3 - Either invalidate quotas or update the quota charges on NTFS 3.x
4 volumes with quota tracking enabled ($Quota).
5 - Checkpoint or disable the user space journal ($UsnJrnl).
6 - Implement aops->set_page_dirty() in order to take control of buffer
7 dirtying. Not having it means if page_has_buffers(), all buffers
8 will be dirtied with the page. And if not they won't be. That is
9 fine for the moment but will break once we enable metadata updates.
10 For now just always using __set_page_dirty_nobuffers() for metadata
11 pages as nothing can dirty a page other than ourselves. Should this
12 change, we will really need to roll our own ->set_page_dirty().
13 - Implement sops->dirty_inode() to implement {a,m,c}time updates and
15 - Implement sops->write_inode().
16 - In between ntfs_prepare/commit_write, need exclusion between
17 simultaneous file extensions. Need perhaps an NInoResizeUnderway()
18 flag which we can set in ntfs_prepare_write() and clear again in
19 ntfs_commit_write(). Just have to be careful in readpage/writepage,
20 as well as in truncate, that we play nice... We might need to have
21 a data_size field in the ntfs_inode to store the real attribute
22 length. Also need to be careful with initialized_size extention in
23 ntfs_prepare_write. Basically, just be _very_ careful in this code...
24 OTOH, perhaps i_sem, which is held accross generic_file_write is
25 sufficient for synchronisation here. We then just need to make sure
26 ntfs_readpage/writepage/truncate interoperate properly with us.
28 2.1.11 - Driver internal cleanups.
30 - Only build logfile.o if building the driver with read-write support.
31 - Really final white space cleanups.
32 - Use generic_ffs() instead of ffs() in logfile.c which allows the
33 log_page_size variable to be optimized by gcc into a constant.
34 - Rename uchar_t to ntfschar everywhere as uchar_t is unsigned 1-byte
35 char as defined by POSIX and as found on some systems.
37 2.1.10 - Force read-only (re)mounting of volumes with unsupported volume flags.
39 - Finish off the white space cleanups (remove trailing spaces, etc).
40 - Clean up ntfs_fill_super() and ntfs_read_inode_mount() by removing
41 the kludges around the first iget(). Instead of (re)setting ->s_op
42 we have the $MFT inode set up by explicit new_inode() / set ->i_ino /
43 insert_inode_hash() / call ntfs_read_inode_mount() directly. This
44 kills the need for second super_operations and allows to return error
45 from ntfs_read_inode_mount() without resorting to ugly "poisoning"
47 - Force read-only (re)mounting if any of the following bits are set in
48 the volume information flags:
49 VOLUME_IS_DIRTY, VOLUME_RESIZE_LOG_FILE,
50 VOLUME_UPGRADE_ON_MOUNT, VOLUME_DELETE_USN_UNDERWAY,
51 VOLUME_REPAIR_OBJECT_ID, VOLUME_MODIFIED_BY_CHKDSK
52 To make this easier we define VOLUME_MUST_MOUNT_RO_MASK with all the
53 above bits set so the test is made easy.
55 2.1.9 - Fix two bugs in decompression engine.
57 - Fix a bug where we would not always detect that we have reached the
58 end of a compression block because we were ending at minus one byte
59 which is effectively the same as being at the end. The fix is to
60 check whether the uncompressed buffer has been fully filled and if so
61 we assume we have reached the end of the compression block. A big
62 thank you to Marcin GibuĊa for the bug report, the assistance in
63 tracking down the bug and testing the fix.
64 - Fix a possible bug where when a compressed read is truncated to the
65 end of the file, the offset inside the last page was not truncated.
67 2.1.8 - Handle $MFT mirror and $LogFile, improve time handling, and cleanups.
69 - Use get_bh() instead of manual atomic_inc() in fs/ntfs/compress.c.
70 - Modify fs/ntfs/time.c::ntfs2utc(), get_current_ntfs_time(), and
71 utc2ntfs() to work with struct timespec instead of time_t on the
72 Linux UTC time side thus preserving the full precision of the NTFS
73 time and only loosing up to 99 nano-seconds in the Linux UTC time.
74 - Move fs/ntfs/time.c to fs/ntfs/time.h and make the time functions
76 - Remove unused ntfs_dirty_inode().
77 - Cleanup super operations declaration in fs/ntfs/super.c.
78 - Wrap flush_dcache_mft_record_page() in #ifdef NTFS_RW.
79 - Add NInoTestSetFoo() and NInoTestClearFoo() macro magic to
80 fs/ntfs/inode.h and use it to declare NInoTest{Set,Clear}Dirty.
81 - Move typedefs for ntfs_attr and test_t from fs/ntfs/inode.c to
82 fs/ntfs/inode.h so they can be used elsewhere.
83 - Determine the mft mirror size as the number of mirrored mft records
84 and store it in ntfs_volume->mftmirr_size (fs/ntfs/super.c).
85 - Load the mft mirror at mount time and compare the mft records stored
86 in it to the ones in the mft. Force a read-only mount if the two do
87 not match (fs/ntfs/super.c).
