - Enable overwriting of resident files and housekeeping of system files.
636
637 - Implement writing of mft records (fs/ntfs/mft.[hc]), which includes
638 keeping the mft mirror in sync with the mft when mirrored mft records
639 are written. The functions are write_mft_record{,_nolock}(). The
640 implementation is quite rudimentary for now with lots of things not
641 implemented yet but I am not sure any of them can actually occur so
642 I will wait for people to hit each one and only then implement it.
643 - Commit open system inodes at umount time. This should make it
644 virtually impossible for sync_mft_mirror_umount() to ever be needed.
645 - Implement ->write_inode (fs/ntfs/inode.c::ntfs_write_inode()) for the
646 ntfs super operations. This gives us inode writing via the VFS inode
647 dirty code paths. Note: Access time updates are not implemented yet.
648 - Implement fs/ntfs/mft.[hc]::{,__}mark_mft_record_dirty() and make
649 fs/ntfs/aops.c::ntfs_writepage() and ntfs_commit_write() use it, thus
650 finally enabling resident file overwrite! (-8 This also includes a
651 placeholder for ->writepage (ntfs_mft_writepage()), which for now
652 just redirties the page and returns. Also, at umount time, we for
653 now throw away all mft data page cache pages after the last call to
654 ntfs_commit_inode() in the hope that all inodes will have been
655 written out by then and hence no dirty (meta)data will be lost. We
656 also check for this case and emit an error message telling the user
657 to run chkdsk.
658 - Use set_page_writeback() and end_page_writeback() in the resident
659 attribute code path of fs/ntfs/aops.c::ntfs_writepage() otherwise
660 the radix-tree tag PAGECACHE_TAG_DIRTY remains set even though the
661 page is clean.
662 - Implement ntfs_mft_writepage() so it now checks if any of the mft
663 records in the page are dirty and if so redirties the page and
664 returns. Otherwise it just returns (after doing set_page_writeback(),
665 unlock_page(), end_page_writeback() or the radix-tree tag
666 PAGECACHE_TAG_DIRTY remains set even though the page is clean), thus
667 alowing the VM to do with the page as it pleases. Also, at umount
668 time, now only throw away dirty mft (meta)data pages if dirty inodes
669 are present and ask the user to email us if they see this happening.
670 - Add functions ntfs_{clear,set}_volume_flags(), to modify the volume
671 information flags (fs/ntfs/super.c).
672 - Mark the volume dirty when (re)mounting read-write and mark it clean
673 when unmounting or remounting read-only. If any volume errors are
674 found, the volume is left marked dirty to force chkdsk to run.
675 - Add code to set the NT4 compatibility flag when (re)mounting
676 read-write for newer NTFS versions but leave it commented out for now
677 since we do not make any modifications that are NTFS 1.2 specific yet
678 and since setting this flag breaks Captive-NTFS which is not nice.
679 This code must be enabled once we start writing NTFS 1.2 specific
680 changes otherwise Windows NTFS driver might crash / cause corruption.
681
6822.1.12 - Fix the second fix to the decompression engine and some cleanups.
683
684 - Add a new address space operations struct, ntfs_mst_aops, for mst
685 protected attributes. This is because the default ntfs_aops do not
686 make sense with mst protected data and were they to write anything to
687 such an attribute they would cause data corruption so we provide
688 ntfs_mst_aops which does not have any write related operations set.
689 - Cleanup dirty ntfs inode handling (fs/ntfs/inode.[hc]) which also
690 includes an adapted ntfs_commit_inode() and an implementation of
691 ntfs_write_inode() which for now just cleans dirty inodes without
692 writing them (it does emit a warning that this is happening).
693 - Undo the second decompression engine fix (see 2.1.9 release ChangeLog
694 entry) as it was only fixing a theoretical bug but at the same time
695 it badly broke the handling of sparse and uncompressed compression
696 blocks.
697
6982.1.11 - Driver internal cleanups.
699
700 - Only build logfile.o if building the driver with read-write support.
701 - Really final white space cleanups.
702 - Use generic_ffs() instead of ffs() in logfile.c which allows the
703 log_page_size variable to be optimized by gcc into a constant.
704 - Rename uchar_t to ntfschar everywhere as uchar_t is unsigned 1-byte
705 char as defined by POSIX and as found on some systems.
706
7072.1.10 - Force read-only (re)mounting of volumes with unsupported volume flags.
708
709 - Finish off the white space cleanups (remove trailing spaces, etc).
710 - Clean up ntfs_fill_super() and ntfs_read_inode_mount() by removing
711 the kludges around the first iget(). Instead of (re)setting ->s_op
712 we have the $MFT inode set up by explicit new_inode() / set ->i_ino /
713 insert_inode_hash() / call ntfs_read_inode_mount() directly. This
714 kills the need for second super_operations and allows to return error
715 from ntfs_read_inode_mount() without resorting to ugly "poisoning"
716 tricks. (Al Viro)
717 - Force read-only (re)mounting if any of the following bits are set in
718 the volume information flags:
719 VOLUME_IS_DIRTY, VOLUME_RESIZE_LOG_FILE,
720 VOLUME_UPGRADE_ON_MOUNT, VOLUME_DELETE_USN_UNDERWAY,
721 VOLUME_REPAIR_OBJECT_ID, VOLUME_MODIFIED_BY_CHKDSK
722 To make this easier we define VOLUME_MUST_MOUNT_RO_MASK with all the
723 above bits set so the test is made easy.
724
7252.1.9 - Fix two bugs in decompression engine.
726
727 - Fix a bug where we would not always detect that we have reached the
728 end of a compression block because we were ending at minus one byte
729 which is effectively the same as being at the end. The fix is to
730 check whether the uncompressed buffer has been fully filled and if so
731 we assume we have reached the end of the compression block. A big
732 thank you to Marcin Gibuła for the bug report, the assistance in
733 tracking down the bug and testing the fix.
734 - Fix a possible bug where when a compressed read is truncated to the
735 end of the file, the offset inside the last page was not truncated.
736
7372.1.8 - Handle $MFT mirror and $LogFile, improve time handling, and cleanups.
738
739 - Use get_bh() instead of manual atomic_inc() in fs/ntfs/compress.c.
740 - Modify fs/ntfs/time.c::ntfs2utc(), get_current_ntfs_time(), and
741 utc2ntfs() to work with struct timespec instead of time_t on the
742 Linux UTC time side thus preserving the full precision of the NTFS
743 time and only loosing up to 99 nano-seconds in the Linux UTC time.
744 - Move fs/ntfs/time.c to fs/ntfs/time.h and make the time functions
745 static inline.
746 - Remove unused ntfs_dirty_inode().
747 - Cleanup super operations declaration in fs/ntfs/super.c.
748 - Wrap flush_dcache_mft_record_page() in #ifdef NTFS_RW.
