blob: 9d8ffa89e2c21f3d334811fdd253b12eb6a615ba [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001ToDo/Notes:
2 - Find and fix bugs.
Anton Altaparmakov98b27032005-10-11 15:40:40 +01003 - The only places in the kernel where a file is resized are
Anton Altaparmakov3672b632006-02-24 09:55:07 +00004 ntfs_file_write*() and ntfs_truncate() for both of which i_mutex is
Anton Altaparmakov98b27032005-10-11 15:40:40 +01005 held. Just have to be careful in read-/writepage and other helpers
Anton Altaparmakov3672b632006-02-24 09:55:07 +00006 not running under i_mutex that we play nice. Also need to be careful
Anton Altaparmakov98b27032005-10-11 15:40:40 +01007 with initialized_size extension in ntfs_file_write*() and writepage.
8 UPDATE: The only things that need to be checked are the compressed
9 write and the other attribute resize/write cases like index
10 attributes, etc. For now none of these are implemented so are safe.
11 - Implement filling in of holes in aops.c::ntfs_writepage() and its
12 helpers.
Linus Torvalds1da177e2005-04-16 15:20:36 -070013 - Implement mft.c::sync_mft_mirror_umount(). We currently will just
14 leave the volume dirty on umount if the final iput(vol->mft_ino)
15 causes a write of any mirrored mft records due to the mft mirror
16 inode having been discarded already. Whether this can actually ever
17 happen is unclear however so it is worth waiting until someone hits
18 the problem.
19 - Enable the code for setting the NT4 compatibility flag when we start
20 making NTFS 1.2 specific modifications.
21
Anton Altaparmakov3672b632006-02-24 09:55:07 +0000222.1.26 - Minor bug fixes and updates.
23
Anton Altaparmakov78af34f2006-02-24 10:32:33 +000024 - Fix a potential overflow in file.c where a cast to s64 was missing in
25 a left shift of a page index.
26 - The struct inode has had its i_sem semaphore changed to a mutex named
27 i_mutex.
Anton Altaparmakov3672b632006-02-24 09:55:07 +000028 - We have struct kmem_cache now so use it instead of the typedef
29 kmem_cache_t. (Pekka Enberg)
Anton Altaparmakov78af34f2006-02-24 10:32:33 +000030 - Implement support for sector sizes above 512 bytes (up to the maximum
31 supported by NTFS which is 4096 bytes).
Anton Altaparmakov1cf3109f2006-02-24 10:48:14 +000032 - Do more detailed reporting of why we cannot mount read-write by
33 special casing the VOLUME_MODIFIED_BY_CHKDSK flag.
Anton Altaparmakov3672b632006-02-24 09:55:07 +000034 - Miscellaneous updates to layout.h.
35 - Cope with attribute list attribute having invalid flags. Windows
36 copes with this and even chkdsk does not detect or fix this so we
37 have to cope with it, too. Thanks to Pawel Kot for reporting the
38 problem.
39
Anton Altaparmakov98b27032005-10-11 15:40:40 +0100402.1.25 - (Almost) fully implement write(2) and truncate(2).
Anton Altaparmakovfd9d6362005-10-04 13:44:48 +010041
Anton Altaparmakov511bea52005-10-04 14:24:21 +010042 - Change ntfs_map_runlist_nolock(), ntfs_attr_find_vcn_nolock() and
43 {__,}ntfs_cluster_free() to also take an optional attribute search
44 context as argument. This allows calling these functions with the
45 mft record mapped. Update all callers.
46 - Fix potential deadlock in ntfs_mft_data_extend_allocation_nolock()
47 error handling by passing in the active search context when calling
48 ntfs_cluster_free().
Anton Altaparmakovfc0fa7d2005-10-04 14:36:56 +010049 - Change ntfs_cluster_alloc() to take an extra boolean parameter
50 specifying whether the cluster are being allocated to extend an
51 attribute or to fill a hole.
52 - Change ntfs_attr_make_non_resident() to call ntfs_cluster_alloc()
53 with @is_extension set to TRUE and remove the runlist terminator
54 fixup code as this is now done by ntfs_cluster_alloc().
Anton Altaparmakov8925d4f2005-10-04 14:48:20 +010055 - Change ntfs_attr_make_non_resident to take the attribute value size
56 as an extra parameter. This is needed since we need to know the size
57 before we can map the mft record and our callers always know it. The
58 reason we cannot simply read the size from the vfs inode i_size is
59 that this is not necessarily uptodate. This happens when
60 ntfs_attr_make_non_resident() is called in the ->truncate call path.
Anton Altaparmakov2a6fc4e12005-10-04 14:57:15 +010061 - Fix ntfs_attr_make_non_resident() to update the vfs inode i_blocks
62 which is zero for a resident attribute but should no longer be zero
63 once the attribute is non-resident as it then has real clusters
64 allocated.
Anton Altaparmakov2d868292005-10-04 15:18:56 +010065 - Add fs/ntfs/attrib.[hc]::ntfs_attr_extend_allocation(), a function to
66 extend the allocation of an attributes. Optionally, the data size,
67 but not the initialized size can be extended, too.
Anton Altaparmakovdd072332005-10-04 15:39:02 +010068 - Implement fs/ntfs/inode.[hc]::ntfs_truncate(). It only supports
Anton Altaparmakov98b27032005-10-11 15:40:40 +010069 uncompressed and unencrypted files and it never creates sparse files
70 at least for the moment (making a file sparse requires us to modify
71 its directory entries and we do not support directory operations at
72 the moment). Also, support for highly fragmented files, i.e. ones
73 whose data attribute is split across multiple extents, is severly
74 limited. When such a case is encountered, EOPNOTSUPP is returned.
Anton Altaparmakove9438252005-10-04 16:01:06 +010075 - Enable ATTR_SIZE attribute changes in ntfs_setattr(). This completes
76 the initial implementation of file truncation. Now both open(2)ing
77 a file with the O_TRUNC flag and the {,f}truncate(2) system calls
78 will resize a file appropriately. The limitations are that only
79 uncompressed and unencrypted files are supported. Also, there is
80 only very limited support for highly fragmented files (the ones whose
81 $DATA attribute is split into multiple attribute extents).
Anton Altaparmakov29b89902005-10-11 14:54:42 +010082 - In attrib.c::ntfs_attr_set() call balance_dirty_pages_ratelimited()
83 and cond_resched() in the main loop as we could be dirtying a lot of
84 pages and this ensures we play nice with the VM and the system as a
85 whole.
Anton Altaparmakov98b27032005-10-11 15:40:40 +010086 - Implement file operations ->write, ->aio_write, ->writev for regular
87 files. This replaces the old use of generic_file_write(), et al and
88 the address space operations ->prepare_write and ->commit_write.
89 This means that both sparse and non-sparse (unencrypted and
90 uncompressed) files can now be extended using the normal write(2)
91 code path. There are two limitations at present and these are that
92 we never create sparse files and that we only have limited support
93 for highly fragmented files, i.e. ones whose data attribute is split
94 across multiple extents. When such a case is encountered,
95 EOPNOTSUPP is returned.
Anton Altaparmakovd04bd1f2005-10-24 08:41:24 +010096 - $EA attributes can be both resident and non-resident.
97 - Use %z for size_t to fix compilation warnings. (Andrew Morton)
Anton Altaparmakovdda65b92005-10-24 08:57:59 +010098 - Fix compilation warnings with gcc-4.0.2 on SUSE 10.0.
Anton Altaparmakovc9c20092005-10-24 09:00:51 +010099 - Document extended attribute ($EA) NEED_EA flag. (Based on libntfs
100 patch by Yura Pakhuchiy.)
Anton Altaparmakovfd9d6362005-10-04 13:44:48 +0100101
Anton Altaparmakov7d333d62005-09-08 23:01:16 +01001022.1.24 - Lots of bug fixes and support more clean journal states.
Anton Altaparmakove7a10332005-09-08 16:12:28 +0100103
104 - Support journals ($LogFile) which have been modified by chkdsk. This
105 means users can boot into Windows after we marked the volume dirty.
106 The Windows boot will run chkdsk and then reboot. The user can then
107 immediately boot into Linux rather than having to do a full Windows
108 boot first before rebooting into Linux and we will recognize such a
Anton Altaparmakov5a8c0cc2005-09-26 10:48:54 +0100109 journal and empty it as it is clean by definition. Note, this only
110 works if chkdsk left the journal in an obviously clean state.
Anton Altaparmakove7a10332005-09-08 16:12:28 +0100111 - Support journals ($LogFile) with only one restart page as well as
112 journals with two different restart pages. We sanity check both and
113 either use the only sane one or the more recent one of the two in the
114 case that both are valid.
Anton Altaparmakov06d0e3c2005-09-08 16:28:25 +0100115 - Add fs/ntfs/malloc.h::ntfs_malloc_nofs_nofail() which is analogous to
116 ntfs_malloc_nofs() but it performs allocations with __GFP_NOFAIL and
117 hence cannot fail.
Anton Altaparmakov9529d462005-09-08 16:33:12 +0100118 - Use ntfs_malloc_nofs_nofail() in the two critical regions in
119 fs/ntfs/runlist.c::ntfs_runlists_merge(). This means we no longer
120 need to panic() if the allocation fails as it now cannot fail.
Anton Altaparmakov84d6ebe2005-09-08 16:46:55 +0100121 - Fix two nasty runlist merging bugs that had gone unnoticed so far.
122 Thanks to Stefano Picerno for the bug report.
Anton Altaparmakov8bb73522005-09-08 16:48:28 +0100123 - Remove two bogus BUG_ON()s from fs/ntfs/mft.c.
Anton Altaparmakov2b0ada22005-09-08 16:52:31 +0100124 - Fix handling of valid but empty mapping pairs array in
125 fs/ntfs/runlist.c::ntfs_mapping_pairs_decompress().
Anton Altaparmakovf94ad382005-09-08 17:04:11 +0100126 - Report unrepresentable inodes during ntfs_readdir() as KERN_WARNING
127 messages and include the inode number. Thanks to Yura Pakhuchiy for
128 pointing this out.
Anton Altaparmakov3ffc5a42005-09-08 20:23:06 +0100129 - Change ntfs_rl_truncate_nolock() to throw away the runlist if the new
130 length is zero.
Anton Altaparmakov6e483212005-09-08 20:26:34 +0100131 - Add runlist.[hc]::ntfs_rl_punch_nolock() which punches a caller
132 specified hole into a runlist.
Anton Altaparmakov8e08ceae2005-09-08 20:29:50 +0100133 - Fix a bug in fs/ntfs/index.c::ntfs_index_lookup(). When the returned
134 index entry is in the index root, we forgot to set the @ir pointer in
135 the index context. Thanks to Yura Pakhuchiy for finding this bug.
Anton Altaparmakovf25dfb52005-09-08 20:35:33 +0100136 - Remove bogus setting of PageError in ntfs_read_compressed_block().
Anton Altaparmakov0aaccea2005-09-08 20:40:32 +0100137 - Add fs/ntfs/attrib.[hc]::ntfs_resident_attr_value_resize().
Anton Altaparmakov2983d1b2005-09-08 20:56:09 +0100138 - Fix a bug in ntfs_map_runlist_nolock() where we forgot to protect
139 access to the allocated size in the ntfs inode with the size lock.
140 - Fix ntfs_attr_vcn_to_lcn_nolock() and ntfs_attr_find_vcn_nolock() to
141 return LCN_ENOENT when there is no runlist and the allocated size is
142 zero.
143 - Fix load_attribute_list() to handle the case of a NULL runlist.
Anton Altaparmakov807c4532005-09-08 21:01:17 +0100144 - Fix handling of sparse attributes in ntfs_attr_make_non_resident().
145 - Add BUG() checks to ntfs_attr_make_non_resident() and ntfs_attr_set()
146 to ensure that these functions are never called for compressed or
147 encrypted attributes.
Anton Altaparmakovbbf18132005-09-08 21:09:06 +0100148 - Fix cluster (de)allocators to work when the runlist is NULL and more
149 importantly to take a locked runlist rather than them locking it
150 which leads to lock reversal.
Anton Altaparmakov1c7d4692005-09-08 21:15:09 +0100151 - Truncate {a,c,m}time to the ntfs supported time granularity when
152 updating the times in the inode in ntfs_setattr().
Anton Altaparmakov67bb1032005-09-08 21:19:45 +0100153 - Fixup handling of sparse, compressed, and encrypted attributes in
Anton Altaparmakov311120e2005-09-08 22:04:20 +0100154 fs/ntfs/inode.c::ntfs_read_locked_{,attr_,index_}inode(),
155 fs/ntfs/aops.c::ntfs_{read,write}page().
Anton Altaparmakov8dcdeba2005-09-08 21:25:48 +0100156 - Make ntfs_write_block() not instantiate sparse blocks if they contain
157 only zeroes.
Anton Altaparmakov54b02eb2005-09-08 21:43:47 +0100158 - Optimize fs/ntfs/aops.c::ntfs_write_block() by extending the page
159 lock protection over the buffer submission for i/o which allows the
160 removal of the get_bh()/put_bh() pairs for each buffer.