88 - Fix type casting related warnings on 64-bit architectures. Thanks
89 to Meelis Roos for reporting them.
90 - Move %L to %ll as %L is floating point and %ll is integer which is
92 - Read the journal ($LogFile) and determine if the volume has been
93 shutdown cleanly and force a read-only mount if not (fs/ntfs/super.c
94 and fs/ntfs/logfile.c). This is a little bit of a crude check in
95 that we only look at the restart areas and not at the actual log
96 records so that there will be a very small number of cases where we
97 think that a volume is dirty when in fact it is clean. This should
98 only affect volumes that have not been shutdown cleanly and did not
99 have any pending, non-check-pointed i/o.
100 - If the $LogFile indicates a clean shutdown and a read-write (re)mount
101 is requested, empty $LogFile by overwriting it with 0xff bytes to
102 ensure that Windows cannot cause data corruption by replaying a stale
103 journal after Linux has written to the volume.
105 2.1.7 - Enable NFS exporting of mounted NTFS volumes.
107 - Set i_generation in the VFS inode from the seq_no of the NTFS inode.
108 - Make ntfs_lookup() NFS export safe, i.e. use d_splice_alias(), etc.
109 - Implement ->get_dentry() in fs/ntfs/namei.c::ntfs_get_dentry() as the
110 default doesn't allow inode number 0 which is a valid inode on NTFS
111 and even if it did allow that it uses iget() instead of ntfs_iget()
112 which makes it useless for us.
113 - Implement ->get_parent() in fs/ntfs/namei.c::ntfs_get_parent() as the
114 default just returns -EACCES which is not very useful.
115 - Define export operations (->s_export_op) for NTFS (ntfs_export_ops)
116 and set them up in the super block at mount time (super.c) this
117 allows mounted NTFS volumes to be exported via NFS.
118 - Add missing return -EOPNOTSUPP; in
119 fs/ntfs/aops.c::ntfs_commit_nonresident_write().
120 - Enforce no atime and no dir atime updates at mount/remount time as
121 they are not implemented yet anyway.
122 - Move a few assignments in fs/ntfs/attrib.c::load_attribute_list() to
123 after a NULL check. Thanks to Dave Jones for pointing this out.
125 2.1.6 - Fix minor bug in handling of compressed directories.
127 - Fix bug in handling of compressed directories. A compressed
128 directory is not really compressed so when we set the ->i_blocks
129 field of a compressed directory inode we were setting it from the
130 non-existing field ni->itype.compressed.size which gave random
131 results... For directories we now always use ni->allocated_size.
133 2.1.5 - Fix minor bug in attribute list attribute handling.
135 - Fix bug in attribute list handling. Actually it is not as much a bug
136 as too much protection in that we were not allowing attribute lists
137 which waste space on disk while Windows XP clearly allows it and in
138 fact creates such attribute lists so our driver was failing.
139 - Update NTFS documentation ready for 2.6 kernel release.
141 2.1.4 - Reduce compiler requirements.
143 - Remove all uses of unnamed structs and unions in the driver to make
144 old and newer gcc versions happy. Makes it a bit uglier IMO but at
145 least people will stop hassling me about it.
147 2.1.3 - Important bug fixes in corner cases.
149 - super.c::parse_ntfs_boot_sector(): Correct the check for 64-bit
150 clusters. (Philipp Thomas)
151 - attrib.c::load_attribute_list(): Fix bug when initialized_size is a
152 multiple of the block_size but not the cluster size. (Szabolcs
153 Szakacsits <szaka@sienet.hu>)
155 2.1.2 - Important bug fixes aleviating the hangs in statfs.
157 - Fix buggy free cluster and free inode determination logic.
159 2.1.1 - Minor updates.
161 - Add handling for initialized_size != data_size in compressed files.
162 - Reduce function local stack usage from 0x3d4 bytes to just noise in
163 fs/ntfs/upcase.c. (Randy Dunlap <rddunlap@osdl.ord>)
164 - Remove compiler warnings for newer gcc.
165 - Pages are no longer kmapped by mm/filemap.c::generic_file_write()
166 around calls to ->{prepare,commit}_write. Adapt NTFS appropriately
167 in fs/ntfs/aops.c::ntfs_prepare_nonresident_write() by using
168 kmap_atomic(KM_USER0).
170 2.1.0 - First steps towards write support: implement file overwrite.
172 - Add configuration option for developmental write support with an
173 appropriately scary configuration help text.
174 - Initial implementation of fs/ntfs/aops.c::ntfs_writepage() and its
175 helper fs/ntfs/aops.c::ntfs_write_block(). This enables mmap(2) based
176 overwriting of existing files on ntfs. Note: Resident files are
177 only written into memory, and not written out to disk at present, so
178 avoid writing to files smaller than about 1kiB.