749 - Add NInoTestSetFoo() and NInoTestClearFoo() macro magic to
750 fs/ntfs/inode.h and use it to declare NInoTest{Set,Clear}Dirty.
751 - Move typedefs for ntfs_attr and test_t from fs/ntfs/inode.c to
752 fs/ntfs/inode.h so they can be used elsewhere.
753 - Determine the mft mirror size as the number of mirrored mft records
754 and store it in ntfs_volume->mftmirr_size (fs/ntfs/super.c).
755 - Load the mft mirror at mount time and compare the mft records stored
756 in it to the ones in the mft. Force a read-only mount if the two do
757 not match (fs/ntfs/super.c).
758 - Fix type casting related warnings on 64-bit architectures. Thanks
759 to Meelis Roos for reporting them.
760 - Move %L to %ll as %L is floating point and %ll is integer which is
761 what we want.
762 - Read the journal ($LogFile) and determine if the volume has been
763 shutdown cleanly and force a read-only mount if not (fs/ntfs/super.c
764 and fs/ntfs/logfile.c). This is a little bit of a crude check in
765 that we only look at the restart areas and not at the actual log
766 records so that there will be a very small number of cases where we
767 think that a volume is dirty when in fact it is clean. This should
768 only affect volumes that have not been shutdown cleanly and did not
769 have any pending, non-check-pointed i/o.
770 - If the $LogFile indicates a clean shutdown and a read-write (re)mount
771 is requested, empty $LogFile by overwriting it with 0xff bytes to
772 ensure that Windows cannot cause data corruption by replaying a stale
773 journal after Linux has written to the volume.
774
7752.1.7 - Enable NFS exporting of mounted NTFS volumes.
776
777 - Set i_generation in the VFS inode from the seq_no of the NTFS inode.
778 - Make ntfs_lookup() NFS export safe, i.e. use d_splice_alias(), etc.
779 - Implement ->get_dentry() in fs/ntfs/namei.c::ntfs_get_dentry() as the
780 default doesn't allow inode number 0 which is a valid inode on NTFS
781 and even if it did allow that it uses iget() instead of ntfs_iget()
782 which makes it useless for us.
783 - Implement ->get_parent() in fs/ntfs/namei.c::ntfs_get_parent() as the
784 default just returns -EACCES which is not very useful.
785 - Define export operations (->s_export_op) for NTFS (ntfs_export_ops)
786 and set them up in the super block at mount time (super.c) this
787 allows mounted NTFS volumes to be exported via NFS.
788 - Add missing return -EOPNOTSUPP; in
789 fs/ntfs/aops.c::ntfs_commit_nonresident_write().
790 - Enforce no atime and no dir atime updates at mount/remount time as
791 they are not implemented yet anyway.
792 - Move a few assignments in fs/ntfs/attrib.c::load_attribute_list() to
793 after a NULL check. Thanks to Dave Jones for pointing this out.
794
7952.1.6 - Fix minor bug in handling of compressed directories.
796
797 - Fix bug in handling of compressed directories. A compressed
798 directory is not really compressed so when we set the ->i_blocks
799 field of a compressed directory inode we were setting it from the
800 non-existing field ni->itype.compressed.size which gave random
801 results... For directories we now always use ni->allocated_size.
802
8032.1.5 - Fix minor bug in attribute list attribute handling.
804
805 - Fix bug in attribute list handling. Actually it is not as much a bug
806 as too much protection in that we were not allowing attribute lists
807 which waste space on disk while Windows XP clearly allows it and in
808 fact creates such attribute lists so our driver was failing.
809 - Update NTFS documentation ready for 2.6 kernel release.
810
8112.1.4 - Reduce compiler requirements.
812
813 - Remove all uses of unnamed structs and unions in the driver to make
814 old and newer gcc versions happy. Makes it a bit uglier IMO but at
815 least people will stop hassling me about it.
816
8172.1.3 - Important bug fixes in corner cases.
818
819 - super.c::parse_ntfs_boot_sector(): Correct the check for 64-bit
820 clusters. (Philipp Thomas)
821 - attrib.c::load_attribute_list(): Fix bug when initialized_size is a
822 multiple of the block_size but not the cluster size. (Szabolcs
823 Szakacsits <szaka@sienet.hu>)
824
8252.1.2 - Important bug fixes aleviating the hangs in statfs.
826
827 - Fix buggy free cluster and free inode determination logic.
828
8292.1.1 - Minor updates.
830
831 - Add handling for initialized_size != data_size in compressed files.
832 - Reduce function local stack usage from 0x3d4 bytes to just noise in
833 fs/ntfs/upcase.c. (Randy Dunlap <rddunlap@osdl.ord>)
834 - Remove compiler warnings for newer gcc.
835 - Pages are no longer kmapped by mm/filemap.c::generic_file_write()
836 around calls to ->{prepare,commit}_write. Adapt NTFS appropriately
837 in fs/ntfs/aops.c::ntfs_prepare_nonresident_write() by using
838 kmap_atomic(KM_USER0).
839
8402.1.0 - First steps towards write support: implement file overwrite.
841
842 - Add configuration option for developmental write support with an
843 appropriately scary configuration help text.
844 - Initial implementation of fs/ntfs/aops.c::ntfs_writepage() and its
845 helper fs/ntfs/aops.c::ntfs_write_block(). This enables mmap(2) based
846 overwriting of existing files on ntfs. Note: Resident files are
847 only written into memory, and not written out to disk at present, so
848 avoid writing to files smaller than about 1kiB.
849 - Initial implementation of fs/ntfs/aops.c::ntfs_prepare_write(), its
850 helper fs/ntfs/aops.c::ntfs_prepare_nonresident_write() and their
851 counterparts, fs/ntfs/aops.c::ntfs_commit_write(), and
852 fs/ntfs/aops.c::ntfs_commit_nonresident_write(), respectively. Also,
853 add generic_file_write() to the ntfs file operations (fs/ntfs/file.c).
854 This enables write(2) based overwriting of existing files on ntfs.
855 Note: As with mmap(2) based overwriting, resident files are only
856 written into memory, and not written out to disk at present, so avoid
857 writing to files smaller than about 1kiB.
858 - Implement ->truncate (fs/ntfs/inode.c::ntfs_truncate()) and
859 ->setattr() (fs/ntfs/inode.c::ntfs_setattr()) inode operations for
860 files with the purpose of intercepting and aborting all i_size
861 changes which we do not support yet. ntfs_truncate() actually only
862 emits a warning message but AFAICS our interception of i_size changes
863 elsewhere means ntfs_truncate() never gets called for i_size changes.
864 It is only called from generic_file_write() when we fail in
865 ntfs_prepare_{,nonresident_}write() in order to discard any
866 instantiated buffers beyond i_size. Thus i_size is not actually
867 changed so our warning message is enough. Unfortunately it is not
868 possible to easily determine if i_size is being changed or not hence
869 we just emit an appropriately worded error message.