Anton Altaparmakov8273d5d2005-09-08 22:00:33 +0100161 - Fix fs/ntfs/aops.c::ntfs_{read,write}_block() to handle the case
162 where a concurrent truncate has truncated the runlist under our feet.
Anton Altaparmakova01ac532005-09-08 22:08:11 +0100163 - Fix page_has_buffers()/page_buffers() handling in fs/ntfs/aops.c.
Anton Altaparmakove6046352005-09-08 22:13:02 +0100164 - In fs/ntfs/aops.c::ntfs_end_buffer_async_read(), use a bit spin lock
165 in the first buffer head instead of a driver global spin lock to
166 improve scalability.
Anton Altaparmakov7d333d62005-09-08 23:01:16 +0100167 - Minor fix to error handling and error message display in
Anton Altaparmakov5d467702005-09-12 14:33:47 +0100168 fs/ntfs/aops.c::ntfs_prepare_nonresident_write().
169 - Change the mount options {u,f,d}mask to always parse the number as
170 an octal number to conform to how chmod(1) works, too. Thanks to
171 Giuseppe Bilotta and Horst von Brand for pointing out the errors of
172 my ways.
Anton Altaparmakov5c9f6de2005-09-19 09:33:40 +0100173 - Fix various bugs in the runlist merging code. (Based on libntfs
174 changes by Richard Russon.)
Anton Altaparmakov5a8c0cc2005-09-26 10:48:54 +0100175 - Fix sparse warnings that have crept in over time.
176 - Change ntfs_cluster_free() to require a write locked runlist on entry
177 since we otherwise get into a lock reversal deadlock if a read locked
178 runlist is passed in. In the process also change it to take an ntfs
179 inode instead of a vfs inode as parameter.
180 - Fix the definition of the CHKD ntfs record magic. It had an off by
181 two error causing it to be CHKB instead of CHKD.
Anton Altaparmakov18efefa2005-10-04 13:06:00 +0100182 - Fix a stupid bug in __ntfs_bitmap_set_bits_in_run() which caused the
183 count to become negative and hence we had a wild memset() scribbling
184 all over the system's ram.
Anton Altaparmakove7a10332005-09-08 16:12:28 +0100185
Anton Altaparmakovaf859a42005-06-25 21:07:27 +01001862.1.23 - Implement extension of resident files and make writing safe as well as
187 many bug fixes, cleanups, and enhancements...
Linus Torvalds1da177e2005-04-16 15:20:36 -0700188
189 - Add printk rate limiting for ntfs_warning() and ntfs_error() when
190 compiled without debug. This avoids a possible denial of service
191 attack. Thanks to Carl-Daniel Hailfinger from SuSE for pointing this
192 out.
Randy Dunlap89075472005-03-03 11:19:53 +0000193 - Fix compilation warnings on ia64. (Randy Dunlap)
Anton Altaparmakovaf859a42005-06-25 21:07:27 +0100194 - Use i_size_{read,write}() instead of reading i_size by hand and cache
195 the value where apropriate.
Anton Altaparmakov36763672004-11-18 13:46:45 +0000196 - Add size_lock to the ntfs_inode structure. This is an rw spinlock
197 and it locks against access to the inode sizes. Note, ->size_lock
198 is also accessed from irq context so you must use the _irqsave and
Anton Altaparmakovaf859a42005-06-25 21:07:27 +0100199 _irqrestore lock and unlock functions, respectively. Protect all
200 accesses to allocated_size, initialized_size, and compressed_size.
Anton Altaparmakov218357f2004-11-18 20:34:59 +0000201 - Minor optimization to fs/ntfs/super.c::ntfs_statfs() and its helpers.
Anton Altaparmakov946929d2005-01-13 15:26:29 +0000202 - Implement extension of resident files in the regular file write code
203 paths (fs/ntfs/aops.c::ntfs_{prepare,commit}_write()). At present
204 this only works until the data attribute becomes too big for the mft
205 record after which we abort the write returning -EOPNOTSUPP from
206 ntfs_prepare_write().
Anton Altaparmakovc002f422005-02-03 12:02:56 +0000207 - Add disable_sparse mount option together with a per volume sparse
208 enable bit which is set appropriately and a per inode sparse disable
209 bit which is preset on some system file inodes as appropriate.
210 - Enforce that sparse support is disabled on NTFS volumes pre 3.0.
Anton Altaparmakov1a0df152005-02-03 12:04:36 +0000211 - Fix a bug in fs/ntfs/runlist.c::ntfs_mapping_pairs_decompress() in
212 the creation of the unmapped runlist element for the base attribute
213 extent.
Anton Altaparmakovb6ad6c52005-02-15 10:08:43 +0000214 - Split ntfs_map_runlist() into ntfs_map_runlist() and a non-locking
215 helper ntfs_map_runlist_nolock() which is used by ntfs_map_runlist().
216 This allows us to map runlist fragments with the runlist lock already
217 held without having to drop and reacquire it around the call. Adapt
218 all callers.
219 - Change ntfs_find_vcn() to ntfs_find_vcn_nolock() which takes a locked
220 runlist. This allows us to find runlist elements with the runlist
221 lock already held without having to drop and reacquire it around the
222 call. Adapt all callers.
Anton Altaparmakov37e4c132005-02-18 10:03:13 +0000223 - Change time to u64 in time.h::ntfs2utc() as it otherwise generates a
Anton Altaparmakovd8ec7852005-02-18 09:23:39 +0000224 warning in the do_div() call on sparc32. Thanks to Meelis Roos for
225 the report and analysis of the warning.
Anton Altaparmakov37e4c132005-02-18 10:03:13 +0000226 - Fix a nasty runlist merge bug when merging two holes.
Anton Altaparmakov5ae9fcf2005-03-02 17:03:24 +0000227 - Set the ntfs_inode->allocated_size to the real allocated size in the
228 mft record for resident attributes (fs/ntfs/inode.c).
229 - Small readability cleanup to use "a" instead of "ctx->attr"
230 everywhere (fs/ntfs/inode.c).
Anton Altaparmakov41382682005-03-03 13:44:15 +0000231 - Make fs/ntfs/namei.c::ntfs_get_{parent,dentry} static and move the
232 definition of ntfs_export_ops from fs/ntfs/super.c to namei.c. Also,
233 declare ntfs_export_ops in fs/ntfs/ntfs.h.
Anton Altaparmakov9451f852005-03-03 14:43:43 +0000234 - Correct sparse file handling. The compressed values need to be
235 checked and set in the ntfs inode as done for compressed files and
236 the compressed size needs to be used for vfs inode->i_blocks instead
237 of the allocated size, again, as done for compressed files.
Anton Altaparmakov271849a2005-03-07 21:36:18 +0000238 - Add AT_EA in addition to AT_DATA to whitelist for being allowed to be
239 non-resident in fs/ntfs/attrib.c::ntfs_attr_can_be_non_resident().
240 - Add fs/ntfs/attrib.c::ntfs_attr_vcn_to_lcn_nolock() used by the new
241 write code.
Anton Altaparmakovc0c1cc02005-03-07 21:43:38 +0000242 - Fix bug in fs/ntfs/attrib.c::ntfs_find_vcn_nolock() where after
243 dropping the read lock and taking the write lock we were not checking
244 whether someone else did not already do the work we wanted to do.
245 - Rename fs/ntfs/attrib.c::ntfs_find_vcn_nolock() to
246 ntfs_attr_find_vcn_nolock() and update all callers.
Anton Altaparmakov2bfb4ff2005-03-09 15:15:06 +0000247 - Add fs/ntfs/attrib.[hc]::ntfs_attr_make_non_resident().
Anton Altaparmakov43b01fd2005-03-09 15:18:43 +0000248 - Fix sign of various error return values to be negative in
249 fs/ntfs/lcnalloc.c.
Anton Altaparmakov905685f2005-03-10 11:06:19 +0000250 - Modify ->readpage and ->writepage (fs/ntfs/aops.c) so they detect and
251 handle the case where an attribute is converted from resident to
252 non-resident by a concurrent file write.
Jesper Juhl251c8422005-04-04 14:59:56 +0100253 - Remove checks for NULL before calling kfree() since kfree() does the
254 checking itself. (Jesper Juhl)
Anton Altaparmakovb0d23742005-04-04 16:20:14 +0100255 - Some utilities modify the boot sector but do not update the checksum.
256 Thus, relax the checking in fs/ntfs/super.c::is_boot_sector_ntfs() to
257 only emit a warning when the checksum is incorrect rather than
258 refusing the mount. Thanks to Bernd Casimir for pointing this
259 problem out.
Anton Altaparmakovbb3cf332005-04-06 13:34:31 +0100260 - Update attribute definition handling.
Anton Altaparmakov7fafb8b2005-04-06 16:09:21 +0100261 - Add NTFS_MAX_CLUSTER_SIZE and NTFS_MAX_PAGES_PER_CLUSTER constants.
262 - Use NTFS_MAX_CLUSTER_SIZE in super.c instead of hard coding 0x10000.
Anton Altaparmakovd53ee322005-04-06 16:11:20 +0100263 - Use MAX_BUF_PER_PAGE instead of variable sized array allocation for
264 better code generation and one less sparse warning in fs/ntfs/aops.c.
Pekka Enberg2fb21db2005-05-25 21:15:34 +0300265 - Remove spurious void pointer casts from fs/ntfs/. (Pekka Enberg)
Anton Altaparmakov442d2072005-05-27 16:42:56 +0100266 - Use C99 style structure initialization after memory allocation where
267 possible (fs/ntfs/{attrib.c,index.c,super.c}). Thanks to Al Viro and
268 Pekka Enberg.
Anton Altaparmakov3f2faef2005-06-25 15:28:56 +0100269 - Stamp the transaction log ($UsnJrnl), aka user space journal, if it
270 is active on the volume and we are mounting read-write or remounting
271 from read-only to read-write.
Anton Altaparmakov9f993fe2005-06-25 16:15:36 +0100272 - Fix a bug in address space operations error recovery code paths where
273 if the runlist was not mapped at all and a mapping error occured we
274 would leave the runlist locked on exit to the function so that the
275 next access to the same file would try to take the lock and deadlock.
Anton Altaparmakovca8fd7a2005-06-25 16:31:27 +0100276 - Detect the case when Windows has been suspended to disk on the volume
277 to be mounted and if this is the case do not allow (re)mounting
278 read-write. This is done by parsing hiberfil.sys if present.
Anton Altaparmakov3bd1f4a2005-06-25 16:51:58 +0100279 - Fix several occurences of a bug where we would perform 'var & ~const'
280 with a 64-bit variable and a int, i.e. 32-bit, constant. This causes
281 the higher order 32-bits of the 64-bit variable to be zeroed. To fix
282 this cast the 'const' to the same 64-bit type as 'var'.
Anton Altaparmakov1d58b272005-06-25 17:04:55 +0100283 - Change the runlist terminator of the newly allocated cluster(s) to
284 LCN_ENOENT in ntfs_attr_make_non_resident(). Otherwise the runlist
285 code gets confused.
Anton Altaparmakovfa3be9232005-06-25 17:15:36 +0100286 - Add an extra parameter @last_vcn to ntfs_get_size_for_mapping_pairs()
287 and ntfs_mapping_pairs_build() to allow the runlist encoding to be
288 partial which is desirable when filling holes in sparse attributes.
289 Update all callers.
Anton Altaparmakov4757d7df2005-06-25 17:24:08 +0100290 - Change ntfs_map_runlist_nolock() to only decompress the mapping pairs
291 if the requested vcn is inside it. Otherwise we get into problems
292 when we try to map an out of bounds vcn because we then try to map
293 the already mapped runlist fragment which causes
294 ntfs_mapping_pairs_decompress() to fail and return error. Update
295 ntfs_attr_find_vcn_nolock() accordingly.
Anton Altaparmakovba6d2372005-06-26 22:12:02 +0100296 - Fix a nasty deadlock that appeared in recent kernels.
297 The situation: VFS inode X on a mounted ntfs volume is dirty. For
298 same inode X, the ntfs_inode is dirty and thus corresponding on-disk
299 inode, i.e. mft record, which is in a dirty PAGE_CACHE_PAGE belonging
300 to the table of inodes, i.e. $MFT, inode 0.
301 What happens:
302 Process 1: sys_sync()/umount()/whatever... calls
303 __sync_single_inode() for $MFT -> do_writepages() -> write_page for
304 the dirty page containing the on-disk inode X, the page is now locked
305 -> ntfs_write_mst_block() which clears PageUptodate() on the page to
306 prevent anyone else getting hold of it whilst it does the write out.
307 This is necessary as the on-disk inode needs "fixups" applied before
308 the write to disk which are removed again after the write and
309 PageUptodate is then set again. It then analyses the page looking
310 for dirty on-disk inodes and when it finds one it calls
311 ntfs_may_write_mft_record() to see if it is safe to write this
312 on-disk inode. This then calls ilookup5() to check if the
313 corresponding VFS inode is in icache(). This in turn calls ifind()
314 which waits on the inode lock via wait_on_inode whilst holding the
315 global inode_lock.