179 - Initial implementation of fs/ntfs/aops.c::ntfs_prepare_write(), its
180 helper fs/ntfs/aops.c::ntfs_prepare_nonresident_write() and their
181 counterparts, fs/ntfs/aops.c::ntfs_commit_write(), and
182 fs/ntfs/aops.c::ntfs_commit_nonresident_write(), respectively. Also,
183 add generic_file_write() to the ntfs file operations (fs/ntfs/file.c).
184 This enables write(2) based overwriting of existing files on ntfs.
185 Note: As with mmap(2) based overwriting, resident files are only
186 written into memory, and not written out to disk at present, so avoid
187 writing to files smaller than about 1kiB.
188 - Implement ->truncate (fs/ntfs/inode.c::ntfs_truncate()) and
189 ->setattr() (fs/ntfs/inode.c::ntfs_setattr()) inode operations for
190 files with the purpose of intercepting and aborting all i_size
191 changes which we do not support yet. ntfs_truncate() actually only
192 emits a warning message but AFAICS our interception of i_size changes
193 elsewhere means ntfs_truncate() never gets called for i_size changes.
194 It is only called from generic_file_write() when we fail in
195 ntfs_prepare_{,nonresident_}write() in order to discard any
196 instantiated buffers beyond i_size. Thus i_size is not actually
197 changed so our warning message is enough. Unfortunately it is not
198 possible to easily determine if i_size is being changed or not hence
199 we just emit an appropriately worded error message.
201 2.0.25 - Small bug fixes and cleanups.
203 - Unlock the page in an out of memory error code path in
204 fs/ntfs/aops.c::ntfs_read_block().
205 - If fs/ntfs/aops.c::ntfs_read_page() is called on an uptodate page,
206 just unlock the page and return. (This can happen due to ->writepage
207 clearing PageUptodate() during write out of MstProtected()
209 - Remove leaked write code again.
213 - Treat BUG_ON() as ASSERT() not VERIFY(), i.e. do not use side effects
214 inside BUG_ON(). (Adam J. Richter)
215 - Split logical OR expressions inside BUG_ON() into individual BUG_ON()
216 calls for improved debugging. (Adam J. Richter)
217 - Add errors flag to the ntfs volume state, accessed via
218 NVol{,Set,Clear}Errors(vol).
219 - Do not allow read-write remounts of read-only volumes with errors.
220 - Clarify comment for ntfs file operation sendfile which was added by
221 Christoph Hellwig a while ago (just using generic_file_sendfile())
222 to say that ntfs ->sendfile is only used for the case where the
223 source data is on the ntfs partition and the destination is
224 somewhere else, i.e. nothing we need to concern ourselves with.
225 - Add generic_file_write() as our ntfs file write operation.
227 2.0.23 - Major bug fixes (races, deadlocks, non-i386 architectures).
229 - Massive internal locking changes to mft record locking. Fixes lock
230 recursion and replaces the mrec_lock read/write semaphore with a
231 mutex. Also removes the now superfluous mft_count. This fixes several
232 race conditions and deadlocks, especially in the future write code.
233 - Fix ntfs over loopback for compressed files by adding an
234 optimization barrier. (gcc was screwing up otherwise ?)
235 - Miscellaneous cleanups all over the code and a fix or two in error
237 Thanks go to Christoph Hellwig for pointing out the following two:
238 - Remove now unused function fs/ntfs/malloc.h::vmalloc_nofs().
239 - Fix ntfs_free() for ia64 and parisc by checking for VMALLOC_END, too.
241 2.0.22 - Cleanups, mainly to ntfs_readdir(), and use C99 initializers.
243 - Change fs/ntfs/dir.c::ntfs_reddir() to only read/write ->f_pos once
244 at entry/exit respectively.
245 - Use C99 initializers for structures.
246 - Remove unused variable blocks from fs/ntfs/aops.c::ntfs_read_block().
248 2.0.21 - Check for, and refuse to work with too large files/directories/volumes.
250 - Limit volume size at mount time to 2TiB on architectures where
251 unsigned long is 32-bits (fs/ntfs/super.c::parse_ntfs_boot_sector()).
252 This is the most we can do without overflowing the 32-bit limit of
253 the block device size imposed on us by sb_bread() and sb_getblk()
255 - Limit file/directory size at open() time to 16TiB on architectures
256 where unsigned long is 32-bits (fs/ntfs/file.c::ntfs_file_open() and
257 fs/ntfs/dir.c::ntfs_dir_open()). This is the most we can do without
258 overflowing the page cache page index.
260 2.0.20 - Support non-resident directory index bitmaps, fix page leak in readdir.
262 - Move the directory index bitmap to use an attribute inode instead of
263 having special fields for it inside the ntfs inode structure. This
264 means that the index bitmaps now use the page cache for i/o, too,
265 and also as a side effect we get support for non-resident index
267 - Simplify/cleanup error handling in fs/ntfs/dir.c::ntfs_readdir() and
268 fix a page leak that manifested itself in some cases.
269 - Add fs/ntfs/inode.c::ntfs_put_inode(), which we need to release the
270 index bitmap inode on the final iput().
272 2.0.19 - Fix race condition, improvements, and optimizations in i/o interface.