870
8712.0.25 - Small bug fixes and cleanups.
872
873 - Unlock the page in an out of memory error code path in
874 fs/ntfs/aops.c::ntfs_read_block().
875 - If fs/ntfs/aops.c::ntfs_read_page() is called on an uptodate page,
876 just unlock the page and return. (This can happen due to ->writepage
877 clearing PageUptodate() during write out of MstProtected()
878 attributes.
879 - Remove leaked write code again.
880
8812.0.24 - Cleanups.
882
883 - Treat BUG_ON() as ASSERT() not VERIFY(), i.e. do not use side effects
884 inside BUG_ON(). (Adam J. Richter)
885 - Split logical OR expressions inside BUG_ON() into individual BUG_ON()
886 calls for improved debugging. (Adam J. Richter)
887 - Add errors flag to the ntfs volume state, accessed via
888 NVol{,Set,Clear}Errors(vol).
889 - Do not allow read-write remounts of read-only volumes with errors.
890 - Clarify comment for ntfs file operation sendfile which was added by
891 Christoph Hellwig a while ago (just using generic_file_sendfile())
892 to say that ntfs ->sendfile is only used for the case where the
893 source data is on the ntfs partition and the destination is
894 somewhere else, i.e. nothing we need to concern ourselves with.
895 - Add generic_file_write() as our ntfs file write operation.
896
8972.0.23 - Major bug fixes (races, deadlocks, non-i386 architectures).
898
899 - Massive internal locking changes to mft record locking. Fixes lock
900 recursion and replaces the mrec_lock read/write semaphore with a
901 mutex. Also removes the now superfluous mft_count. This fixes several
902 race conditions and deadlocks, especially in the future write code.
903 - Fix ntfs over loopback for compressed files by adding an
904 optimization barrier. (gcc was screwing up otherwise ?)
905 - Miscellaneous cleanups all over the code and a fix or two in error
906 handling code paths.
907 Thanks go to Christoph Hellwig for pointing out the following two:
908 - Remove now unused function fs/ntfs/malloc.h::vmalloc_nofs().
909 - Fix ntfs_free() for ia64 and parisc by checking for VMALLOC_END, too.
910
9112.0.22 - Cleanups, mainly to ntfs_readdir(), and use C99 initializers.
912
913 - Change fs/ntfs/dir.c::ntfs_reddir() to only read/write ->f_pos once
914 at entry/exit respectively.
915 - Use C99 initializers for structures.
916 - Remove unused variable blocks from fs/ntfs/aops.c::ntfs_read_block().
917
9182.0.21 - Check for, and refuse to work with too large files/directories/volumes.
919
920 - Limit volume size at mount time to 2TiB on architectures where
921 unsigned long is 32-bits (fs/ntfs/super.c::parse_ntfs_boot_sector()).
922 This is the most we can do without overflowing the 32-bit limit of
923 the block device size imposed on us by sb_bread() and sb_getblk()
924 for the time being.
925 - Limit file/directory size at open() time to 16TiB on architectures
926 where unsigned long is 32-bits (fs/ntfs/file.c::ntfs_file_open() and
927 fs/ntfs/dir.c::ntfs_dir_open()). This is the most we can do without
928 overflowing the page cache page index.
929
9302.0.20 - Support non-resident directory index bitmaps, fix page leak in readdir.
931
932 - Move the directory index bitmap to use an attribute inode instead of
933 having special fields for it inside the ntfs inode structure. This
934 means that the index bitmaps now use the page cache for i/o, too,
935 and also as a side effect we get support for non-resident index
936 bitmaps for free.
937 - Simplify/cleanup error handling in fs/ntfs/dir.c::ntfs_readdir() and
938 fix a page leak that manifested itself in some cases.
939 - Add fs/ntfs/inode.c::ntfs_put_inode(), which we need to release the
940 index bitmap inode on the final iput().
941
9422.0.19 - Fix race condition, improvements, and optimizations in i/o interface.
943
944 - Apply block optimization added to fs/ntfs/aops.c::ntfs_read_block()
945 to fs/ntfs/compress.c::ntfs_file_read_compressed_block() as well.
946 - Drop the "file" from ntfs_file_read_compressed_block().
947 - Rename fs/ntfs/aops.c::ntfs_enb_buffer_read_async() to
948 ntfs_end_buffer_async_read() (more like the fs/buffer.c counterpart).
949 - Update ntfs_end_buffer_async_read() with the improved logic from
950 its updated counterpart fs/buffer.c::end_buffer_async_read(). Apply
951 further logic improvements to better determine when we set PageError.
952 - Update submission of buffers in fs/ntfs/aops.c::ntfs_read_block() to
953 check for the buffers being uptodate first in line with the updated
954 fs/buffer.c::block_read_full_page(). This plugs a small race
955 condition.
956
9572.0.18 - Fix race condition in reading of compressed files.
958
959 - There was a narrow window between checking a buffer head for being
960 uptodate and locking it in ntfs_file_read_compressed_block(). We now
961 lock the buffer and then check whether it is uptodate or not.
962
9632.0.17 - Cleanups and optimizations - shrinking the ToDo list.
964
965 - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to return an error
966 code and update callers, i.e. ntfs_iget(), to pass that error code
967 up instead of just using -EIO.
968 - Modifications to super.c to ensure that both mount and remount
969 cannot set any write related options when the driver is compiled
970 read-only.
971 - Optimize block resolution in fs/ntfs/aops.c::ntfs_read_block() to
972 cache the current runlist element. This should improve performance
973 when reading very large and/or very fragmented data.
974
9752.0.16 - Convert access to $MFT/$BITMAP to attribute inode API.
976
977 - Fix a stupid bug introduced in 2.0.15 where we were unmapping the
978 wrong inode in fs/ntfs/inode.c::ntfs_attr_iget().
979 - Fix debugging check in fs/ntfs/aops.c::ntfs_read_block().
980 - Convert $MFT/$BITMAP access to attribute inode API and remove all
981 remnants of the ugly mftbmp address space and operations hack. This
982 means we finally have only one readpage function as well as only one
983 async io completion handler. Yey! The mft bitmap is now just an
984 attribute inode and is accessed from vol->mftbmp_ino just as if it
985 were a normal file. Fake inodes rule. (-:
986
9872.0.15 - Fake inodes based attribute i/o via the pagecache, fixes and cleanups.
988
989 - Fix silly bug in fs/ntfs/super.c::parse_options() which was causing
990 remounts to fail when the partition had an entry in /etc/fstab and
991 the entry specified the nls= option.
992 - Apply same macro magic used in fs/ntfs/inode.h to fs/ntfs/volume.h to
993 expand all the helper functions NVolFoo(), NVolSetFoo(), and
994 NVolClearFoo().
995 - Move copyright statement from driver initialisation message to
996 module description (fs/super.c). This makes the initialisation
997 message fit on one line and fits in better with rest of kernel.