316 Process 2: pdflush results in a call to __sync_single_inode for the
317 same VFS inode X on the ntfs volume. This locks the inode (I_LOCK)
318 then calls write-inode -> ntfs_write_inode -> map_mft_record() ->
319 read_cache_page() for the page (in page cache of table of inodes
320 $MFT, inode 0) containing the on-disk inode. This page has
321 PageUptodate() clear because of Process 1 (see above) so
322 read_cache_page() blocks when it tries to take the page lock for the
323 page so it can call ntfs_read_page().
324 Thus Process 1 is holding the page lock on the page containing the
325 on-disk inode X and it is waiting on the inode X to be unlocked in
326 ifind() so it can write the page out and then unlock the page.
327 And Process 2 is holding the inode lock on inode X and is waiting for
328 the page to be unlocked so it can call ntfs_readpage() or discover
329 that Process 1 set PageUptodate() again and use the page.
330 Thus we have a deadlock due to ifind() waiting on the inode lock.
331 The solution: The fix is to use the newly introduced
332 ilookup5_nowait() which does not wait on the inode's lock and hence
333 avoids the deadlock. This is safe as we do not care about the VFS
334 inode and only use the fact that it is in the VFS inode cache and the
335 fact that the vfs and ntfs inodes are one struct in memory to find
336 the ntfs inode in memory if present. Also, the ntfs inode has its
337 own locking so it does not matter if the vfs inode is locked.
Anton Altaparmakove74589a2005-08-16 16:38:28 +0100338 - Fix bug in mft record writing where we forgot to set the device in
Anton Altaparmakov481d0372005-08-16 19:42:56 +0100339 the buffers when mapping them after the VM had discarded them.
Anton Altaparmakove74589a2005-08-16 16:38:28 +0100340 Thanks to Martin MOKREJÅ  for the bug report.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700341
3422.1.22 - Many bug and race fixes and error handling improvements.
343
344 - Improve error handling in fs/ntfs/inode.c::ntfs_truncate().
345 - Change fs/ntfs/inode.c::ntfs_truncate() to return an error code
346 instead of void and provide a helper ntfs_truncate_vfs() for the
347 vfs ->truncate method.
348 - Add a new ntfs inode flag NInoTruncateFailed() and modify
349 fs/ntfs/inode.c::ntfs_truncate() to set and clear it appropriately.
350 - Fix min_size and max_size definitions in ATTR_DEF structure in
351 fs/ntfs/layout.h to be signed.
352 - Add attribute definition handling helpers to fs/ntfs/attrib.[hc]:
353 ntfs_attr_size_bounds_check(), ntfs_attr_can_be_non_resident(), and
354 ntfs_attr_can_be_resident(), which in turn use the new private helper
355 ntfs_attr_find_in_attrdef().
356 - In fs/ntfs/aops.c::mark_ntfs_record_dirty(), take the
357 mapping->private_lock around the dirtying of the buffer heads
358 analagous to the way it is done in __set_page_dirty_buffers().
359 - Ensure the mft record size does not exceed the PAGE_CACHE_SIZE at
360 mount time as this cannot work with the current implementation.
361 - Check for location of attribute name and improve error handling in
362 general in fs/ntfs/inode.c::ntfs_read_locked_inode() and friends.
363 - In fs/ntfs/aops.c::ntfs_writepage(), if the page is fully outside
364 i_size, i.e. race with truncate, invalidate the buffers on the page
365 so that they become freeable and hence the page does not leak.
366 - Remove unused function fs/ntfs/runlist.c::ntfs_rl_merge(). (Adrian
367 Bunk)
368 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_find() that resulted in
369 a NULL pointer dereference in the error code path when a corrupt
370 attribute was found. (Thanks to Domen Puncer for the bug report.)
371 - Add MODULE_VERSION() to fs/ntfs/super.c.
372 - Make several functions and variables static. (Adrian Bunk)
373 - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() so it allocates
374 buffers for the page if they are not present and then marks the
375 buffers belonging to the ntfs record dirty. This causes the buffers
376 to become busy and hence they are safe from removal until the page
377 has been written out.
378 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find() in the
379 error handling code path that resulted in a BUG() due to trying to
380 unmap an extent mft record when the mapping of it had failed and it
381 thus was not mapped. (Thanks to Ken MacFerrin for the bug report.)
382 - Drop the runlist lock after the vcn has been read in
383 fs/ntfs/lcnalloc.c::__ntfs_cluster_free().
384 - Rewrite handling of multi sector transfer errors. We now do not set
385 PageError() when such errors are detected in the async i/o handler
386 fs/ntfs/aops.c::ntfs_end_buffer_async_read(). All users of mst
387 protected attributes now check the magic of each ntfs record as they
388 use it and act appropriately. This has the effect of making errors
389 granular per ntfs record rather than per page which solves the case
390 where we cannot access any of the ntfs records in a page when a
391 single one of them had an mst error. (Thanks to Ken MacFerrin for
392 the bug report.)
393 - Fix error handling in fs/ntfs/quota.c::ntfs_mark_quotas_out_of_date()
Anton Altaparmakov3672b632006-02-24 09:55:07 +0000394 where we failed to release i_mutex on the $Quota/$Q attribute inode.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700395 - Fix bug in handling of bad inodes in fs/ntfs/namei.c::ntfs_lookup().
396 - Add mapping of unmapped buffers to all remaining code paths, i.e.
397 fs/ntfs/aops.c::ntfs_write_mst_block(), mft.c::ntfs_sync_mft_mirror(),
398 and write_mft_record_nolock(). From now on we require that the
399 complete runlist for the mft mirror is always mapped into memory.
400 - Add creation of buffers to fs/ntfs/mft.c::ntfs_sync_mft_mirror().
401 - Improve error handling in fs/ntfs/aops.c::ntfs_{read,write}_block().
402 - Cleanup fs/ntfs/aops.c::ntfs_{read,write}page() since we know that a
403 resident attribute will be smaller than a page which makes the code
404 simpler. Also make the code more tolerant to concurrent ->truncate.
405
4062.1.21 - Fix some races and bugs, rewrite mft write code, add mft allocator.
407
408 - Implement extent mft record deallocation
409 fs/ntfs/mft.c::ntfs_extent_mft_record_free().
410 - Splitt runlist related functions off from attrib.[hc] to runlist.[hc].
411 - Add vol->mft_data_pos and initialize it at mount time.
412 - Rename init_runlist() to ntfs_init_runlist(), ntfs_vcn_to_lcn() to
413 ntfs_rl_vcn_to_lcn(), decompress_mapping_pairs() to
414 ntfs_mapping_pairs_decompress(), ntfs_merge_runlists() to
415 ntfs_runlists_merge() and adapt all callers.
416 - Add fs/ntfs/runlist.[hc]::ntfs_get_nr_significant_bytes(),
417 ntfs_get_size_for_mapping_pairs(), ntfs_write_significant_bytes(),
418 and ntfs_mapping_pairs_build(), adapted from libntfs.
419 - Make fs/ntfs/lcnalloc.c::ntfs_cluster_free_from_rl_nolock() not
420 static and add a declaration for it to lcnalloc.h.
421 - Add fs/ntfs/lcnalloc.h::ntfs_cluster_free_from_rl() which is a static
422 inline wrapper for ntfs_cluster_free_from_rl_nolock() which takes the
423 cluster bitmap lock for the duration of the call.
424 - Add fs/ntfs/attrib.[hc]::ntfs_attr_record_resize().
425 - Implement the equivalent of memset() for an ntfs attribute in
426 fs/ntfs/attrib.[hc]::ntfs_attr_set() and switch
427 fs/ntfs/logfile.c::ntfs_empty_logfile() to using it.
428 - Remove unnecessary casts from LCN_* constants.
429 - Implement fs/ntfs/runlist.c::ntfs_rl_truncate_nolock().
430 - Add MFT_RECORD_OLD as a copy of MFT_RECORD in fs/ntfs/layout.h and
431 change MFT_RECORD to contain the NTFS 3.1+ specific fields.
432 - Add a helper function fs/ntfs/aops.c::mark_ntfs_record_dirty() which
433 marks all buffers belonging to an ntfs record dirty, followed by
434 marking the page the ntfs record is in dirty and also marking the vfs
435 inode containing the ntfs record dirty (I_DIRTY_PAGES).
436 - Switch fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to using the
437 new helper fs/ntfs/aops.c::mark_ntfs_record_dirty() and remove the no
438 longer needed fs/ntfs/index.[hc]::__ntfs_index_entry_mark_dirty().
439 - Move ntfs_{un,}map_page() from ntfs.h to aops.h and fix resulting
440 include errors.
441 - Move the typedefs for runlist_element and runlist from types.h to
442 runlist.h and fix resulting include errors.
443 - Remove unused {__,}format_mft_record() from fs/ntfs/mft.c.
444 - Modify fs/ntfs/mft.c::__mark_mft_record_dirty() to use the helper
445 mark_ntfs_record_dirty() which also changes the behaviour in that we
446 now set the buffers belonging to the mft record dirty as well as the
447 page itself.
448 - Update fs/ntfs/mft.c::write_mft_record_nolock() and sync_mft_mirror()
449 to cope with the fact that there now are dirty buffers in mft pages.
450 - Update fs/ntfs/inode.c::ntfs_write_inode() to also use the helper
451 mark_ntfs_record_dirty() and thus to set the buffers belonging to the
452 mft record dirty as well as the page itself.
453 - Fix compiler warnings on x86-64 in fs/ntfs/dir.c. (Randy Dunlap,
454 slightly modified by me)
455 - Add fs/ntfs/mft.c::try_map_mft_record() which fails with -EALREADY if
456 the mft record is already locked and otherwise behaves the same way
457 as fs/ntfs/mft.c::map_mft_record().
458 - Modify fs/ntfs/mft.c::write_mft_record_nolock() so that it only
459 writes the mft record if the buffers belonging to it are dirty.
460 Otherwise we assume that it was written out by other means already.
461 - Attempting to write outside initialized size is _not_ a bug so remove
462 the bug check from fs/ntfs/aops.c::ntfs_write_mst_block(). It is in
463 fact required to write outside initialized size when preparing to
464 extend the initialized size.
465 - Map the page instead of using page_address() before writing to it in
466 fs/ntfs/aops.c::ntfs_mft_writepage().
467 - Provide exclusion between opening an inode / mapping an mft record
468 and accessing the mft record in fs/ntfs/mft.c::ntfs_mft_writepage()
469 by setting the page not uptodate throughout ntfs_mft_writepage().
470 - Clear the page uptodate flag in fs/ntfs/aops.c::ntfs_write_mst_block()
471 to ensure noone can see the page whilst the mst fixups are applied.
472 - Add the helper fs/ntfs/mft.c::ntfs_may_write_mft_record() which
473 checks if an mft record may be written out safely obtaining any
474 necessary locks in the process. This is used by
475 fs/ntfs/aops.c::ntfs_write_mst_block().
476 - Modify fs/ntfs/aops.c::ntfs_write_mst_block() to also work for
477 writing mft records and improve its error handling in the process.
478 Now if any of the records in the page fail to be written out, all
479 other records will be written out instead of aborting completely.
480 - Remove ntfs_mft_aops and update all users to use ntfs_mst_aops.
481 - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to set the
482 ntfs_mst_aops for all inodes which are NInoMstProtected() and
483 ntfs_aops for all other inodes.
484 - Rename fs/ntfs/mft.c::sync_mft_mirror{,_umount}() to
485 ntfs_sync_mft_mirror{,_umount}() and change their parameters so they
486 no longer require an ntfs inode to be present. Update all callers.
487 - Cleanup the error handling in fs/ntfs/mft.c::ntfs_sync_mft_mirror().
488 - Clear the page uptodate flag in fs/ntfs/mft.c::ntfs_sync_mft_mirror()
489 to ensure noone can see the page whilst the mst fixups are applied.
490 - Remove the no longer needed fs/ntfs/mft.c::ntfs_mft_writepage() and
491 fs/ntfs/mft.c::try_map_mft_record().
492 - Fix callers of fs/ntfs/aops.c::mark_ntfs_record_dirty() to call it
493 with the ntfs inode which contains the page rather than the ntfs
494 inode the mft record of which is in the page.
495 - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by moving the
496 index inode bitmap inode release code from there to
497 fs/ntfs/inode.c::ntfs_clear_big_inode(). (Thanks to Christoph
498 Hellwig for spotting this.)
499 - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by taking the
500 inode semaphore around the code that sets ni->itype.index.bmp_ino to
501 NULL and reorganize the code to optimize it a bit. (Thanks to
502 Christoph Hellwig for spotting this.)
503 - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() to no longer take the
504 ntfs inode as a parameter as this is confusing and misleading and the
505 needed ntfs inode is available via NTFS_I(page->mapping->host).
506 Adapt all callers to this change.
507 - Modify fs/ntfs/mft.c::write_mft_record_nolock() and
508 fs/ntfs/aops.c::ntfs_write_mst_block() to only check the dirty state
509 of the first buffer in a record and to take this as the ntfs record
510 dirty state. We cannot look at the dirty state for subsequent
511 buffers because we might be racing with
512 fs/ntfs/aops.c::mark_ntfs_record_dirty().