274 - Apply block optimization added to fs/ntfs/aops.c::ntfs_read_block()
275 to fs/ntfs/compress.c::ntfs_file_read_compressed_block() as well.
276 - Drop the "file" from ntfs_file_read_compressed_block().
277 - Rename fs/ntfs/aops.c::ntfs_enb_buffer_read_async() to
278 ntfs_end_buffer_async_read() (more like the fs/buffer.c counterpart).
279 - Update ntfs_end_buffer_async_read() with the improved logic from
280 its updated counterpart fs/buffer.c::end_buffer_async_read(). Apply
281 further logic improvements to better determine when we set PageError.
282 - Update submission of buffers in fs/ntfs/aops.c::ntfs_read_block() to
283 check for the buffers being uptodate first in line with the updated
284 fs/buffer.c::block_read_full_page(). This plugs a small race
287 2.0.18 - Fix race condition in reading of compressed files.
289 - There was a narrow window between checking a buffer head for being
290 uptodate and locking it in ntfs_file_read_compressed_block(). We now
291 lock the buffer and then check whether it is uptodate or not.
293 2.0.17 - Cleanups and optimizations - shrinking the ToDo list.
295 - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to return an error
296 code and update callers, i.e. ntfs_iget(), to pass that error code
297 up instead of just using -EIO.
298 - Modifications to super.c to ensure that both mount and remount
299 cannot set any write related options when the driver is compiled
301 - Optimize block resolution in fs/ntfs/aops.c::ntfs_read_block() to
302 cache the current run list element. This should improve performance
303 when reading very large and/or very fragmented data.
305 2.0.16 - Convert access to $MFT/$BITMAP to attribute inode API.
307 - Fix a stupid bug introduced in 2.0.15 where we were unmapping the
308 wrong inode in fs/ntfs/inode.c::ntfs_attr_iget().
309 - Fix debugging check in fs/ntfs/aops.c::ntfs_read_block().
310 - Convert $MFT/$BITMAP access to attribute inode API and remove all
311 remnants of the ugly mftbmp address space and operations hack. This
312 means we finally have only one readpage function as well as only one
313 async io completion handler. Yey! The mft bitmap is now just an
314 attribute inode and is accessed from vol->mftbmp_ino just as if it
315 were a normal file. Fake inodes rule. (-:
317 2.0.15 - Fake inodes based attribute i/o via the pagecache, fixes and cleanups.
319 - Fix silly bug in fs/ntfs/super.c::parse_options() which was causing
320 remounts to fail when the partition had an entry in /etc/fstab and
321 the entry specified the nls= option.
322 - Apply same macro magic used in fs/ntfs/inode.h to fs/ntfs/volume.h to
323 expand all the helper functions NVolFoo(), NVolSetFoo(), and
325 - Move copyright statement from driver initialisation message to
326 module description (fs/super.c). This makes the initialisation
327 message fit on one line and fits in better with rest of kernel.
328 - Update fs/ntfs/attrib.c::map_run_list() to work on both real and
329 attribute inodes, and both for files and directories.
330 - Implement fake attribute inodes allowing all attribute i/o to go via
331 the page cache and to use all the normal vfs/mm functionality:
332 - Add ntfs_attr_iget() and its helper ntfs_read_locked_attr_inode()
334 - Add needed cleanup code to ntfs_clear_big_inode().
335 - Merge address space operations for files and directories (aops.c),
336 now just have ntfs_aops:
338 end_buffer_read_attr_async() -> ntfs_end_buffer_read_async(),
339 ntfs_attr_read_block() -> ntfs_read_block(),
340 ntfs_file_read_page() -> ntfs_readpage().
341 - Rewrite fs/ntfs/aops.c::ntfs_readpage() to work on both real and
342 attribute inodes, and both for files and directories.
343 - Remove obsolete fs/ntfs/aops.c::ntfs_mst_readpage().
345 2.0.14 - Run list merging code cleanup, minor locking changes, typo fixes.
347 - Change fs/ntfs/super.c::ntfs_statfs() to not rely on BKL by moving
348 the locking out of super.c::get_nr_free_mft_records() and taking and
349 dropping the mftbmp_lock rw_semaphore in ntfs_statfs() itself.
350 - Bring attribute run list merging code (fs/ntfs/attrib.c) in sync with
351 current userspace ntfs library code. This means that if a merge
352 fails the original run lists are always left unmodified instead of
353 being silently corrupted.
356 2.0.13 - Use iget5_locked() in preparation for fake inodes and small cleanups.
358 - Remove nr_mft_bits and the now superfluous union with nr_mft_records
359 from ntfs_volume structure.
360 - Remove nr_lcn_bits and the now superfluous union with nr_clusters
361 from ntfs_volume structure.
362 - Use iget5_locked() and friends instead of conventional iget(). Wrap
363 the call in fs/ntfs/inode.c::ntfs_iget() and update callers of iget()
364 to use ntfs_iget(). Leave only one iget() call at mount time so we
365 don't need an ntfs_iget_mount().