998 - Update fs/ntfs/attrib.c::map_run_list() to work on both real and
999 attribute inodes, and both for files and directories.
1000 - Implement fake attribute inodes allowing all attribute i/o to go via
1001 the page cache and to use all the normal vfs/mm functionality:
1002 - Add ntfs_attr_iget() and its helper ntfs_read_locked_attr_inode()
1003 to fs/ntfs/inode.c.
1004 - Add needed cleanup code to ntfs_clear_big_inode().
1005 - Merge address space operations for files and directories (aops.c),
1006 now just have ntfs_aops:
1007 - Rename:
1008 end_buffer_read_attr_async() -> ntfs_end_buffer_read_async(),
1009 ntfs_attr_read_block() -> ntfs_read_block(),
1010 ntfs_file_read_page() -> ntfs_readpage().
1011 - Rewrite fs/ntfs/aops.c::ntfs_readpage() to work on both real and
1012 attribute inodes, and both for files and directories.
1013 - Remove obsolete fs/ntfs/aops.c::ntfs_mst_readpage().
1014
10152.0.14 - Run list merging code cleanup, minor locking changes, typo fixes.
1016
1017 - Change fs/ntfs/super.c::ntfs_statfs() to not rely on BKL by moving
1018 the locking out of super.c::get_nr_free_mft_records() and taking and
1019 dropping the mftbmp_lock rw_semaphore in ntfs_statfs() itself.
1020 - Bring attribute runlist merging code (fs/ntfs/attrib.c) in sync with
1021 current userspace ntfs library code. This means that if a merge
1022 fails the original runlists are always left unmodified instead of
1023 being silently corrupted.
1024 - Misc typo fixes.
1025
10262.0.13 - Use iget5_locked() in preparation for fake inodes and small cleanups.
1027
1028 - Remove nr_mft_bits and the now superfluous union with nr_mft_records
1029 from ntfs_volume structure.
1030 - Remove nr_lcn_bits and the now superfluous union with nr_clusters
1031 from ntfs_volume structure.
1032 - Use iget5_locked() and friends instead of conventional iget(). Wrap
1033 the call in fs/ntfs/inode.c::ntfs_iget() and update callers of iget()
1034 to use ntfs_iget(). Leave only one iget() call at mount time so we
1035 don't need an ntfs_iget_mount().
1036 - Change fs/ntfs/inode.c::ntfs_new_extent_inode() to take mft_no as an
1037 additional argument.
1038
10392.0.12 - Initial cleanup of address space operations following 2.0.11 changes.
1040
1041 - Merge fs/ntfs/aops.c::end_buffer_read_mst_async() and
1042 fs/ntfs/aops.c::end_buffer_read_file_async() into one function
1043 fs/ntfs/aops.c::end_buffer_read_attr_async() using NInoMstProtected()
1044 to determine whether to apply mst fixups or not.
1045 - Above change allows merging fs/ntfs/aops.c::ntfs_file_read_block()
1046 and fs/ntfs/aops.c::ntfs_mst_readpage() into one function
1047 fs/ntfs/aops.c::ntfs_attr_read_block(). Also, create a tiny wrapper
1048 fs/ntfs/aops.c::ntfs_mst_readpage() to transform the parameters from
1049 the VFS readpage function prototype to the ntfs_attr_read_block()
1050 function prototype.
1051
10522.0.11 - Initial preparations for fake inode based attribute i/o.
1053
1054 - Move definition of ntfs_inode_state_bits to fs/ntfs/inode.h and
1055 do some macro magic (adapted from include/linux/buffer_head.h) to
1056 expand all the helper functions NInoFoo(), NInoSetFoo(), and
1057 NInoClearFoo().
1058 - Add new flag to ntfs_inode_state_bits: NI_Sparse.
1059 - Add new fields to ntfs_inode structure to allow use of fake inodes
1060 for attribute i/o: type, name, name_len. Also add new state bits:
1061 NI_Attr, which, if set, indicates the inode is a fake inode, and
1062 NI_MstProtected, which, if set, indicates the attribute uses multi
1063 sector transfer protection, i.e. fixups need to be applied after
1064 reads and before/after writes.
1065 - Rename fs/ntfs/inode.c::ntfs_{new,clear,destroy}_inode() to
1066 ntfs_{new,clear,destroy}_extent_inode() and update callers.
1067 - Use ntfs_clear_extent_inode() in fs/ntfs/inode.c::__ntfs_clear_inode()
1068 instead of ntfs_destroy_extent_inode().
1069 - Cleanup memory deallocations in {__,}ntfs_clear_{,big_}inode().
1070 - Make all operations on ntfs inode state bits use the NIno* functions.
1071 - Set up the new ntfs inode fields and state bits in
1072 fs/ntfs/inode.c::ntfs_read_inode() and add appropriate cleanup of
1073 allocated memory to __ntfs_clear_inode().
1074 - Cleanup ntfs_inode structure a bit for better ordering of elements
1075 w.r.t. their size to allow better packing of the structure in memory.
1076
10772.0.10 - There can only be 2^32 - 1 inodes on an NTFS volume.
1078
1079 - Add check at mount time to verify that the number of inodes on the
1080 volume does not exceed 2^32 - 1, which is the maximum allowed for
1081 NTFS according to Microsoft.
1082 - Change mft_no member of ntfs_inode structure to be unsigned long.
1083 Update all users. This makes ntfs_inode->mft_no just a copy of struct
1084 inode->i_ino. But we can't just always use struct inode->i_ino and
1085 remove mft_no because extent inodes do not have an attached struct
1086 inode.
1087
10882.0.9 - Decompression engine now uses a single buffer and other cleanups.
1089
1090 - Change decompression engine to use a single buffer protected by a
1091 spin lock instead of per-CPU buffers. (Rusty Russell)
1092 - Do not update cb_pos when handling a partial final page during
1093 decompression of a sparse compression block, as the value is later
1094 reset without being read/used. (Rusty Russell)
1095 - Switch to using the new KM_BIO_SRC_IRQ for atomic kmap()s. (Andrew
1096 Morton)
1097 - Change buffer size in ntfs_readdir()/ntfs_filldir() to use
1098 NLS_MAX_CHARSET_SIZE which makes the buffers almost 1kiB each but
1099 it also makes everything safer so it is a good thing.
1100 - Miscellaneous minor cleanups to comments.
1101
11022.0.8 - Major updates for handling of case sensitivity and dcache aliasing.
1103
1104 Big thanks go to Al Viro and other inhabitants of #kernel for investing
1105 their time to discuss the case sensitivity and dcache aliasing issues.
1106
1107 - Remove unused source file fs/ntfs/attraops.c.
1108 - Remove show_inodes mount option(s), thus dropping support for
1109 displaying of short file names.