513 - Move the static inline ntfs_init_big_inode() from fs/ntfs/inode.c to
514 inode.h and make fs/ntfs/inode.c::__ntfs_init_inode() non-static and
515 add a declaration for it to inode.h. Fix some compilation issues
516 that resulted due to #includes and header file interdependencies.
517 - Simplify setup of i_mode in fs/ntfs/inode.c::ntfs_read_locked_inode().
518 - Add helpers fs/ntfs/layout.h::MK_MREF() and MK_LE_MREF().
519 - Modify fs/ntfs/mft.c::map_extent_mft_record() to only verify the mft
520 record sequence number if it is specified (i.e. not zero).
521 - Add fs/ntfs/mft.[hc]::ntfs_mft_record_alloc() and various helper
522 functions used by it.
523 - Update Documentation/filesystems/ntfs.txt with instructions on how to
524 use the Device-Mapper driver with NTFS ftdisk/LDM raid. This removes
525 the linear raid problem with the Software RAID / MD driver when one
526 or more of the devices has an odd number of sectors.
527
5282.1.20 - Fix two stupid bugs introduced in 2.1.18 release.
529
530 - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_reinit_search_ctx()
531 where we did not clear ctx->al_entry but it was still set due to
532 changes in ntfs_attr_lookup() and ntfs_external_attr_find() in
533 particular.
534 - Fix another stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find()
535 where we forgot to unmap the extent mft record when we had finished
536 enumerating an attribute which caused a bug check to trigger when the
537 VFS calls ->clear_inode.
538
5392.1.19 - Many cleanups, improvements, and a minor bug fix.
540
541 - Update ->setattr (fs/ntfs/inode.c::ntfs_setattr()) to refuse to
542 change the uid, gid, and mode of an inode as we do not support NTFS
543 ACLs yet.
544 - Remove BKL use from ntfs_setattr() syncing up with the rest of the
545 kernel.
546 - Get rid of the ugly transparent union in fs/ntfs/dir.c::ntfs_readdir()
547 and ntfs_filldir() as per suggestion from Al Viro.
548 - Change '\0' and L'\0' to simply 0 as per advice from Linus Torvalds.
549 - Update ->truncate (fs/ntfs/inode.c::ntfs_truncate()) to check if the
550 inode size has changed and to only output an error if so.
551 - Rename fs/ntfs/attrib.h::attribute_value_length() to ntfs_attr_size().
552 - Add le{16,32,64} as well as sle{16,32,64} data types to
553 fs/ntfs/types.h.
554 - Change ntfschar to be le16 instead of u16 in fs/ntfs/types.h.
555 - Add le versions of VCN, LCN, and LSN called leVCN, leLCN, and leLSN,
556 respectively, to fs/ntfs/types.h.
557 - Update endianness conversion macros in fs/ntfs/endian.h to use the
558 new types as appropriate.
559 - Do proper type casting when using sle64_to_cpup() in fs/ntfs/dir.c
560 and index.c.
561 - Add leMFT_REF data type to fs/ntfs/layout.h.
562 - Update all NTFS header files with the new little endian data types.
563 Affected files are fs/ntfs/layout.h, logfile.h, and time.h.
564 - Do proper type casting when using ntfs_is_*_recordp() in
565 fs/ntfs/logfile.c, mft.c, and super.c.
566 - Fix all the sparse bitwise warnings. Had to change all the typedef
567 enums storing little endian values to simple enums plus a typedef for
568 the datatype to make sparse happy.
569 - Fix a bug found by the new sparse bitwise warnings where the default
570 upcase table was defined as a pointer to wchar_t rather than ntfschar
571 in fs/ntfs/ntfs.h and super.c.
572 - Change {const_,}cpu_to_le{16,32}(0) to just 0 as suggested by Al Viro.
573
5742.1.18 - Fix scheduling latencies at mount time as well as an endianness bug.
575
576 - Remove vol->nr_mft_records as it was pretty meaningless and optimize
577 the calculation of total/free inodes as used by statfs().
578 - Fix scheduling latencies in ntfs_fill_super() by dropping the BKL
579 because the code itself is using the ntfs_lock semaphore which
580 provides safe locking. (Ingo Molnar)
581 - Fix a potential bug in fs/ntfs/mft.c::map_extent_mft_record() that
582 could occur in the future for when we start closing/freeing extent
583 inodes if we don't set base_ni->ext.extent_ntfs_inos to NULL after
584 we free it.
585 - Rename {find,lookup}_attr() to ntfs_attr_{find,lookup}() as well as
586 find_external_attr() to ntfs_external_attr_find() to cleanup the
587 namespace a bit and to be more consistent with libntfs.
588 - Rename {{re,}init,get,put}_attr_search_ctx() to
589 ntfs_attr_{{re,}init,get,put}_search_ctx() as well as the type
590 attr_search_context to ntfs_attr_search_ctx.
591 - Force use of ntfs_attr_find() in ntfs_attr_lookup() when searching
592 for the attribute list attribute itself.
593 - Fix endianness bug in ntfs_external_attr_find().
594 - Change ntfs_{external_,}attr_find() to return 0 on success, -ENOENT
595 if the attribute is not found, and -EIO on real error. In the case
596 of -ENOENT, the search context is updated to describe the attribute
597 before which the attribute being searched for would need to be
598 inserted if such an action were to be desired and in the case of
599 ntfs_external_attr_find() the search context is also updated to
600 indicate the attribute list entry before which the attribute list
601 entry of the attribute being searched for would need to be inserted
602 if such an action were to be desired. Also make ntfs_find_attr()
603 static and remove its prototype from attrib.h as it is not used
604 anywhere other than attrib.c. Update ntfs_attr_lookup() and all
605 callers of ntfs_{external,}attr_{find,lookup}() for the new return
606 values.
607 - Minor cleanup of fs/ntfs/inode.c::ntfs_init_locked_inode().
608
6092.1.17 - Fix bugs in mount time error code paths and other updates.
610
611 - Implement bitmap modification code (fs/ntfs/bitmap.[hc]). This
612 includes functions to set/clear a single bit or a run of bits.
613 - Add fs/ntfs/attrib.[hc]::ntfs_find_vcn() which returns the locked
614 runlist element containing a particular vcn. It also takes care of
615 mapping any needed runlist fragments.
616 - Implement cluster (de-)allocation code (fs/ntfs/lcnalloc.[hc]).
617 - Load attribute definition table from $AttrDef at mount time.
618 - Fix bugs in mount time error code paths involving (de)allocation of
619 the default and volume upcase tables.
620 - Remove ntfs_nr_mounts as it is no longer used.
621
6222.1.16 - Implement access time updates, file sync, async io, and read/writev.
623
624 - Add support for readv/writev and aio_read/aio_write (fs/ntfs/file.c).
625 This is done by setting the appropriate file operations pointers to
626 the generic helper functions provided by mm/filemap.c.
627 - Implement fsync, fdatasync, and msync both for files (fs/ntfs/file.c)
628 and directories (fs/ntfs/dir.c).
629 - Add support for {a,m,c}time updates to inode.c::ntfs_write_inode().
630 Note, except for the root directory and any other system files opened
631 by the user, the system files will not have their access times
632 updated as they are only accessed at the inode level an hence the
633 file level functions which cause the times to be updated are never
634 invoked.
635
6362.1.15 - Invalidate quotas when (re)mounting read-write.
637
638 - Add new element itype.index.collation_rule to the ntfs inode
639 structure and set it appropriately in ntfs_read_locked_inode().
640 - Implement a new inode type "index" to allow efficient access to the
641 indices found in various system files and adapt inode handling
642 accordingly (fs/ntfs/inode.[hc]). An index inode is essentially an
643 attribute inode (NInoAttr() is true) with an attribute type of
644 AT_INDEX_ALLOCATION. As such, it is no longer allowed to call
645 ntfs_attr_iget() with an attribute type of AT_INDEX_ALLOCATION as
646 there would be no way to distinguish between normal attribute inodes
647 and index inodes. The function to obtain an index inode is
648 ntfs_index_iget() and it uses the helper function
649 ntfs_read_locked_index_inode(). Note, we do not overload
650 ntfs_attr_iget() as indices consist of multiple attributes so using
651 ntfs_attr_iget() to obtain an index inode would be confusing.
652 - Ensure that there is no overflow when doing page->index <<
653 PAGE_CACHE_SHIFT by casting page->index to s64 in fs/ntfs/aops.c.
654 - Use atomic kmap instead of kmap() in fs/ntfs/aops.c::ntfs_read_page()
655 and ntfs_read_block().
656 - Use case sensitive attribute lookups instead of case insensitive ones.
657 - Lock all page cache pages belonging to mst protected attributes while
658 accessing them to ensure we never see corrupt data while the page is
659 under writeout.
660 - Add framework for generic ntfs collation (fs/ntfs/collation.[hc]).
661 We have ntfs_is_collation_rule_supported() to check if the collation
662 rule you want to use is supported and ntfs_collation() which actually
663 collates two data items. We currently only support COLLATION_BINARY
664 and COLLATION_NTOFS_ULONG but support for other collation rules will
665 be added as the need arises.
666 - Add a new type, ntfs_index_context, to allow retrieval of an index
667 entry using the corresponding index key. To get an index context,
668 use ntfs_index_ctx_get() and to release it, use ntfs_index_ctx_put().
669 This also adds a new slab cache for the index contexts. To lookup a
670 key in an index inode, use ntfs_index_lookup(). After modifying an
671 index entry, call ntfs_index_entry_flush_dcache_page() followed by
672 ntfs_index_entry_mark_dirty() to ensure the changes are written out
673 to disk. For details see fs/ntfs/index.[hc]. Note, at present, if
674 an index entry is in the index allocation attribute rather than the
675 index root attribute it will not be written out (you will get a
676 warning message about discarded changes instead).
677 - Load the quota file ($Quota) and check if quota tracking is enabled
678 and if so, mark the quotas out of date. This causes windows to
679 rescan the volume on boot and update all quota entries.
680 - Add a set_page_dirty address space operation for ntfs_m[fs]t_aops.
681 It is simply set to __set_page_dirty_nobuffers() to make sure that
682 running set_page_dirty() on a page containing mft/ntfs records will
683 not affect the dirty state of the page buffers.
684 - Add fs/ntfs/index.c::__ntfs_index_entry_mark_dirty() which sets all
685 buffers that are inside the ntfs record in the page dirty after which
686 it sets the page dirty. This allows ->writepage to only write the
687 dirty index records rather than having to write all the records in
688 the page. Modify fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to
689 use this rather than __set_page_dirty_nobuffers().
690 - Implement fs/ntfs/aops.c::ntfs_write_mst_block() which enables the
691 writing of page cache pages belonging to mst protected attributes
692 like the index allocation attribute in directory indices and other
693 indices like $Quota/$Q, etc. This means that the quota is now marked
694 out of date on all volumes rather than only on ones where the quota
695 defaults entry is in the index root attribute of the $Quota/$Q index.
696
6972.1.14 - Fix an NFSd caused deadlock reported by several users.
698
699 - Modify fs/ntfs/ntfs_readdir() to copy the index root attribute value
700 to a buffer so that we can put the search context and unmap the mft
701 record before calling the filldir() callback. We need to do this
702 because of NFSd which calls ->lookup() from its filldir callback()
703 and this causes NTFS to deadlock as ntfs_lookup() maps the mft record
704 of the directory and since ntfs_readdir() has got it mapped already
705 ntfs_lookup() deadlocks.
706
7072.1.13 - Enable overwriting of resident files and housekeeping of system files.
708
709 - Implement writing of mft records (fs/ntfs/mft.[hc]), which includes
710 keeping the mft mirror in sync with the mft when mirrored mft records
711 are written. The functions are write_mft_record{,_nolock}(). The
712 implementation is quite rudimentary for now with lots of things not
713 implemented yet but I am not sure any of them can actually occur so
714 I will wait for people to hit each one and only then implement it.
715 - Commit open system inodes at umount time. This should make it
716 virtually impossible for sync_mft_mirror_umount() to ever be needed.
717 - Implement ->write_inode (fs/ntfs/inode.c::ntfs_write_inode()) for the
718 ntfs super operations. This gives us inode writing via the VFS inode
719 dirty code paths. Note: Access time updates are not implemented yet.
720 - Implement fs/ntfs/mft.[hc]::{,__}mark_mft_record_dirty() and make
721 fs/ntfs/aops.c::ntfs_writepage() and ntfs_commit_write() use it, thus
722 finally enabling resident file overwrite! (-8 This also includes a
723 placeholder for ->writepage (ntfs_mft_writepage()), which for now
724 just redirties the page and returns. Also, at umount time, we for
725 now throw away all mft data page cache pages after the last call to
726 ntfs_commit_inode() in the hope that all inodes will have been
727 written out by then and hence no dirty (meta)data will be lost. We
728 also check for this case and emit an error message telling the user
729 to run chkdsk.
730 - Use set_page_writeback() and end_page_writeback() in the resident
731 attribute code path of fs/ntfs/aops.c::ntfs_writepage() otherwise
732 the radix-tree tag PAGECACHE_TAG_DIRTY remains set even though the
733 page is clean.