366 - Change fs/ntfs/inode.c::ntfs_new_extent_inode() to take mft_no as an
369 2.0.12 - Initial cleanup of address space operations following 2.0.11 changes.
371 - Merge fs/ntfs/aops.c::end_buffer_read_mst_async() and
372 fs/ntfs/aops.c::end_buffer_read_file_async() into one function
373 fs/ntfs/aops.c::end_buffer_read_attr_async() using NInoMstProtected()
374 to determine whether to apply mst fixups or not.
375 - Above change allows merging fs/ntfs/aops.c::ntfs_file_read_block()
376 and fs/ntfs/aops.c::ntfs_mst_readpage() into one function
377 fs/ntfs/aops.c::ntfs_attr_read_block(). Also, create a tiny wrapper
378 fs/ntfs/aops.c::ntfs_mst_readpage() to transform the parameters from
379 the VFS readpage function prototype to the ntfs_attr_read_block()
382 2.0.11 - Initial preparations for fake inode based attribute i/o.
384 - Move definition of ntfs_inode_state_bits to fs/ntfs/inode.h and
385 do some macro magic (adapted from include/linux/buffer_head.h) to
386 expand all the helper functions NInoFoo(), NInoSetFoo(), and
388 - Add new flag to ntfs_inode_state_bits: NI_Sparse.
389 - Add new fields to ntfs_inode structure to allow use of fake inodes
390 for attribute i/o: type, name, name_len. Also add new state bits:
391 NI_Attr, which, if set, indicates the inode is a fake inode, and
392 NI_MstProtected, which, if set, indicates the attribute uses multi
393 sector transfer protection, i.e. fixups need to be applied after
394 reads and before/after writes.
395 - Rename fs/ntfs/inode.c::ntfs_{new,clear,destroy}_inode() to
396 ntfs_{new,clear,destroy}_extent_inode() and update callers.
397 - Use ntfs_clear_extent_inode() in fs/ntfs/inode.c::__ntfs_clear_inode()
398 instead of ntfs_destroy_extent_inode().
399 - Cleanup memory deallocations in {__,}ntfs_clear_{,big_}inode().
400 - Make all operations on ntfs inode state bits use the NIno* functions.
401 - Set up the new ntfs inode fields and state bits in
402 fs/ntfs/inode.c::ntfs_read_inode() and add appropriate cleanup of
403 allocated memory to __ntfs_clear_inode().
404 - Cleanup ntfs_inode structure a bit for better ordering of elements
405 w.r.t. their size to allow better packing of the structure in memory.
407 2.0.10 - There can only be 2^32 - 1 inodes on an NTFS volume.
409 - Add check at mount time to verify that the number of inodes on the
410 volume does not exceed 2^32 - 1, which is the maximum allowed for
411 NTFS according to Microsoft.
412 - Change mft_no member of ntfs_inode structure to be unsigned long.
413 Update all users. This makes ntfs_inode->mft_no just a copy of struct
414 inode->i_ino. But we can't just always use struct inode->i_ino and
415 remove mft_no because extent inodes do not have an attached struct
418 2.0.9 - Decompression engine now uses a single buffer and other cleanups.
420 - Change decompression engine to use a single buffer protected by a
421 spin lock instead of per-CPU buffers. (Rusty Russell)
422 - Do not update cb_pos when handling a partial final page during
423 decompression of a sparse compression block, as the value is later
424 reset without being read/used. (Rusty Russell)
425 - Switch to using the new KM_BIO_SRC_IRQ for atomic kmap()s. (Andrew
427 - Change buffer size in ntfs_readdir()/ntfs_filldir() to use
428 NLS_MAX_CHARSET_SIZE which makes the buffers almost 1kiB each but
429 it also makes everything safer so it is a good thing.
430 - Miscellaneous minor cleanups to comments.
432 2.0.8 - Major updates for handling of case sensitivity and dcache aliasing.
434 Big thanks go to Al Viro and other inhabitants of #kernel for investing
435 their time to discuss the case sensitivity and dcache aliasing issues.
437 - Remove unused source file fs/ntfs/attraops.c.
438 - Remove show_inodes mount option(s), thus dropping support for
439 displaying of short file names.
440 - Remove deprecated mount option posix.
441 - Restore show_sys_files mount option.
442 - Add new mount option case_sensitive, to determine if the driver
443 treats file names as case sensitive or not. If case sensitive, create
444 file names in the POSIX namespace. Otherwise create file names in the
445 LONG/WIN32 namespace. Note, files remain accessible via their short
446 file name, if it exists.
447 - Remove really dumb logic bug in boot sector recovery code.
448 - Fix dcache aliasing issues wrt short/long file names via changes
449 to fs/ntfs/dir.c::ntfs_lookup_inode_by_name() and
450 fs/ntfs/namei.c::ntfs_lookup():
451 - Add additional argument to ntfs_lookup_inode_by_name() in which we
452 return information about the matching file name if the case is not
453 matching or the match is a short file name. See comments above the
454 function definition for details.