1110 - Remove deprecated mount option posix.
1111 - Restore show_sys_files mount option.
1112 - Add new mount option case_sensitive, to determine if the driver
1113 treats file names as case sensitive or not. If case sensitive, create
1114 file names in the POSIX namespace. Otherwise create file names in the
1115 LONG/WIN32 namespace. Note, files remain accessible via their short
1116 file name, if it exists.
1117 - Remove really dumb logic bug in boot sector recovery code.
1118 - Fix dcache aliasing issues wrt short/long file names via changes
1119 to fs/ntfs/dir.c::ntfs_lookup_inode_by_name() and
1120 fs/ntfs/namei.c::ntfs_lookup():
1121 - Add additional argument to ntfs_lookup_inode_by_name() in which we
1122 return information about the matching file name if the case is not
1123 matching or the match is a short file name. See comments above the
1124 function definition for details.
1125 - Change ntfs_lookup() to only create dcache entries for the correctly
1126 cased file name and only for the WIN32 namespace counterpart of DOS
1127 namespace file names. This ensures we have only one dentry per
1128 directory and also removes all dcache aliasing issues between short
1129 and long file names once we add write support. See comments above
1130 function for details.
1131 - Fix potential 1 byte overflow in fs/ntfs/unistr.c::ntfs_ucstonls().
1132
11332.0.7 - Minor cleanups and updates for changes in core kernel code.
1134
1135 - Remove much of the NULL struct element initializers.
1136 - Various updates to make compatible with recent kernels.
1137 - Remove defines of MAX_BUF_PER_PAGE and include linux/buffer_head.h
1138 in fs/ntfs/ntfs.h instead.
1139 - Remove no longer needed KERNEL_VERSION checks. We are now in the
1140 kernel proper so they are no longer needed.
1141
11422.0.6 - Major bugfix to make compatible with other kernel changes.
1143
1144 - Initialize the mftbmp address space properly now that there are more
1145 fields in the struct address_space. This was leading to hangs and
1146 oopses on umount since 2.5.12 because of changes to other parts of
1147 the kernel. We probably want a kernel generic init_address_space()
1148 function...
1149 - Drop BKL from ntfs_readdir() after consultation with Al Viro. The
1150 only caller of ->readdir() is vfs_readdir() which holds i_sem during
1151 the call, and i_sem is sufficient protection against changes in the
1152 directory inode (including ->i_size).
1153 - Use generic_file_llseek() for directories (as opposed to
1154 default_llseek()) as this downs i_sem instead of the BKL which is
1155 what we now need for exclusion against ->f_pos changes considering we
1156 no longer take the BKL in ntfs_readdir().
1157
11582.0.5 - Major bugfix. Buffer overflow in extent inode handling.
1159
1160 - No need to set old blocksize in super.c::ntfs_fill_super() as the
1161 VFS does so via invocation of deactivate_super() calling
1162 fs->fill_super() calling block_kill_super() which does it.
1163 - BKL moved from VFS into dir.c::ntfs_readdir(). (Linus Torvalds)
1164 -> Do we really need it? I don't think so as we have exclusion on
1165 the directory ntfs_inode rw_semaphore mrec_lock. We mmight have to
1166 move the ->f_pos accesses under the mrec_lock though. Check this...
1167 - Fix really, really, really stupid buffer overflow in extent inode
1168 handling in mft.c::map_extent_mft_record().
1169
11702.0.4 - Cleanups and updates for kernel 2.5.11.
1171
1172 - Add documentation on how to use the MD driver to be able to use NTFS
1173 stripe and volume sets in Linux and generally cleanup documentation
1174 a bit.
1175 Remove all uses of kdev_t in favour of struct block_device *:
1176 - Change compress.c::ntfs_file_read_compressed_block() to use
1177 sb_getblk() instead of getblk().
1178 - Change super.c::ntfs_fill_super() to use bdev_hardsect_size() instead
1179 of get_hardsect_size().
1180 - No need to get old blocksize in super.c::ntfs_fill_super() as
1181 fs/super.c::get_sb_bdev() already does this.
1182 - Set bh->b_bdev instead of bh->b_dev throughout aops.c.
1183
11842.0.3 - Small bug fixes, cleanups, and performance improvements.
1185
1186 - Remove some dead code from mft.c.
1187 - Optimize readpage and read_block functions throughout aops.c so that
1188 only initialized blocks are read. Non-initialized ones have their
1189 buffer head mapped, zeroed, and set up to date, without scheduling
1190 any i/o. Thanks to Al Viro for advice on how to avoid the device i/o.
1191 Thanks go to Andrew Morton for spotting the below:
1192 - Fix buglet in allocate_compression_buffers() error code path.
1193 - Call flush_dcache_page() after modifying page cache page contents in
1194 ntfs_file_readpage().
1195 - Check for existence of page buffers throughout aops.c before calling
1196 create_empty_buffers(). This happens when an I/O error occurs and the
1197 read is retried. (It also happens once writing is implemented so that
1198 needed doing anyway but I had left it for later...)
1199 - Don't BUG_ON() uptodate and/or mapped buffers throughout aops.c in
1200 readpage and read_block functions. Reasoning same as above (i.e. I/O
1201 error retries and future write code paths.)
1202
12032.0.2 - Minor updates and cleanups.
1204
1205 - Cleanup: rename mst.c::__post_read_mst_fixup to post_write_mst_fixup
1206 and cleanup the code a bit, removing the unused size parameter.
1207 - Change default fmask to 0177 and update documentation.
1208 - Change attrib.c::get_attr_search_ctx() to return the search context
1209 directly instead of taking the address of a pointer. A return value
1210 of NULL means the allocation failed. Updated all callers
1211 appropriately.
1212 - Update to 2.5.9 kernel (preserving backwards compatibility) by
1213 replacing all occurences of page->buffers with page_buffers(page).
1214 - Fix minor bugs in runlist merging, also minor cleanup.
1215 - Updates to bootsector layout and mft mirror contents descriptions.
1216 - Small bug fix in error detection in unistr.c and some cleanups.
1217 - Grow name buffer allocations in unistr.c in aligned mutlipled of 64
1218 bytes.
1219
12202.0.1 - Minor updates.
1221
1222 - Make default umask correspond to documentation.
1223 - Improve documentation.
1224 - Set default mode to include execute bit. The {u,f,d}mask can be used
1225 to take it away if desired. This allows binaries to be executed from
1226 a mounted ntfs partition.
1227
12282.0.0 - New version number. Remove TNG from the name. Now in the kernel.
1229
1230 - Add kill_super, just keeping up with the vfs changes in the kernel.
1231 - Repeat some changes from tng-0.0.8 that somehow got lost on the way
1232 from the CVS import into BitKeeper.