734 - Implement ntfs_mft_writepage() so it now checks if any of the mft
735 records in the page are dirty and if so redirties the page and
736 returns. Otherwise it just returns (after doing set_page_writeback(),
737 unlock_page(), end_page_writeback() or the radix-tree tag
738 PAGECACHE_TAG_DIRTY remains set even though the page is clean), thus
739 alowing the VM to do with the page as it pleases. Also, at umount
740 time, now only throw away dirty mft (meta)data pages if dirty inodes
741 are present and ask the user to email us if they see this happening.
742 - Add functions ntfs_{clear,set}_volume_flags(), to modify the volume
743 information flags (fs/ntfs/super.c).
744 - Mark the volume dirty when (re)mounting read-write and mark it clean
745 when unmounting or remounting read-only. If any volume errors are
746 found, the volume is left marked dirty to force chkdsk to run.
747 - Add code to set the NT4 compatibility flag when (re)mounting
748 read-write for newer NTFS versions but leave it commented out for now
749 since we do not make any modifications that are NTFS 1.2 specific yet
750 and since setting this flag breaks Captive-NTFS which is not nice.
751 This code must be enabled once we start writing NTFS 1.2 specific
752 changes otherwise Windows NTFS driver might crash / cause corruption.
753
7542.1.12 - Fix the second fix to the decompression engine and some cleanups.
755
756 - Add a new address space operations struct, ntfs_mst_aops, for mst
757 protected attributes. This is because the default ntfs_aops do not
758 make sense with mst protected data and were they to write anything to
759 such an attribute they would cause data corruption so we provide
760 ntfs_mst_aops which does not have any write related operations set.
761 - Cleanup dirty ntfs inode handling (fs/ntfs/inode.[hc]) which also
762 includes an adapted ntfs_commit_inode() and an implementation of
763 ntfs_write_inode() which for now just cleans dirty inodes without
764 writing them (it does emit a warning that this is happening).
765 - Undo the second decompression engine fix (see 2.1.9 release ChangeLog
766 entry) as it was only fixing a theoretical bug but at the same time
767 it badly broke the handling of sparse and uncompressed compression
768 blocks.
769
7702.1.11 - Driver internal cleanups.
771
772 - Only build logfile.o if building the driver with read-write support.
773 - Really final white space cleanups.
774 - Use generic_ffs() instead of ffs() in logfile.c which allows the
775 log_page_size variable to be optimized by gcc into a constant.
776 - Rename uchar_t to ntfschar everywhere as uchar_t is unsigned 1-byte
777 char as defined by POSIX and as found on some systems.
778
7792.1.10 - Force read-only (re)mounting of volumes with unsupported volume flags.
780
781 - Finish off the white space cleanups (remove trailing spaces, etc).
782 - Clean up ntfs_fill_super() and ntfs_read_inode_mount() by removing
783 the kludges around the first iget(). Instead of (re)setting ->s_op
784 we have the $MFT inode set up by explicit new_inode() / set ->i_ino /
785 insert_inode_hash() / call ntfs_read_inode_mount() directly. This
786 kills the need for second super_operations and allows to return error
787 from ntfs_read_inode_mount() without resorting to ugly "poisoning"
788 tricks. (Al Viro)
789 - Force read-only (re)mounting if any of the following bits are set in
790 the volume information flags:
791 VOLUME_IS_DIRTY, VOLUME_RESIZE_LOG_FILE,
792 VOLUME_UPGRADE_ON_MOUNT, VOLUME_DELETE_USN_UNDERWAY,
793 VOLUME_REPAIR_OBJECT_ID, VOLUME_MODIFIED_BY_CHKDSK
794 To make this easier we define VOLUME_MUST_MOUNT_RO_MASK with all the
795 above bits set so the test is made easy.
796
7972.1.9 - Fix two bugs in decompression engine.
798
799 - Fix a bug where we would not always detect that we have reached the
800 end of a compression block because we were ending at minus one byte
801 which is effectively the same as being at the end. The fix is to
802 check whether the uncompressed buffer has been fully filled and if so
803 we assume we have reached the end of the compression block. A big
804 thank you to Marcin Gibuła for the bug report, the assistance in
805 tracking down the bug and testing the fix.
806 - Fix a possible bug where when a compressed read is truncated to the
807 end of the file, the offset inside the last page was not truncated.
808
8092.1.8 - Handle $MFT mirror and $LogFile, improve time handling, and cleanups.
810
811 - Use get_bh() instead of manual atomic_inc() in fs/ntfs/compress.c.
812 - Modify fs/ntfs/time.c::ntfs2utc(), get_current_ntfs_time(), and
813 utc2ntfs() to work with struct timespec instead of time_t on the
814 Linux UTC time side thus preserving the full precision of the NTFS
815 time and only loosing up to 99 nano-seconds in the Linux UTC time.
816 - Move fs/ntfs/time.c to fs/ntfs/time.h and make the time functions
817 static inline.
818 - Remove unused ntfs_dirty_inode().
819 - Cleanup super operations declaration in fs/ntfs/super.c.
820 - Wrap flush_dcache_mft_record_page() in #ifdef NTFS_RW.
821 - Add NInoTestSetFoo() and NInoTestClearFoo() macro magic to
822 fs/ntfs/inode.h and use it to declare NInoTest{Set,Clear}Dirty.
823 - Move typedefs for ntfs_attr and test_t from fs/ntfs/inode.c to
824 fs/ntfs/inode.h so they can be used elsewhere.
825 - Determine the mft mirror size as the number of mirrored mft records
826 and store it in ntfs_volume->mftmirr_size (fs/ntfs/super.c).
827 - Load the mft mirror at mount time and compare the mft records stored
828 in it to the ones in the mft. Force a read-only mount if the two do
829 not match (fs/ntfs/super.c).
830 - Fix type casting related warnings on 64-bit architectures. Thanks
831 to Meelis Roos for reporting them.
832 - Move %L to %ll as %L is floating point and %ll is integer which is
833 what we want.
834 - Read the journal ($LogFile) and determine if the volume has been
835 shutdown cleanly and force a read-only mount if not (fs/ntfs/super.c
836 and fs/ntfs/logfile.c). This is a little bit of a crude check in
837 that we only look at the restart areas and not at the actual log
838 records so that there will be a very small number of cases where we
839 think that a volume is dirty when in fact it is clean. This should
840 only affect volumes that have not been shutdown cleanly and did not
841 have any pending, non-check-pointed i/o.
842 - If the $LogFile indicates a clean shutdown and a read-write (re)mount
843 is requested, empty $LogFile by overwriting it with 0xff bytes to
844 ensure that Windows cannot cause data corruption by replaying a stale
845 journal after Linux has written to the volume.
846
8472.1.7 - Enable NFS exporting of mounted NTFS volumes.
848
849 - Set i_generation in the VFS inode from the seq_no of the NTFS inode.
850 - Make ntfs_lookup() NFS export safe, i.e. use d_splice_alias(), etc.
851 - Implement ->get_dentry() in fs/ntfs/namei.c::ntfs_get_dentry() as the
852 default doesn't allow inode number 0 which is a valid inode on NTFS
853 and even if it did allow that it uses iget() instead of ntfs_iget()
854 which makes it useless for us.
855 - Implement ->get_parent() in fs/ntfs/namei.c::ntfs_get_parent() as the
856 default just returns -EACCES which is not very useful.
857 - Define export operations (->s_export_op) for NTFS (ntfs_export_ops)
858 and set them up in the super block at mount time (super.c) this
859 allows mounted NTFS volumes to be exported via NFS.
860 - Add missing return -EOPNOTSUPP; in
861 fs/ntfs/aops.c::ntfs_commit_nonresident_write().
862 - Enforce no atime and no dir atime updates at mount/remount time as
863 they are not implemented yet anyway.
864 - Move a few assignments in fs/ntfs/attrib.c::load_attribute_list() to
865 after a NULL check. Thanks to Dave Jones for pointing this out.
866
8672.1.6 - Fix minor bug in handling of compressed directories.
868
869 - Fix bug in handling of compressed directories. A compressed
870 directory is not really compressed so when we set the ->i_blocks
871 field of a compressed directory inode we were setting it from the
872 non-existing field ni->itype.compressed.size which gave random
873 results... For directories we now always use ni->allocated_size.
874
8752.1.5 - Fix minor bug in attribute list attribute handling.
876
877 - Fix bug in attribute list handling. Actually it is not as much a bug
878 as too much protection in that we were not allowing attribute lists
879 which waste space on disk while Windows XP clearly allows it and in
880 fact creates such attribute lists so our driver was failing.
881 - Update NTFS documentation ready for 2.6 kernel release.
882
8832.1.4 - Reduce compiler requirements.
884
885 - Remove all uses of unnamed structs and unions in the driver to make
886 old and newer gcc versions happy. Makes it a bit uglier IMO but at
887 least people will stop hassling me about it.
888
8892.1.3 - Important bug fixes in corner cases.
890
891 - super.c::parse_ntfs_boot_sector(): Correct the check for 64-bit
892 clusters. (Philipp Thomas)
893 - attrib.c::load_attribute_list(): Fix bug when initialized_size is a
894 multiple of the block_size but not the cluster size. (Szabolcs
Anton Altaparmakov3672b632006-02-24 09:55:07 +0000895 Szakacsits)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700896
8972.1.2 - Important bug fixes aleviating the hangs in statfs.
898
899 - Fix buggy free cluster and free inode determination logic.
900
9012.1.1 - Minor updates.
902
903 - Add handling for initialized_size != data_size in compressed files.
904 - Reduce function local stack usage from 0x3d4 bytes to just noise in
Anton Altaparmakov3672b632006-02-24 09:55:07 +0000905 fs/ntfs/upcase.c. (Randy Dunlap)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700906 - Remove compiler warnings for newer gcc.
907 - Pages are no longer kmapped by mm/filemap.c::generic_file_write()
908 around calls to ->{prepare,commit}_write. Adapt NTFS appropriately
909 in fs/ntfs/aops.c::ntfs_prepare_nonresident_write() by using
910 kmap_atomic(KM_USER0).
911
9122.1.0 - First steps towards write support: implement file overwrite.
913
914 - Add configuration option for developmental write support with an
915 appropriately scary configuration help text.
916 - Initial implementation of fs/ntfs/aops.c::ntfs_writepage() and its
917 helper fs/ntfs/aops.c::ntfs_write_block(). This enables mmap(2) based
918 overwriting of existing files on ntfs. Note: Resident files are
919 only written into memory, and not written out to disk at present, so
920 avoid writing to files smaller than about 1kiB.
921 - Initial implementation of fs/ntfs/aops.c::ntfs_prepare_write(), its
922 helper fs/ntfs/aops.c::ntfs_prepare_nonresident_write() and their
923 counterparts, fs/ntfs/aops.c::ntfs_commit_write(), and
924 fs/ntfs/aops.c::ntfs_commit_nonresident_write(), respectively. Also,
925 add generic_file_write() to the ntfs file operations (fs/ntfs/file.c).
926 This enables write(2) based overwriting of existing files on ntfs.
927 Note: As with mmap(2) based overwriting, resident files are only
928 written into memory, and not written out to disk at present, so avoid
929 writing to files smaller than about 1kiB.
930 - Implement ->truncate (fs/ntfs/inode.c::ntfs_truncate()) and
931 ->setattr() (fs/ntfs/inode.c::ntfs_setattr()) inode operations for
932 files with the purpose of intercepting and aborting all i_size
933 changes which we do not support yet. ntfs_truncate() actually only
934 emits a warning message but AFAICS our interception of i_size changes
935 elsewhere means ntfs_truncate() never gets called for i_size changes.
936 It is only called from generic_file_write() when we fail in
937 ntfs_prepare_{,nonresident_}write() in order to discard any
938 instantiated buffers beyond i_size. Thus i_size is not actually
939 changed so our warning message is enough. Unfortunately it is not
940 possible to easily determine if i_size is being changed or not hence
941 we just emit an appropriately worded error message.
942
9432.0.25 - Small bug fixes and cleanups.
944
945 - Unlock the page in an out of memory error code path in
946 fs/ntfs/aops.c::ntfs_read_block().
947 - If fs/ntfs/aops.c::ntfs_read_page() is called on an uptodate page,
948 just unlock the page and return. (This can happen due to ->writepage
949 clearing PageUptodate() during write out of MstProtected()
950 attributes.
951 - Remove leaked write code again.
952
9532.0.24 - Cleanups.
954
955 - Treat BUG_ON() as ASSERT() not VERIFY(), i.e. do not use side effects
956 inside BUG_ON(). (Adam J. Richter)
957 - Split logical OR expressions inside BUG_ON() into individual BUG_ON()
958 calls for improved debugging. (Adam J. Richter)
959 - Add errors flag to the ntfs volume state, accessed via
960 NVol{,Set,Clear}Errors(vol).
961 - Do not allow read-write remounts of read-only volumes with errors.
962 - Clarify comment for ntfs file operation sendfile which was added by
963 Christoph Hellwig a while ago (just using generic_file_sendfile())
964 to say that ntfs ->sendfile is only used for the case where the
965 source data is on the ntfs partition and the destination is
966 somewhere else, i.e. nothing we need to concern ourselves with.
967 - Add generic_file_write() as our ntfs file write operation.