455 - Change ntfs_lookup() to only create dcache entries for the correctly
456 cased file name and only for the WIN32 namespace counterpart of DOS
457 namespace file names. This ensures we have only one dentry per
458 directory and also removes all dcache aliasing issues between short
459 and long file names once we add write support. See comments above
460 function for details.
461 - Fix potential 1 byte overflow in fs/ntfs/unistr.c::ntfs_ucstonls().
463 2.0.7 - Minor cleanups and updates for changes in core kernel code.
465 - Remove much of the NULL struct element initializers.
466 - Various updates to make compatible with recent kernels.
467 - Remove defines of MAX_BUF_PER_PAGE and include linux/buffer_head.h
468 in fs/ntfs/ntfs.h instead.
469 - Remove no longer needed KERNEL_VERSION checks. We are now in the
470 kernel proper so they are no longer needed.
472 2.0.6 - Major bugfix to make compatible with other kernel changes.
474 - Initialize the mftbmp address space properly now that there are more
475 fields in the struct address_space. This was leading to hangs and
476 oopses on umount since 2.5.12 because of changes to other parts of
477 the kernel. We probably want a kernel generic init_address_space()
479 - Drop BKL from ntfs_readdir() after consultation with Al Viro. The
480 only caller of ->readdir() is vfs_readdir() which holds i_sem during
481 the call, and i_sem is sufficient protection against changes in the
482 directory inode (including ->i_size).
483 - Use generic_file_llseek() for directories (as opposed to
484 default_llseek()) as this downs i_sem instead of the BKL which is
485 what we now need for exclusion against ->f_pos changes considering we
486 no longer take the BKL in ntfs_readdir().
488 2.0.5 - Major bugfix. Buffer overflow in extent inode handling.
490 - No need to set old blocksize in super.c::ntfs_fill_super() as the
491 VFS does so via invocation of deactivate_super() calling
492 fs->fill_super() calling block_kill_super() which does it.
493 - BKL moved from VFS into dir.c::ntfs_readdir(). (Linus Torvalds)
494 -> Do we really need it? I don't think so as we have exclusion on
495 the directory ntfs_inode rw_semaphore mrec_lock. We mmight have to
496 move the ->f_pos accesses under the mrec_lock though. Check this...
497 - Fix really, really, really stupid buffer overflow in extent inode
498 handling in mft.c::map_extent_mft_record().
500 2.0.4 - Cleanups and updates for kernel 2.5.11.
502 - Add documentation on how to use the MD driver to be able to use NTFS
503 stripe and volume sets in Linux and generally cleanup documentation
505 Remove all uses of kdev_t in favour of struct block_device *:
506 - Change compress.c::ntfs_file_read_compressed_block() to use
507 sb_getblk() instead of getblk().
508 - Change super.c::ntfs_fill_super() to use bdev_hardsect_size() instead
509 of get_hardsect_size().
510 - No need to get old blocksize in super.c::ntfs_fill_super() as
511 fs/super.c::get_sb_bdev() already does this.
512 - Set bh->b_bdev instead of bh->b_dev throughout aops.c.
514 2.0.3 - Small bug fixes, cleanups, and performance improvements.
516 - Remove some dead code from mft.c.
517 - Optimize readpage and read_block functions throughout aops.c so that
518 only initialized blocks are read. Non-initialized ones have their
519 buffer head mapped, zeroed, and set up to date, without scheduling
520 any i/o. Thanks to Al Viro for advice on how to avoid the device i/o.
521 Thanks go to Andrew Morton for spotting the below:
522 - Fix buglet in allocate_compression_buffers() error code path.
523 - Call flush_dcache_page() after modifying page cache page contents in
524 ntfs_file_readpage().
525 - Check for existence of page buffers throughout aops.c before calling
526 create_empty_buffers(). This happens when an I/O error occurs and the
527 read is retried. (It also happens once writing is implemented so that
528 needed doing anyway but I had left it for later...)
529 - Don't BUG_ON() uptodate and/or mapped buffers throughout aops.c in
530 readpage and read_block functions. Reasoning same as above (i.e. I/O
531 error retries and future write code paths.)
533 2.0.2 - Minor updates and cleanups.
535 - Cleanup: rename mst.c::__post_read_mst_fixup to post_write_mst_fixup
536 and cleanup the code a bit, removing the unused size parameter.
537 - Change default fmask to 0177 and update documentation.
538 - Change attrib.c::get_attr_search_ctx() to return the search context
539 directly instead of taking the address of a pointer. A return value
540 of NULL means the allocation failed. Updated all callers
542 - Update to 2.5.9 kernel (preserving backwards compatibility) by
543 replacing all occurences of page->buffers with page_buffers(page).
544 - Fix minor bugs in run list merging, also minor cleanup.
545 - Updates to bootsector layout and mft mirror contents descriptions.
546 - Small bug fix in error detection in unistr.c and some cleanups.
547 - Grow name buffer allocations in unistr.c in aligned mutlipled of 64
550 2.0.1 - Minor updates.