1233 - Begin to implement proper handling of allocated_size vs
1234 initialized_size vs data_size (i.e. i_size). Done are
1235 mft.c::ntfs_mft_readpage(), aops.c::end_buffer_read_index_async(),
1236 and attrib.c::load_attribute_list().
1237 - Lock the runlist in attrib.c::load_attribute_list() while using it.
1238 - Fix memory leak in ntfs_file_read_compressed_block() and generally
1239 clean up compress.c a little, removing some uncommented/unused debug
1240 code.
1241 - Tidy up dir.c a little bit.
1242 - Don't bother getting the runlist in inode.c::ntfs_read_inode().
1243 - Merge mft.c::ntfs_mft_readpage() and aops.c::ntfs_index_readpage()
1244 creating aops.c::ntfs_mst_readpage(), improving the handling of
1245 holes and overflow in the process and implementing the correct
1246 equivalent of ntfs_file_get_block() in ntfs_mst_readpage() itself.
1247 I am aiming for correctness at the moment. Modularisation can come
1248 later.
1249 - Rename aops.c::end_buffer_read_index_async() to
1250 end_buffer_read_mst_async() and optimize the overflow checking and
1251 handling.
1252 - Use the host of the mftbmp address space mapping to hold the ntfs
1253 volume. This is needed so the async i/o completion handler can
1254 retrieve a pointer to the volume. Hopefully this will not cause
1255 problems elsewhere in the kernel... Otherwise will need to use a
1256 fake inode.
1257 - Complete implementation of proper handling of allocated_size vs
1258 initialized_size vs data_size (i.e. i_size) in whole driver.
1259 Basically aops.c is now completely rewritten.
1260 - Change NTFS driver name to just NTFS and set version number to 2.0.0
1261 to make a clear distinction from the old driver which is still on
1262 version 1.1.22.
1263
1264tng-0.0.8 - 08/03/2002 - Now using BitKeeper, http://linux-ntfs.bkbits.net/
1265
1266 - Replace bdevname(sb->s_dev) with sb->s_id.
1267 - Remove now superfluous new-line characters in all callers of
1268 ntfs_debug().
1269 - Apply kludge in ntfs_read_inode(), setting i_nlink to 1 for
1270 directories. Without this the "find" utility gets very upset which is
1271 fair enough as Linux/Unix do not support directory hard links.
1272 - Further runlist merging work. (Richard Russon)
1273 - Backwards compatibility for gcc-2.95. (Richard Russon)
1274 - Update to kernel 2.5.5-pre1 and rediff the now tiny patch.
Anton Altaparmakovc002f422005-02-03 12:02:56 +00001275 - Convert to new filesystem declaration using ->ntfs_get_sb() and
Linus Torvalds1da177e2005-04-16 15:20:36 -07001276 replacing ntfs_read_super() with ntfs_fill_super().
1277 - Set s_maxbytes to MAX_LFS_FILESIZE to avoid page cache page index
1278 overflow on 32-bit architectures.
1279 - Cleanup upcase loading code to use ntfs_(un)map_page().
1280 - Disable/reenable preemtion in critical sections of compession engine.
1281 - Replace device size determination in ntfs_fill_super() with
1282 sb->s_bdev->bd_inode->i_size (in bytes) and remove now superfluous
1283 function super.c::get_nr_blocks().
1284 - Implement a mount time option (show_inodes) allowing choice of which
1285 types of inode names readdir() returns and modify ntfs_filldir()
1286 accordingly. There are several parameters to show_inodes:
1287 system: system files
1288 win32: long file names (including POSIX file names) [DEFAULT]
1289 long: same as win32
1290 dos: short file names only (excluding POSIX file names)
1291 short: same as dos
1292 posix: same as both win32 and dos
1293 all: all file names
1294 Note that the options are additive, i.e. specifying:
1295 -o show_inodes=system,show_inodes=win32,show_inodes=dos
1296 is the same as specifying:
1297 -o show_inodes=all
1298 Note that the "posix" and "all" options will show all directory
1299 names, BUT the link count on each directory inode entry is set to 1,
1300 due to Linux not supporting directory hard links. This may well
1301 confuse some userspace applications, since the directory names will
1302 have the same inode numbers. Thus it is NOT advisable to use the
1303 "posix" or "all" options. We provide them only for completeness sake.
1304 - Add copies of allocated_size, initialized_size, and compressed_size to
1305 the ntfs inode structure and set them up in
1306 inode.c::ntfs_read_inode(). These reflect the unnamed data attribute
1307 for files and the index allocation attribute for directories.
1308 - Add copies of allocated_size and initialized_size to ntfs inode for
1309 $BITMAP attribute of large directories and set them up in
1310 inode.c::ntfs_read_inode().
1311 - Add copies of allocated_size and initialized_size to ntfs volume for
1312 $BITMAP attribute of $MFT and set them up in
1313 super.c::load_system_files().
1314 - Parse deprecated ntfs driver options (iocharset, show_sys_files,
1315 posix, and utf8) and tell user what the new options to use are. Note
1316 we still do support them but they will be removed with kernel 2.7.x.
1317 - Change all occurences of integer long long printf formatting to hex
1318 as printk() will not support long long integer format if/when the
1319 div64 patch goes into the kernel.
1320 - Make slab caches have stable names and change the names to what they
1321 were intended to be. These changes are required/made possible by the
1322 new slab cache name handling which removes the length limitation by
1323 requiring the caller of kmem_cache_create() to supply a stable name
1324 which is then referenced but not copied.
1325 - Rename run_list structure to run_list_element and create a new
1326 run_list structure containing a pointer to a run_list_element
1327 structure and a read/write semaphore. Adapt all users of runlists
1328 to new scheme and take and release the lock as needed. This fixes a
1329 nasty race as the run_list changes even when inodes are locked for
1330 reading and even when the inode isn't locked at all, so we really
1331 needed the serialization. We use a semaphore rather than a spinlock
1332 as memory allocations can sleep and doing everything GFP_ATOMIC
1333 would be silly.
1334 - Cleanup read_inode() removing all code checking for lowest_vcn != 0.
1335 This can never happen due to the nature of lookup_attr() and how we
1336 support attribute lists. If it did happen it would imply the inode
1337 being corrupt.
1338 - Check for lowest_vcn != 0 in ntfs_read_inode() and mark the inode as
1339 bad if found.
1340 - Update to 2.5.6-pre2 changes in struct address_space.
1341 - Use parent_ino() when accessing d_parent inode number in dir.c.
1342 - Import Sourceforge CVS repository into BitKeeper repository:
1343 http://linux-ntfs.bkbits.net/ntfs-tng-2.5
1344 - Update fs/Makefile, fs/Config.help, fs/Config.in, and
1345 Documentation/filesystems/ntfs.txt for NTFS TNG.
1346 - Create kernel configuration option controlling whether debugging
1347 is enabled or not.
1348 - Add the required export of end_buffer_io_sync() from the patches
1349 directory to the kernel code.