968
9692.0.23 - Major bug fixes (races, deadlocks, non-i386 architectures).
970
971 - Massive internal locking changes to mft record locking. Fixes lock
972 recursion and replaces the mrec_lock read/write semaphore with a
973 mutex. Also removes the now superfluous mft_count. This fixes several
974 race conditions and deadlocks, especially in the future write code.
975 - Fix ntfs over loopback for compressed files by adding an
976 optimization barrier. (gcc was screwing up otherwise ?)
977 - Miscellaneous cleanups all over the code and a fix or two in error
978 handling code paths.
979 Thanks go to Christoph Hellwig for pointing out the following two:
980 - Remove now unused function fs/ntfs/malloc.h::vmalloc_nofs().
981 - Fix ntfs_free() for ia64 and parisc by checking for VMALLOC_END, too.
982
9832.0.22 - Cleanups, mainly to ntfs_readdir(), and use C99 initializers.
984
985 - Change fs/ntfs/dir.c::ntfs_reddir() to only read/write ->f_pos once
986 at entry/exit respectively.
987 - Use C99 initializers for structures.
988 - Remove unused variable blocks from fs/ntfs/aops.c::ntfs_read_block().
989
9902.0.21 - Check for, and refuse to work with too large files/directories/volumes.
991
992 - Limit volume size at mount time to 2TiB on architectures where
993 unsigned long is 32-bits (fs/ntfs/super.c::parse_ntfs_boot_sector()).
994 This is the most we can do without overflowing the 32-bit limit of
995 the block device size imposed on us by sb_bread() and sb_getblk()
996 for the time being.
997 - Limit file/directory size at open() time to 16TiB on architectures
998 where unsigned long is 32-bits (fs/ntfs/file.c::ntfs_file_open() and
999 fs/ntfs/dir.c::ntfs_dir_open()). This is the most we can do without
1000 overflowing the page cache page index.
1001
10022.0.20 - Support non-resident directory index bitmaps, fix page leak in readdir.
1003
1004 - Move the directory index bitmap to use an attribute inode instead of
1005 having special fields for it inside the ntfs inode structure. This
1006 means that the index bitmaps now use the page cache for i/o, too,
1007 and also as a side effect we get support for non-resident index
1008 bitmaps for free.
1009 - Simplify/cleanup error handling in fs/ntfs/dir.c::ntfs_readdir() and
1010 fix a page leak that manifested itself in some cases.
1011 - Add fs/ntfs/inode.c::ntfs_put_inode(), which we need to release the
1012 index bitmap inode on the final iput().
1013
10142.0.19 - Fix race condition, improvements, and optimizations in i/o interface.
1015
1016 - Apply block optimization added to fs/ntfs/aops.c::ntfs_read_block()
1017 to fs/ntfs/compress.c::ntfs_file_read_compressed_block() as well.
1018 - Drop the "file" from ntfs_file_read_compressed_block().
1019 - Rename fs/ntfs/aops.c::ntfs_enb_buffer_read_async() to
1020 ntfs_end_buffer_async_read() (more like the fs/buffer.c counterpart).
1021 - Update ntfs_end_buffer_async_read() with the improved logic from
1022 its updated counterpart fs/buffer.c::end_buffer_async_read(). Apply
1023 further logic improvements to better determine when we set PageError.
1024 - Update submission of buffers in fs/ntfs/aops.c::ntfs_read_block() to
1025 check for the buffers being uptodate first in line with the updated
1026 fs/buffer.c::block_read_full_page(). This plugs a small race
1027 condition.
1028
10292.0.18 - Fix race condition in reading of compressed files.
1030
1031 - There was a narrow window between checking a buffer head for being
1032 uptodate and locking it in ntfs_file_read_compressed_block(). We now
1033 lock the buffer and then check whether it is uptodate or not.
1034
10352.0.17 - Cleanups and optimizations - shrinking the ToDo list.
1036
1037 - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to return an error
1038 code and update callers, i.e. ntfs_iget(), to pass that error code
1039 up instead of just using -EIO.
1040 - Modifications to super.c to ensure that both mount and remount
1041 cannot set any write related options when the driver is compiled
1042 read-only.
1043 - Optimize block resolution in fs/ntfs/aops.c::ntfs_read_block() to
1044 cache the current runlist element. This should improve performance
1045 when reading very large and/or very fragmented data.
1046
10472.0.16 - Convert access to $MFT/$BITMAP to attribute inode API.
1048
1049 - Fix a stupid bug introduced in 2.0.15 where we were unmapping the
1050 wrong inode in fs/ntfs/inode.c::ntfs_attr_iget().
1051 - Fix debugging check in fs/ntfs/aops.c::ntfs_read_block().
1052 - Convert $MFT/$BITMAP access to attribute inode API and remove all
1053 remnants of the ugly mftbmp address space and operations hack. This
1054 means we finally have only one readpage function as well as only one
1055 async io completion handler. Yey! The mft bitmap is now just an
1056 attribute inode and is accessed from vol->mftbmp_ino just as if it
1057 were a normal file. Fake inodes rule. (-:
1058
10592.0.15 - Fake inodes based attribute i/o via the pagecache, fixes and cleanups.
1060
1061 - Fix silly bug in fs/ntfs/super.c::parse_options() which was causing
1062 remounts to fail when the partition had an entry in /etc/fstab and
1063 the entry specified the nls= option.
1064 - Apply same macro magic used in fs/ntfs/inode.h to fs/ntfs/volume.h to
1065 expand all the helper functions NVolFoo(), NVolSetFoo(), and
1066 NVolClearFoo().
1067 - Move copyright statement from driver initialisation message to
1068 module description (fs/super.c). This makes the initialisation
1069 message fit on one line and fits in better with rest of kernel.
1070 - Update fs/ntfs/attrib.c::map_run_list() to work on both real and
1071 attribute inodes, and both for files and directories.
1072 - Implement fake attribute inodes allowing all attribute i/o to go via
1073 the page cache and to use all the normal vfs/mm functionality:
1074 - Add ntfs_attr_iget() and its helper ntfs_read_locked_attr_inode()
1075 to fs/ntfs/inode.c.
1076 - Add needed cleanup code to ntfs_clear_big_inode().
1077 - Merge address space operations for files and directories (aops.c),
1078 now just have ntfs_aops:
1079 - Rename:
1080 end_buffer_read_attr_async() -> ntfs_end_buffer_read_async(),
1081 ntfs_attr_read_block() -> ntfs_read_block(),
1082 ntfs_file_read_page() -> ntfs_readpage().
1083 - Rewrite fs/ntfs/aops.c::ntfs_readpage() to work on both real and
1084 attribute inodes, and both for files and directories.
1085 - Remove obsolete fs/ntfs/aops.c::ntfs_mst_readpage().
1086
10872.0.14 - Run list merging code cleanup, minor locking changes, typo fixes.
1088
1089 - Change fs/ntfs/super.c::ntfs_statfs() to not rely on BKL by moving
1090 the locking out of super.c::get_nr_free_mft_records() and taking and
1091 dropping the mftbmp_lock rw_semaphore in ntfs_statfs() itself.
1092 - Bring attribute runlist merging code (fs/ntfs/attrib.c) in sync with
1093 current userspace ntfs library code. This means that if a merge
1094 fails the original runlists are always left unmodified instead of
1095 being silently corrupted.
1096 - Misc typo fixes.
1097
10982.0.13 - Use iget5_locked() in preparation for fake inodes and small cleanups.
1099
1100 - Remove nr_mft_bits and the now superfluous union with nr_mft_records
1101 from ntfs_volume structure.
1102 - Remove nr_lcn_bits and the now superfluous union with nr_clusters
1103 from ntfs_volume structure.
1104 - Use iget5_locked() and friends instead of conventional iget(). Wrap
1105 the call in fs/ntfs/inode.c::ntfs_iget() and update callers of iget()
1106 to use ntfs_iget(). Leave only one iget() call at mount time so we
1107 don't need an ntfs_iget_mount().
1108 - Change fs/ntfs/inode.c::ntfs_new_extent_inode() to take mft_no as an
1109 additional argument.
1110
11112.0.12 - Initial cleanup of address space operations following 2.0.11 changes.
1112
1113 - Merge fs/ntfs/aops.c::end_buffer_read_mst_async() and
1114 fs/ntfs/aops.c::end_buffer_read_file_async() into one function
1115 fs/ntfs/aops.c::end_buffer_read_attr_async() using NInoMstProtected()
1116 to determine whether to apply mst fixups or not.
1117 - Above change allows merging fs/ntfs/aops.c::ntfs_file_read_block()
1118 and fs/ntfs/aops.c::ntfs_mst_readpage() into one function
1119 fs/ntfs/aops.c::ntfs_attr_read_block(). Also, create a tiny wrapper
1120 fs/ntfs/aops.c::ntfs_mst_readpage() to transform the parameters from
1121 the VFS readpage function prototype to the ntfs_attr_read_block()
1122 function prototype.
1123
11242.0.11 - Initial preparations for fake inode based attribute i/o.
1125
1126 - Move definition of ntfs_inode_state_bits to fs/ntfs/inode.h and
1127 do some macro magic (adapted from include/linux/buffer_head.h) to
1128 expand all the helper functions NInoFoo(), NInoSetFoo(), and
1129 NInoClearFoo().
1130 - Add new flag to ntfs_inode_state_bits: NI_Sparse.
1131 - Add new fields to ntfs_inode structure to allow use of fake inodes
1132 for attribute i/o: type, name, name_len. Also add new state bits:
1133 NI_Attr, which, if set, indicates the inode is a fake inode, and
1134 NI_MstProtected, which, if set, indicates the attribute uses multi
1135 sector transfer protection, i.e. fixups need to be applied after
1136 reads and before/after writes.
1137 - Rename fs/ntfs/inode.c::ntfs_{new,clear,destroy}_inode() to
1138 ntfs_{new,clear,destroy}_extent_inode() and update callers.
1139 - Use ntfs_clear_extent_inode() in fs/ntfs/inode.c::__ntfs_clear_inode()
1140 instead of ntfs_destroy_extent_inode().
1141 - Cleanup memory deallocations in {__,}ntfs_clear_{,big_}inode().
1142 - Make all operations on ntfs inode state bits use the NIno* functions.
1143 - Set up the new ntfs inode fields and state bits in
1144 fs/ntfs/inode.c::ntfs_read_inode() and add appropriate cleanup of
1145 allocated memory to __ntfs_clear_inode().
1146 - Cleanup ntfs_inode structure a bit for better ordering of elements
1147 w.r.t. their size to allow better packing of the structure in memory.
1148
11492.0.10 - There can only be 2^32 - 1 inodes on an NTFS volume.
1150
1151 - Add check at mount time to verify that the number of inodes on the
1152 volume does not exceed 2^32 - 1, which is the maximum allowed for
1153 NTFS according to Microsoft.
1154 - Change mft_no member of ntfs_inode structure to be unsigned long.
1155 Update all users. This makes ntfs_inode->mft_no just a copy of struct
1156 inode->i_ino. But we can't just always use struct inode->i_ino and
1157 remove mft_no because extent inodes do not have an attached struct
1158 inode.
1159
11602.0.9 - Decompression engine now uses a single buffer and other cleanups.
1161
1162 - Change decompression engine to use a single buffer protected by a
1163 spin lock instead of per-CPU buffers. (Rusty Russell)
1164 - Do not update cb_pos when handling a partial final page during
1165 decompression of a sparse compression block, as the value is later
1166 reset without being read/used. (Rusty Russell)
1167 - Switch to using the new KM_BIO_SRC_IRQ for atomic kmap()s. (Andrew
1168 Morton)
1169 - Change buffer size in ntfs_readdir()/ntfs_filldir() to use
1170 NLS_MAX_CHARSET_SIZE which makes the buffers almost 1kiB each but
1171 it also makes everything safer so it is a good thing.
1172 - Miscellaneous minor cleanups to comments.
1173
11742.0.8 - Major updates for handling of case sensitivity and dcache aliasing.
1175
1176 Big thanks go to Al Viro and other inhabitants of #kernel for investing
1177 their time to discuss the case sensitivity and dcache aliasing issues.
1178
1179 - Remove unused source file fs/ntfs/attraops.c.
1180 - Remove show_inodes mount option(s), thus dropping support for
1181 displaying of short file names.
1182 - Remove deprecated mount option posix.
1183 - Restore show_sys_files mount option.
1184 - Add new mount option case_sensitive, to determine if the driver
1185 treats file names as case sensitive or not. If case sensitive, create
1186 file names in the POSIX namespace. Otherwise create file names in the
1187 LONG/WIN32 namespace. Note, files remain accessible via their short
1188 file name, if it exists.
1189 - Remove really dumb logic bug in boot sector recovery code.
1190 - Fix dcache aliasing issues wrt short/long file names via changes
1191 to fs/ntfs/dir.c::ntfs_lookup_inode_by_name() and
1192 fs/ntfs/namei.c::ntfs_lookup():
1193 - Add additional argument to ntfs_lookup_inode_by_name() in which we
1194 return information about the matching file name if the case is not
1195 matching or the match is a short file name. See comments above the
1196 function definition for details.