552 - Make default umask correspond to documentation.
553 - Improve documentation.
554 - Set default mode to include execute bit. The {u,f,d}mask can be used
555 to take it away if desired. This allows binaries to be executed from
556 a mounted ntfs partition.
558 2.0.0 - New version number. Remove TNG from the name. Now in the kernel.
560 - Add kill_super, just keeping up with the vfs changes in the kernel.
561 - Repeat some changes from tng-0.0.8 that somehow got lost on the way
562 from the CVS import into BitKeeper.
563 - Begin to implement proper handling of allocated_size vs
564 initialized_size vs data_size (i.e. i_size). Done are
565 mft.c::ntfs_mft_readpage(), aops.c::end_buffer_read_index_async(),
566 and attrib.c::load_attribute_list().
567 - Lock the run list in attrib.c::load_attribute_list() while using it.
568 - Fix memory leak in ntfs_file_read_compressed_block() and generally
569 clean up compress.c a little, removing some uncommented/unused debug
571 - Tidy up dir.c a little bit.
572 - Don't bother getting the run list in inode.c::ntfs_read_inode().
573 - Merge mft.c::ntfs_mft_readpage() and aops.c::ntfs_index_readpage()
574 creating aops.c::ntfs_mst_readpage(), improving the handling of
575 holes and overflow in the process and implementing the correct
576 equivalent of ntfs_file_get_block() in ntfs_mst_readpage() itself.
577 I am aiming for correctness at the moment. Modularisation can come
579 - Rename aops.c::end_buffer_read_index_async() to
580 end_buffer_read_mst_async() and optimize the overflow checking and
582 - Use the host of the mftbmp address space mapping to hold the ntfs
583 volume. This is needed so the async i/o completion handler can
584 retrieve a pointer to the volume. Hopefully this will not cause
585 problems elsewhere in the kernel... Otherwise will need to use a
587 - Complete implementation of proper handling of allocated_size vs
588 initialized_size vs data_size (i.e. i_size) in whole driver.
589 Basically aops.c is now completely rewritten.
590 - Change NTFS driver name to just NTFS and set version number to 2.0.0
591 to make a clear distinction from the old driver which is still on
594 tng-0.0.8 - 08/03/2002 - Now using BitKeeper, http://linux-ntfs.bkbits.net/
596 - Replace bdevname(sb->s_dev) with sb->s_id.
597 - Remove now superfluous new-line characters in all callers of
599 - Apply kludge in ntfs_read_inode(), setting i_nlink to 1 for
600 directories. Without this the "find" utility gets very upset which is
601 fair enough as Linux/Unix do not support directory hard links.
602 - Further run list merging work. (Richard Russon)
603 - Backwards compatibility for gcc-2.95. (Richard Russon)
604 - Update to kernel 2.5.5-pre1 and rediff the now tiny patch.
605 - Convert to new file system declaration using ->ntfs_get_sb() and
606 replacing ntfs_read_super() with ntfs_fill_super().
607 - Set s_maxbytes to MAX_LFS_FILESIZE to avoid page cache page index
608 overflow on 32-bit architectures.
609 - Cleanup upcase loading code to use ntfs_(un)map_page().
610 - Disable/reenable preemtion in critical sections of compession engine.
611 - Replace device size determination in ntfs_fill_super() with
612 sb->s_bdev->bd_inode->i_size (in bytes) and remove now superfluous
613 function super.c::get_nr_blocks().
614 - Implement a mount time option (show_inodes) allowing choice of which
615 types of inode names readdir() returns and modify ntfs_filldir()
616 accordingly. There are several parameters to show_inodes:
618 win32: long file names (including POSIX file names) [DEFAULT]
620 dos: short file names only (excluding POSIX file names)
622 posix: same as both win32 and dos
624 Note that the options are additive, i.e. specifying:
625 -o show_inodes=system,show_inodes=win32,show_inodes=dos
626 is the same as specifying:
628 Note that the "posix" and "all" options will show all directory
629 names, BUT the link count on each directory inode entry is set to 1,
630 due to Linux not supporting directory hard links. This may well
631 confuse some userspace applications, since the directory names will
632 have the same inode numbers. Thus it is NOT advisable to use the
633 "posix" or "all" options. We provide them only for completeness sake.
634 - Add copies of allocated_size, initialized_size, and compressed_size to
635 the ntfs inode structure and set them up in
636 inode.c::ntfs_read_inode(). These reflect the unnamed data attribute
637 for files and the index allocation attribute for directories.
638 - Add copies of allocated_size and initialized_size to ntfs inode for
639 $BITMAP attribute of large directories and set them up in
640 inode.c::ntfs_read_inode().
641 - Add copies of allocated_size and initialized_size to ntfs volume for
642 $BITMAP attribute of $MFT and set them up in
643 super.c::load_system_files().
644 - Parse deprecated ntfs driver options (iocharset, show_sys_files,
645 posix, and utf8) and tell user what the new options to use are. Note
646 we still do support them but they will be removed with kernel 2.7.x.