1350 - Update inode.c::ntfs_show_options() with show_inodes mount option.
1351 - Update errors mount option.
1352
1353tng-0.0.7 - 13/02/2002 - The driver is now feature complete for read-only!
1354
1355 - Cleanup mft.c and it's debug/error output in particular. Fix a minor
1356 bug in mapping of extent inodes. Update all the comments to fit all
1357 the recent code changes.
1358 - Modify vcn_to_lcn() to cope with entirely unmapped runlists.
1359 - Cleanups in compress.c, mostly comments and folding help.
1360 - Implement attrib.c::map_run_list() as a generic helper.
1361 - Make compress.c::ntfs_file_read_compressed_block() use map_run_list()
1362 thus making code shorter and enabling attribute list support.
1363 - Cleanup incorrect use of [su]64 with %L printf format specifier in
1364 all source files. Type casts to [unsigned] long long added to correct
1365 the mismatches (important for architectures which have long long not
1366 being 64 bits).
1367 - Merge async io completion handlers for directory indexes and $MFT
1368 data into one by setting the index_block_size{_bits} of the ntfs
1369 inode for $MFT to the mft_record_size{_bits} of the ntfs_volume.
1370 - Cleanup aops.c, update comments.
1371 - Make ntfs_file_get_block() use map_run_list() so all files now
1372 support attribute lists.
1373 - Make ntfs_dir_readpage() almost verbatim copy of
1374 block_read_full_page() by using ntfs_file_get_block() with only real
1375 difference being the use of our own async io completion handler
1376 rather than the default one, thus reducing the amount of code and
1377 automatically enabling attribute list support for directory indices.
1378 - Fix bug in load_attribute_list() - forgot to call brelse in error
1379 code path.
1380 - Change parameters to find_attr() and lookup_attr(). We no longer
1381 pass in the upcase table and its length. These can be gotten from
1382 ctx->ntfs_ino->vol->upcase{_len}. Update all callers.
1383 - Cleanups in attrib.c.
1384 - Implement merging of runlists, attrib.c::merge_run_lists() and its
1385 helpers. (Richard Russon)
1386 - Attribute lists part 2, attribute extents and multi part runlists:
1387 enable proper support for LCN_RL_NOT_MAPPED and automatic mapping of
1388 further runlist parts via attrib.c::map_run_list().
1389 - Tiny endianness bug fix in decompress_mapping_pairs().
1390
1391tng-0.0.6 - Encrypted directories, bug fixes, cleanups, debugging enhancements.
1392
1393 - Enable encrypted directories. (Their index root is marked encrypted
1394 to indicate that new files in that directory should be created
1395 encrypted.)
1396 - Fix bug in NInoBmpNonResident() macro. (Cut and paste error.)
1397 - Enable $Extend system directory. Most (if not all) extended system
1398 files do not have unnamed data attributes so ntfs_read_inode() had to
1399 special case them but that is ok, as the special casing recovery
1400 happens inside an error code path so there is zero slow down in the
1401 normal fast path. The special casing is done by introducing a new
1402 function inode.c::ntfs_is_extended_system_file() which checks if any
1403 of the hard links in the inode point to $Extend as being their parent
1404 directory and if they do we assume this is an extended system file.
1405 - Create a sysctl/proc interface to allow {dis,en}abling of debug output
1406 when compiled with -DDEBUG. Default is debug messages to be disabled.
1407 To enable them, one writes a non-zero value to /proc/sys/fs/ntfs-debug
1408 (if /proc is enabled) or uses sysctl(2) to effect the same (if sysctl
1409 interface is enabled). Inspired by old ntfs driver.
1410 - Add debug_msgs insmod/kernel boot parameter to set whether debug
1411 messages are {dis,en}abled. This is useful to enable debug messages
1412 during ntfs initialization and is the only way to activate debugging
1413 when the sysctl interface is not enabled.
1414 - Cleanup debug output in various places.
1415 - Remove all dollar signs ($) from the source (except comments) to
1416 enable compilation on architectures whose gcc compiler does not
1417 support dollar signs in the names of variables/constants. Attribute
1418 types now start with AT_ instead of $ and $I30 is now just I30.
1419 - Cleanup ntfs_lookup() and add consistency check of sequence numbers.
1420 - Load complete runlist for $MFT/$BITMAP during mount and cleanup
1421 access functions. This means we now cope with $MFT/$BITMAP being
1422 spread accross several mft records.
1423 - Disable modification of mft_zone_multiplier on remount. We can always
1424 reenable this later on if we really want to, but we will need to make
1425 sure we readjust the mft_zone size / layout accordingly.
1426
1427tng-0.0.5 - Modernize for 2.5.x and further in line-ing with Al Viro's comments.
1428
1429 - Use sb_set_blocksize() instead of set_blocksize() and verify the
1430 return value.
1431 - Use sb_bread() instead of bread() throughout.
1432 - Add index_vcn_size{_bits} to ntfs_inode structure to store the size
1433 of a directory index block vcn. Apply resulting simplifications in
1434 dir.c everywhere.
1435 - Fix a small bug somewhere (but forgot what it was).
1436 - Change ntfs_{debug,error,warning} to enable gcc to do type checking
1437 on the printf-format parameter list and fix bugs reported by gcc
1438 as a result. (Richard Russon)
1439 - Move inode allocation strategy to Al's new stuff but maintain the
1440 divorce of ntfs_inode from struct inode. To achieve this we have two
1441 separate slab caches, one for big ntfs inodes containing a struct
1442 inode and pure ntfs inodes and at the same time fix some faulty
1443 error code paths in ntfs_read_inode().
1444 - Show mount options in proc (inode.c::ntfs_show_options()).
1445
1446tng-0.0.4 - Big changes, getting in line with Al Viro's comments.
1447
1448 - Modified (un)map_mft_record functions to be common for read and write
1449 case. To specify which is which, added extra parameter at front of
1450 parameter list. Pass either READ or WRITE to this, each has the
1451 obvious meaning.
1452 - General cleanups to allow for easier folding in vi.
1453 - attrib.c::decompress_mapping_pairs() now accepts the old runlist
1454 argument, and invokes attrib.c::merge_run_lists() to merge the old
1455 and the new runlists.
1456 - Removed attrib.c::find_first_attr().
1457 - Implemented loading of attribute list and complete runlist for $MFT.
1458 This means we now cope with $MFT being spread across several mft
1459 records.
1460 - Adapt to 2.5.2-pre9 and the changed create_empty_buffers() syntax.
1461 - Adapt major/minor/kdev_t/[bk]devname stuff to new 2.5.x kernels.
1462 - Make ntfs_volume be allocated via kmalloc() instead of using a slab
1463 cache. There are too little ntfs_volume structures at any one time
1464 to justify a private slab cache.