1197 - Change ntfs_lookup() to only create dcache entries for the correctly
1198 cased file name and only for the WIN32 namespace counterpart of DOS
1199 namespace file names. This ensures we have only one dentry per
1200 directory and also removes all dcache aliasing issues between short
1201 and long file names once we add write support. See comments above
1202 function for details.
1203 - Fix potential 1 byte overflow in fs/ntfs/unistr.c::ntfs_ucstonls().
1204
12052.0.7 - Minor cleanups and updates for changes in core kernel code.
1206
1207 - Remove much of the NULL struct element initializers.
1208 - Various updates to make compatible with recent kernels.
1209 - Remove defines of MAX_BUF_PER_PAGE and include linux/buffer_head.h
1210 in fs/ntfs/ntfs.h instead.
1211 - Remove no longer needed KERNEL_VERSION checks. We are now in the
1212 kernel proper so they are no longer needed.
1213
12142.0.6 - Major bugfix to make compatible with other kernel changes.
1215
1216 - Initialize the mftbmp address space properly now that there are more
1217 fields in the struct address_space. This was leading to hangs and
1218 oopses on umount since 2.5.12 because of changes to other parts of
1219 the kernel. We probably want a kernel generic init_address_space()
1220 function...
1221 - Drop BKL from ntfs_readdir() after consultation with Al Viro. The
Anton Altaparmakov3672b632006-02-24 09:55:07 +00001222 only caller of ->readdir() is vfs_readdir() which holds i_mutex
1223 during the call, and i_mutex is sufficient protection against changes
1224 in the directory inode (including ->i_size).
Linus Torvalds1da177e2005-04-16 15:20:36 -07001225 - Use generic_file_llseek() for directories (as opposed to
Anton Altaparmakov3672b632006-02-24 09:55:07 +00001226 default_llseek()) as this downs i_mutex instead of the BKL which is
Linus Torvalds1da177e2005-04-16 15:20:36 -07001227 what we now need for exclusion against ->f_pos changes considering we
1228 no longer take the BKL in ntfs_readdir().
1229
12302.0.5 - Major bugfix. Buffer overflow in extent inode handling.
1231
1232 - No need to set old blocksize in super.c::ntfs_fill_super() as the
1233 VFS does so via invocation of deactivate_super() calling
1234 fs->fill_super() calling block_kill_super() which does it.
1235 - BKL moved from VFS into dir.c::ntfs_readdir(). (Linus Torvalds)
1236 -> Do we really need it? I don't think so as we have exclusion on
1237 the directory ntfs_inode rw_semaphore mrec_lock. We mmight have to
1238 move the ->f_pos accesses under the mrec_lock though. Check this...
1239 - Fix really, really, really stupid buffer overflow in extent inode
1240 handling in mft.c::map_extent_mft_record().
1241
12422.0.4 - Cleanups and updates for kernel 2.5.11.
1243
1244 - Add documentation on how to use the MD driver to be able to use NTFS
1245 stripe and volume sets in Linux and generally cleanup documentation
1246 a bit.
1247 Remove all uses of kdev_t in favour of struct block_device *:
1248 - Change compress.c::ntfs_file_read_compressed_block() to use
1249 sb_getblk() instead of getblk().
1250 - Change super.c::ntfs_fill_super() to use bdev_hardsect_size() instead
1251 of get_hardsect_size().
1252 - No need to get old blocksize in super.c::ntfs_fill_super() as
1253 fs/super.c::get_sb_bdev() already does this.
1254 - Set bh->b_bdev instead of bh->b_dev throughout aops.c.
1255
12562.0.3 - Small bug fixes, cleanups, and performance improvements.
1257
1258 - Remove some dead code from mft.c.
1259 - Optimize readpage and read_block functions throughout aops.c so that
1260 only initialized blocks are read. Non-initialized ones have their
1261 buffer head mapped, zeroed, and set up to date, without scheduling
1262 any i/o. Thanks to Al Viro for advice on how to avoid the device i/o.
1263 Thanks go to Andrew Morton for spotting the below:
1264 - Fix buglet in allocate_compression_buffers() error code path.
1265 - Call flush_dcache_page() after modifying page cache page contents in
1266 ntfs_file_readpage().
1267 - Check for existence of page buffers throughout aops.c before calling
1268 create_empty_buffers(). This happens when an I/O error occurs and the
1269 read is retried. (It also happens once writing is implemented so that
1270 needed doing anyway but I had left it for later...)
1271 - Don't BUG_ON() uptodate and/or mapped buffers throughout aops.c in
1272 readpage and read_block functions. Reasoning same as above (i.e. I/O
1273 error retries and future write code paths.)
1274
12752.0.2 - Minor updates and cleanups.
1276
1277 - Cleanup: rename mst.c::__post_read_mst_fixup to post_write_mst_fixup
1278 and cleanup the code a bit, removing the unused size parameter.
1279 - Change default fmask to 0177 and update documentation.
1280 - Change attrib.c::get_attr_search_ctx() to return the search context
1281 directly instead of taking the address of a pointer. A return value
1282 of NULL means the allocation failed. Updated all callers
1283 appropriately.
1284 - Update to 2.5.9 kernel (preserving backwards compatibility) by
1285 replacing all occurences of page->buffers with page_buffers(page).
1286 - Fix minor bugs in runlist merging, also minor cleanup.
1287 - Updates to bootsector layout and mft mirror contents descriptions.
1288 - Small bug fix in error detection in unistr.c and some cleanups.
1289 - Grow name buffer allocations in unistr.c in aligned mutlipled of 64
1290 bytes.
1291
12922.0.1 - Minor updates.
1293
1294 - Make default umask correspond to documentation.
1295 - Improve documentation.
1296 - Set default mode to include execute bit. The {u,f,d}mask can be used
1297 to take it away if desired. This allows binaries to be executed from
1298 a mounted ntfs partition.
1299
13002.0.0 - New version number. Remove TNG from the name. Now in the kernel.
1301
1302 - Add kill_super, just keeping up with the vfs changes in the kernel.
1303 - Repeat some changes from tng-0.0.8 that somehow got lost on the way
1304 from the CVS import into BitKeeper.
1305 - Begin to implement proper handling of allocated_size vs
1306 initialized_size vs data_size (i.e. i_size). Done are
1307 mft.c::ntfs_mft_readpage(), aops.c::end_buffer_read_index_async(),
1308 and attrib.c::load_attribute_list().
1309 - Lock the runlist in attrib.c::load_attribute_list() while using it.
1310 - Fix memory leak in ntfs_file_read_compressed_block() and generally
1311 clean up compress.c a little, removing some uncommented/unused debug
1312 code.
1313 - Tidy up dir.c a little bit.
1314 - Don't bother getting the runlist in inode.c::ntfs_read_inode().
1315 - Merge mft.c::ntfs_mft_readpage() and aops.c::ntfs_index_readpage()
1316 creating aops.c::ntfs_mst_readpage(), improving the handling of
1317 holes and overflow in the process and implementing the correct
1318 equivalent of ntfs_file_get_block() in ntfs_mst_readpage() itself.
1319 I am aiming for correctness at the moment. Modularisation can come
1320 later.
1321 - Rename aops.c::end_buffer_read_index_async() to
1322 end_buffer_read_mst_async() and optimize the overflow checking and
1323 handling.
1324 - Use the host of the mftbmp address space mapping to hold the ntfs
1325 volume. This is needed so the async i/o completion handler can
1326 retrieve a pointer to the volume. Hopefully this will not cause
1327 problems elsewhere in the kernel... Otherwise will need to use a
1328 fake inode.
1329 - Complete implementation of proper handling of allocated_size vs
1330 initialized_size vs data_size (i.e. i_size) in whole driver.
1331 Basically aops.c is now completely rewritten.
1332 - Change NTFS driver name to just NTFS and set version number to 2.0.0
1333 to make a clear distinction from the old driver which is still on
1334 version 1.1.22.
1335
1336tng-0.0.8 - 08/03/2002 - Now using BitKeeper, http://linux-ntfs.bkbits.net/
1337
1338 - Replace bdevname(sb->s_dev) with sb->s_id.
1339 - Remove now superfluous new-line characters in all callers of
1340 ntfs_debug().
1341 - Apply kludge in ntfs_read_inode(), setting i_nlink to 1 for
1342 directories. Without this the "find" utility gets very upset which is
1343 fair enough as Linux/Unix do not support directory hard links.
1344 - Further runlist merging work. (Richard Russon)
1345 - Backwards compatibility for gcc-2.95. (Richard Russon)
1346 - Update to kernel 2.5.5-pre1 and rediff the now tiny patch.
Anton Altaparmakovc002f422005-02-03 12:02:56 +00001347 - Convert to new filesystem declaration using ->ntfs_get_sb() and
Linus Torvalds1da177e2005-04-16 15:20:36 -07001348 replacing ntfs_read_super() with ntfs_fill_super().
1349 - Set s_maxbytes to MAX_LFS_FILESIZE to avoid page cache page index
1350 overflow on 32-bit architectures.
1351 - Cleanup upcase loading code to use ntfs_(un)map_page().
1352 - Disable/reenable preemtion in critical sections of compession engine.
1353 - Replace device size determination in ntfs_fill_super() with
1354 sb->s_bdev->bd_inode->i_size (in bytes) and remove now superfluous
1355 function super.c::get_nr_blocks().
1356 - Implement a mount time option (show_inodes) allowing choice of which
1357 types of inode names readdir() returns and modify ntfs_filldir()
1358 accordingly. There are several parameters to show_inodes:
1359 system: system files
1360 win32: long file names (including POSIX file names) [DEFAULT]
1361 long: same as win32
1362 dos: short file names only (excluding POSIX file names)
1363 short: same as dos
1364 posix: same as both win32 and dos
1365 all: all file names
1366 Note that the options are additive, i.e. specifying:
1367 -o show_inodes=system,show_inodes=win32,show_inodes=dos
1368 is the same as specifying:
1369 -o show_inodes=all
1370 Note that the "posix" and "all" options will show all directory
1371 names, BUT the link count on each directory inode entry is set to 1,
1372 due to Linux not supporting directory hard links. This may well
1373 confuse some userspace applications, since the directory names will
1374 have the same inode numbers. Thus it is NOT advisable to use the
1375 "posix" or "all" options. We provide them only for completeness sake.
1376 - Add copies of allocated_size, initialized_size, and compressed_size to
1377 the ntfs inode structure and set them up in
1378 inode.c::ntfs_read_inode(). These reflect the unnamed data attribute
1379 for files and the index allocation attribute for directories.
1380 - Add copies of allocated_size and initialized_size to ntfs inode for
1381 $BITMAP attribute of large directories and set them up in
1382 inode.c::ntfs_read_inode().
1383 - Add copies of allocated_size and initialized_size to ntfs volume for
1384 $BITMAP attribute of $MFT and set them up in
1385 super.c::load_system_files().
1386 - Parse deprecated ntfs driver options (iocharset, show_sys_files,
1387 posix, and utf8) and tell user what the new options to use are. Note
1388 we still do support them but they will be removed with kernel 2.7.x.
1389 - Change all occurences of integer long long printf formatting to hex
1390 as printk() will not support long long integer format if/when the
1391 div64 patch goes into the kernel.
1392 - Make slab caches have stable names and change the names to what they
1393 were intended to be. These changes are required/made possible by the
1394 new slab cache name handling which removes the length limitation by
1395 requiring the caller of kmem_cache_create() to supply a stable name
1396 which is then referenced but not copied.
1397 - Rename run_list structure to run_list_element and create a new
1398 run_list structure containing a pointer to a run_list_element
1399 structure and a read/write semaphore. Adapt all users of runlists
1400 to new scheme and take and release the lock as needed. This fixes a
1401 nasty race as the run_list changes even when inodes are locked for
1402 reading and even when the inode isn't locked at all, so we really
1403 needed the serialization. We use a semaphore rather than a spinlock
1404 as memory allocations can sleep and doing everything GFP_ATOMIC
1405 would be silly.
1406 - Cleanup read_inode() removing all code checking for lowest_vcn != 0.
1407 This can never happen due to the nature of lookup_attr() and how we
1408 support attribute lists. If it did happen it would imply the inode
1409 being corrupt.
1410 - Check for lowest_vcn != 0 in ntfs_read_inode() and mark the inode as
1411 bad if found.
1412 - Update to 2.5.6-pre2 changes in struct address_space.
1413 - Use parent_ino() when accessing d_parent inode number in dir.c.
1414 - Import Sourceforge CVS repository into BitKeeper repository:
1415 http://linux-ntfs.bkbits.net/ntfs-tng-2.5
1416 - Update fs/Makefile, fs/Config.help, fs/Config.in, and
1417 Documentation/filesystems/ntfs.txt for NTFS TNG.
1418 - Create kernel configuration option controlling whether debugging
1419 is enabled or not.
1420 - Add the required export of end_buffer_io_sync() from the patches
1421 directory to the kernel code.
1422 - Update inode.c::ntfs_show_options() with show_inodes mount option.
1423 - Update errors mount option.
1424
1425tng-0.0.7 - 13/02/2002 - The driver is now feature complete for read-only!
1426
1427 - Cleanup mft.c and it's debug/error output in particular. Fix a minor
1428 bug in mapping of extent inodes. Update all the comments to fit all
1429 the recent code changes.
1430 - Modify vcn_to_lcn() to cope with entirely unmapped runlists.
1431 - Cleanups in compress.c, mostly comments and folding help.
1432 - Implement attrib.c::map_run_list() as a generic helper.
1433 - Make compress.c::ntfs_file_read_compressed_block() use map_run_list()
1434 thus making code shorter and enabling attribute list support.
1435 - Cleanup incorrect use of [su]64 with %L printf format specifier in
1436 all source files. Type casts to [unsigned] long long added to correct
1437 the mismatches (important for architectures which have long long not
1438 being 64 bits).
1439 - Merge async io completion handlers for directory indexes and $MFT
1440 data into one by setting the index_block_size{_bits} of the ntfs
1441 inode for $MFT to the mft_record_size{_bits} of the ntfs_volume.
1442 - Cleanup aops.c, update comments.
1443 - Make ntfs_file_get_block() use map_run_list() so all files now
1444 support attribute lists.
1445 - Make ntfs_dir_readpage() almost verbatim copy of
1446 block_read_full_page() by using ntfs_file_get_block() with only real
1447 difference being the use of our own async io completion handler
1448 rather than the default one, thus reducing the amount of code and
1449 automatically enabling attribute list support for directory indices.
1450 - Fix bug in load_attribute_list() - forgot to call brelse in error
1451 code path.
1452 - Change parameters to find_attr() and lookup_attr(). We no longer
1453 pass in the upcase table and its length. These can be gotten from
1454 ctx->ntfs_ino->vol->upcase{_len}. Update all callers.
1455 - Cleanups in attrib.c.
1456 - Implement merging of runlists, attrib.c::merge_run_lists() and its
1457 helpers. (Richard Russon)
1458 - Attribute lists part 2, attribute extents and multi part runlists:
1459 enable proper support for LCN_RL_NOT_MAPPED and automatic mapping of
1460 further runlist parts via attrib.c::map_run_list().
1461 - Tiny endianness bug fix in decompress_mapping_pairs().
1462
1463tng-0.0.6 - Encrypted directories, bug fixes, cleanups, debugging enhancements.
1464
1465 - Enable encrypted directories. (Their index root is marked encrypted
1466 to indicate that new files in that directory should be created
1467 encrypted.)
1468 - Fix bug in NInoBmpNonResident() macro. (Cut and paste error.)
1469 - Enable $Extend system directory. Most (if not all) extended system
1470 files do not have unnamed data attributes so ntfs_read_inode() had to
1471 special case them but that is ok, as the special casing recovery
1472 happens inside an error code path so there is zero slow down in the
1473 normal fast path. The special casing is done by introducing a new
1474 function inode.c::ntfs_is_extended_system_file() which checks if any
1475 of the hard links in the inode point to $Extend as being their parent
1476 directory and if they do we assume this is an extended system file.
1477 - Create a sysctl/proc interface to allow {dis,en}abling of debug output
1478 when compiled with -DDEBUG. Default is debug messages to be disabled.
1479 To enable them, one writes a non-zero value to /proc/sys/fs/ntfs-debug
1480 (if /proc is enabled) or uses sysctl(2) to effect the same (if sysctl
1481 interface is enabled). Inspired by old ntfs driver.
1482 - Add debug_msgs insmod/kernel boot parameter to set whether debug
1483 messages are {dis,en}abled. This is useful to enable debug messages
1484 during ntfs initialization and is the only way to activate debugging
1485 when the sysctl interface is not enabled.
1486 - Cleanup debug output in various places.
1487 - Remove all dollar signs ($) from the source (except comments) to
1488 enable compilation on architectures whose gcc compiler does not
1489 support dollar signs in the names of variables/constants. Attribute
1490 types now start with AT_ instead of $ and $I30 is now just I30.
1491 - Cleanup ntfs_lookup() and add consistency check of sequence numbers.
1492 - Load complete runlist for $MFT/$BITMAP during mount and cleanup
1493 access functions. This means we now cope with $MFT/$BITMAP being
1494 spread accross several mft records.
1495 - Disable modification of mft_zone_multiplier on remount. We can always
1496 reenable this later on if we really want to, but we will need to make
1497 sure we readjust the mft_zone size / layout accordingly.
1498
1499tng-0.0.5 - Modernize for 2.5.x and further in line-ing with Al Viro's comments.
1500
1501 - Use sb_set_blocksize() instead of set_blocksize() and verify the
1502 return value.
1503 - Use sb_bread() instead of bread() throughout.
1504 - Add index_vcn_size{_bits} to ntfs_inode structure to store the size
1505 of a directory index block vcn. Apply resulting simplifications in
1506 dir.c everywhere.
1507 - Fix a small bug somewhere (but forgot what it was).
1508 - Change ntfs_{debug,error,warning} to enable gcc to do type checking
1509 on the printf-format parameter list and fix bugs reported by gcc
1510 as a result. (Richard Russon)
1511 - Move inode allocation strategy to Al's new stuff but maintain the
1512 divorce of ntfs_inode from struct inode. To achieve this we have two
1513 separate slab caches, one for big ntfs inodes containing a struct
1514 inode and pure ntfs inodes and at the same time fix some faulty
1515 error code paths in ntfs_read_inode().
1516 - Show mount options in proc (inode.c::ntfs_show_options()).
1517
1518tng-0.0.4 - Big changes, getting in line with Al Viro's comments.
1519
1520 - Modified (un)map_mft_record functions to be common for read and write
1521 case. To specify which is which, added extra parameter at front of
1522 parameter list. Pass either READ or WRITE to this, each has the
1523 obvious meaning.
1524 - General cleanups to allow for easier folding in vi.
1525 - attrib.c::decompress_mapping_pairs() now accepts the old runlist
1526 argument, and invokes attrib.c::merge_run_lists() to merge the old
1527 and the new runlists.
1528 - Removed attrib.c::find_first_attr().
1529 - Implemented loading of attribute list and complete runlist for $MFT.
1530 This means we now cope with $MFT being spread across several mft
1531 records.
1532 - Adapt to 2.5.2-pre9 and the changed create_empty_buffers() syntax.
1533 - Adapt major/minor/kdev_t/[bk]devname stuff to new 2.5.x kernels.
1534 - Make ntfs_volume be allocated via kmalloc() instead of using a slab
1535 cache. There are too little ntfs_volume structures at any one time
1536 to justify a private slab cache.
1537 - Fix bogus kmap() use in async io completion. Now use kmap_atomic().
1538 Use KM_BIO_IRQ on advice from IRC/kernel...
1539 - Use ntfs_map_page() in map_mft_record() and create ->readpage method
1540 for reading $MFT (ntfs_mft_readpage). In the process create dedicated
1541 address space operations (ntfs_mft_aops) for $MFT inode mapping. Also
1542 removed the now superfluous exports from the kernel core patch.
1543 - Fix a bug where kfree() was used insted of ntfs_free().
1544 - Change map_mft_record() to take ntfs_inode as argument instead of
1545 vfs inode. Dito for unmap_mft_record(). Adapt all callers.
1546 - Add pointer to ntfs_volume to ntfs_inode.
1547 - Add mft record number and sequence number to ntfs_inode. Stop using
1548 i_ino and i_generation for in-driver purposes.
1549 - Implement attrib.c::merge_run_lists(). (Richard Russon)
1550 - Remove use of proper inodes by extent inodes. Move i_ino and
1551 i_generation to ntfs_inode to do this. Apply simplifications that
1552 result and remove iget_no_wait(), etc.
1553 - Pass ntfs_inode everywhere in the driver (used to be struct inode).
1554 - Add reference counting in ntfs_inode for the ntfs inode itself and
1555 for the mapped mft record.
1556 - Extend mft record mapping so we can (un)map extent mft records (new
1557 functions (un)map_extent_mft_record), and so mappings are reference
1558 counted and don't have to happen twice if already mapped - just ref
1559 count increases.
1560 - Add -o iocharset as alias to -o nls for backwards compatibility.
1561 - The latest core patch is now tiny. In fact just a single additional
1562 export is necessary over the base kernel.
1563
1564tng-0.0.3 - Cleanups, enhancements, bug fixes.
1565
1566 - Work on attrib.c::decompress_mapping_pairs() to detect base extents
1567 and setup the runlist appropriately using knowledge provided by the
1568 sizes in the base attribute record.
1569 - Balance the get_/put_attr_search_ctx() calls so we don't leak memory
1570 any more.
1571 - Introduce ntfs_malloc_nofs() and ntfs_free() to allocate/free a single
1572 page or use vmalloc depending on the amount of memory requested.
1573 - Cleanup error output. The __FUNCTION__ "(): " is now added
1574 automatically. Introduced a new header file debug.h to support this
1575 and also moved ntfs_debug() function into it.
1576 - Make reading of compressed files more intelligent and especially get
1577 rid of the vmalloc_nofs() from readpage(). This now uses per CPU
1578 buffers (allocated at first mount with cluster size <= 4kiB and
1579 deallocated on last umount with cluster size <= 4kiB), and
1580 asynchronous io for the compressed data using a list of buffer heads.
1581 Er, we use synchronous io as async io only works on whole pages
1582 covered by buffers and not on individual buffer heads...
1583 - Bug fix for reading compressed files with sparse compression blocks.
1584
1585tng-0.0.2 - Now handles larger/fragmented/compressed volumes/files/dirs.
1586
1587 - Fixed handling of directories when cluster size exceeds index block
1588 size.
1589 - Hide DOS only name space directory entries from readdir() but allow
1590 them in lookup(). This should fix the problem that Linux doesn't
1591 support directory hard links, while still allowing access to entries
1592 via their short file name. This also has the benefit of mimicking
1593 what Windows users are used to, so it is the ideal solution.
1594 - Implemented sync_page everywhere so no more hangs in D state when
1595 waiting for a page.
1596 - Stop using bforget() in favour of brelse().
1597 - Stop locking buffers unnecessarily.
1598 - Implemented compressed files (inode->mapping contains uncompressed
1599 data, raw compressed data is currently bread() into a vmalloc()ed
1600 memory buffer).
1601 - Enable compressed directories. (Their index root is marked compressed
1602 to indicate that new files in that directory should be created
1603 compressed.)
1604 - Use vsnprintf rather than vsprintf in the ntfs_error and ntfs_warning
1605 functions. (Thanks to Will Dyson for pointing this out.)
1606 - Moved the ntfs_inode and ntfs_volume (the former ntfs_inode_info and
1607 ntfs_sb_info) out of the common inode and super_block structures and
1608 started using the generic_ip and generic_sbp pointers instead. This
1609 makes ntfs entirely private with respect to the kernel tree.
1610 - Detect compiler version and abort with error message if gcc less than
1611 2.96 is used.
1612 - Fix bug in name comparison function in unistr.c.
1613 - Implement attribute lists part 1, the infrastructure: search contexts
1614 and operations, find_external_attr(), lookup_attr()) and make the
1615 code use the infrastructure.
1616 - Fix stupid buffer overflow bug that became apparent on larger run
1617 list containing attributes.
1618 - Fix bugs in readdir() that became apparent on larger directories.
1619
1620 The driver is now really useful and survives the test
1621 find . -type f -exec md5sum "{}" \;
1622 without any error messages on a over 1GiB sized partition with >16k
1623 files on it, including compressed files and directories and many files
1624 and directories with attribute lists.
1625
1626tng-0.0.1 - The first useful version.
1627
1628 - Added ntfs_lookup().
1629 - Added default upcase generation and handling.
1630 - Added compile options to be shown on module init.
1631 - Many bug fixes that were "hidden" before.
1632 - Update to latest kernel.
1633 - Added ntfs_readdir().
1634 - Added file operations for mmap(), read(), open() and llseek(). We just
1635 use the generic ones. The whole point of going through implementing
1636 readpage() methods and where possible get_block() call backs is that
1637 this allows us to make use of the generic high level methods provided
1638 by the kernel.
1639
1640 The driver is now actually useful! Yey. (-: It undoubtedly has got bugs
1641 though and it doesn't implement accesssing compressed files yet. Also,
1642 accessing files with attribute list attributes is not implemented yet
Anton Altaparmakovc002f422005-02-03 12:02:56 +00001643 either. But for small or simple filesystems it should work and allow
Linus Torvalds1da177e2005-04-16 15:20:36 -07001644 you to list directories, use stat on directory entries and the file
1645 system, open, read, mmap and llseek around in files. A big mile stone
1646 has been reached!
1647
1648tng-0.0.0 - Initial version tag.
1649
1650 Initial driver implementation. The driver can mount and umount simple
Anton Altaparmakovc002f422005-02-03 12:02:56 +00001651 NTFS filesystems (i.e. ones without attribute lists in the system
Linus Torvalds1da177e2005-04-16 15:20:36 -07001652 files). If the mount fails there might be problems in the error handling
1653 code paths, so be warned. Otherwise it seems to be loading the system
1654 files nicely and the mft record read mapping/unmapping seems to be
1655 working nicely, too. Proof of inode metadata in the page cache and non-
1656 resident file unnamed stream data in the page cache concepts is thus
1657 complete.