1465 - Fix bogus kmap() use in async io completion. Now use kmap_atomic().
1466 Use KM_BIO_IRQ on advice from IRC/kernel...
1467 - Use ntfs_map_page() in map_mft_record() and create ->readpage method
1468 for reading $MFT (ntfs_mft_readpage). In the process create dedicated
1469 address space operations (ntfs_mft_aops) for $MFT inode mapping. Also
1470 removed the now superfluous exports from the kernel core patch.
1471 - Fix a bug where kfree() was used insted of ntfs_free().
1472 - Change map_mft_record() to take ntfs_inode as argument instead of
1473 vfs inode. Dito for unmap_mft_record(). Adapt all callers.
1474 - Add pointer to ntfs_volume to ntfs_inode.
1475 - Add mft record number and sequence number to ntfs_inode. Stop using
1476 i_ino and i_generation for in-driver purposes.
1477 - Implement attrib.c::merge_run_lists(). (Richard Russon)
1478 - Remove use of proper inodes by extent inodes. Move i_ino and
1479 i_generation to ntfs_inode to do this. Apply simplifications that
1480 result and remove iget_no_wait(), etc.
1481 - Pass ntfs_inode everywhere in the driver (used to be struct inode).
1482 - Add reference counting in ntfs_inode for the ntfs inode itself and
1483 for the mapped mft record.
1484 - Extend mft record mapping so we can (un)map extent mft records (new
1485 functions (un)map_extent_mft_record), and so mappings are reference
1486 counted and don't have to happen twice if already mapped - just ref
1487 count increases.
1488 - Add -o iocharset as alias to -o nls for backwards compatibility.
1489 - The latest core patch is now tiny. In fact just a single additional
1490 export is necessary over the base kernel.
1491
1492tng-0.0.3 - Cleanups, enhancements, bug fixes.
1493
1494 - Work on attrib.c::decompress_mapping_pairs() to detect base extents
1495 and setup the runlist appropriately using knowledge provided by the
1496 sizes in the base attribute record.
1497 - Balance the get_/put_attr_search_ctx() calls so we don't leak memory
1498 any more.
1499 - Introduce ntfs_malloc_nofs() and ntfs_free() to allocate/free a single
1500 page or use vmalloc depending on the amount of memory requested.
1501 - Cleanup error output. The __FUNCTION__ "(): " is now added
1502 automatically. Introduced a new header file debug.h to support this
1503 and also moved ntfs_debug() function into it.
1504 - Make reading of compressed files more intelligent and especially get
1505 rid of the vmalloc_nofs() from readpage(). This now uses per CPU
1506 buffers (allocated at first mount with cluster size <= 4kiB and
1507 deallocated on last umount with cluster size <= 4kiB), and
1508 asynchronous io for the compressed data using a list of buffer heads.
1509 Er, we use synchronous io as async io only works on whole pages
1510 covered by buffers and not on individual buffer heads...
1511 - Bug fix for reading compressed files with sparse compression blocks.
1512
1513tng-0.0.2 - Now handles larger/fragmented/compressed volumes/files/dirs.
1514
1515 - Fixed handling of directories when cluster size exceeds index block
1516 size.
1517 - Hide DOS only name space directory entries from readdir() but allow
1518 them in lookup(). This should fix the problem that Linux doesn't
1519 support directory hard links, while still allowing access to entries
1520 via their short file name. This also has the benefit of mimicking
1521 what Windows users are used to, so it is the ideal solution.
1522 - Implemented sync_page everywhere so no more hangs in D state when
1523 waiting for a page.
1524 - Stop using bforget() in favour of brelse().
1525 - Stop locking buffers unnecessarily.
1526 - Implemented compressed files (inode->mapping contains uncompressed
1527 data, raw compressed data is currently bread() into a vmalloc()ed
1528 memory buffer).
1529 - Enable compressed directories. (Their index root is marked compressed
1530 to indicate that new files in that directory should be created
1531 compressed.)
1532 - Use vsnprintf rather than vsprintf in the ntfs_error and ntfs_warning
1533 functions. (Thanks to Will Dyson for pointing this out.)
1534 - Moved the ntfs_inode and ntfs_volume (the former ntfs_inode_info and
1535 ntfs_sb_info) out of the common inode and super_block structures and
1536 started using the generic_ip and generic_sbp pointers instead. This
1537 makes ntfs entirely private with respect to the kernel tree.
1538 - Detect compiler version and abort with error message if gcc less than
1539 2.96 is used.
1540 - Fix bug in name comparison function in unistr.c.
1541 - Implement attribute lists part 1, the infrastructure: search contexts
1542 and operations, find_external_attr(), lookup_attr()) and make the
1543 code use the infrastructure.
1544 - Fix stupid buffer overflow bug that became apparent on larger run
1545 list containing attributes.
1546 - Fix bugs in readdir() that became apparent on larger directories.
1547
1548 The driver is now really useful and survives the test
1549 find . -type f -exec md5sum "{}" \;
1550 without any error messages on a over 1GiB sized partition with >16k
1551 files on it, including compressed files and directories and many files
1552 and directories with attribute lists.
1553
1554tng-0.0.1 - The first useful version.
1555
1556 - Added ntfs_lookup().
1557 - Added default upcase generation and handling.
1558 - Added compile options to be shown on module init.
1559 - Many bug fixes that were "hidden" before.
1560 - Update to latest kernel.
1561 - Added ntfs_readdir().
1562 - Added file operations for mmap(), read(), open() and llseek(). We just
1563 use the generic ones. The whole point of going through implementing
1564 readpage() methods and where possible get_block() call backs is that
1565 this allows us to make use of the generic high level methods provided
1566 by the kernel.
1567
1568 The driver is now actually useful! Yey. (-: It undoubtedly has got bugs
1569 though and it doesn't implement accesssing compressed files yet. Also,
1570 accessing files with attribute list attributes is not implemented yet
Anton Altaparmakovc002f422005-02-03 12:02:56 +00001571 either. But for small or simple filesystems it should work and allow
Linus Torvalds1da177e2005-04-16 15:20:36 -07001572 you to list directories, use stat on directory entries and the file
1573 system, open, read, mmap and llseek around in files. A big mile stone
1574 has been reached!
1575
1576tng-0.0.0 - Initial version tag.
1577
1578 Initial driver implementation. The driver can mount and umount simple
Anton Altaparmakovc002f422005-02-03 12:02:56 +00001579 NTFS filesystems (i.e. ones without attribute lists in the system
Linus Torvalds1da177e2005-04-16 15:20:36 -07001580 files). If the mount fails there might be problems in the error handling
1581 code paths, so be warned. Otherwise it seems to be loading the system
1582 files nicely and the mft record read mapping/unmapping seems to be
1583 working nicely, too. Proof of inode metadata in the page cache and non-
1584 resident file unnamed stream data in the page cache concepts is thus
1585 complete.