647 - Change all occurences of integer long long printf formatting to hex
648 as printk() will not support long long integer format if/when the
649 div64 patch goes into the kernel.
650 - Make slab caches have stable names and change the names to what they
651 were intended to be. These changes are required/made possible by the
652 new slab cache name handling which removes the length limitation by
653 requiring the caller of kmem_cache_create() to supply a stable name
654 which is then referenced but not copied.
655 - Rename run_list structure to run_list_element and create a new
656 run_list structure containing a pointer to a run_list_element
657 structure and a read/write semaphore. Adapt all users of run lists
658 to new scheme and take and release the lock as needed. This fixes a
659 nasty race as the run_list changes even when inodes are locked for
660 reading and even when the inode isn't locked at all, so we really
661 needed the serialization. We use a semaphore rather than a spinlock
662 as memory allocations can sleep and doing everything GFP_ATOMIC
664 - Cleanup read_inode() removing all code checking for lowest_vcn != 0.
665 This can never happen due to the nature of lookup_attr() and how we
666 support attribute lists. If it did happen it would imply the inode
668 - Check for lowest_vcn != 0 in ntfs_read_inode() and mark the inode as
670 - Update to 2.5.6-pre2 changes in struct address_space.
671 - Use parent_ino() when accessing d_parent inode number in dir.c.
672 - Import Sourceforge CVS repository into BitKeeper repository:
673 http://linux-ntfs.bkbits.net/ntfs-tng-2.5
674 - Update fs/Makefile, fs/Config.help, fs/Config.in, and
675 Documentation/filesystems/ntfs.txt for NTFS TNG.
676 - Create kernel configuration option controlling whether debugging
678 - Add the required export of end_buffer_io_sync() from the patches
679 directory to the kernel code.
680 - Update inode.c::ntfs_show_options() with show_inodes mount option.
681 - Update errors mount option.
683 tng-0.0.7 - 13/02/2002 - The driver is now feature complete for read-only!
685 - Cleanup mft.c and it's debug/error output in particular. Fix a minor
686 bug in mapping of extent inodes. Update all the comments to fit all
687 the recent code changes.
688 - Modify vcn_to_lcn() to cope with entirely unmapped run lists.
689 - Cleanups in compress.c, mostly comments and folding help.
690 - Implement attrib.c::map_run_list() as a generic helper.
691 - Make compress.c::ntfs_file_read_compressed_block() use map_run_list()
692 thus making code shorter and enabling attribute list support.
693 - Cleanup incorrect use of [su]64 with %L printf format specifier in
694 all source files. Type casts to [unsigned] long long added to correct
695 the mismatches (important for architectures which have long long not
697 - Merge async io completion handlers for directory indexes and $MFT
698 data into one by setting the index_block_size{_bits} of the ntfs
699 inode for $MFT to the mft_record_size{_bits} of the ntfs_volume.
700 - Cleanup aops.c, update comments.
701 - Make ntfs_file_get_block() use map_run_list() so all files now
702 support attribute lists.
703 - Make ntfs_dir_readpage() almost verbatim copy of
704 block_read_full_page() by using ntfs_file_get_block() with only real
705 difference being the use of our own async io completion handler
706 rather than the default one, thus reducing the amount of code and
707 automatically enabling attribute list support for directory indices.
708 - Fix bug in load_attribute_list() - forgot to call brelse in error
710 - Change parameters to find_attr() and lookup_attr(). We no longer
711 pass in the upcase table and its length. These can be gotten from
712 ctx->ntfs_ino->vol->upcase{_len}. Update all callers.
713 - Cleanups in attrib.c.
714 - Implement merging of run lists, attrib.c::merge_run_lists() and its
715 helpers. (Richard Russon)
716 - Attribute lists part 2, attribute extents and multi part run lists:
717 enable proper support for LCN_RL_NOT_MAPPED and automatic mapping of
718 further run list parts via attrib.c::map_run_list().
719 - Tiny endianness bug fix in decompress_mapping_pairs().
721 tng-0.0.6 - Encrypted directories, bug fixes, cleanups, debugging enhancements.
723 - Enable encrypted directories. (Their index root is marked encrypted
724 to indicate that new files in that directory should be created
726 - Fix bug in NInoBmpNonResident() macro. (Cut and paste error.)
727 - Enable $Extend system directory. Most (if not all) extended system
728 files do not have unnamed data attributes so ntfs_read_inode() had to
729 special case them but that is ok, as the special casing recovery
730 happens inside an error code path so there is zero slow down in the
731 normal fast path. The special casing is done by introducing a new
732 function inode.c::ntfs_is_extended_system_file() which checks if any
733 of the hard links in the inode point to $Extend as being their parent
734 directory and if they do we assume this is an extended system file.
735 - Create a sysctl/proc interface to allow {dis,en}abling of debug output
736 when compiled with -DDEBUG. Default is debug messages to be disabled.
737 To enable them, one writes a non-zero value to /proc/sys/fs/ntfs-debug
738 (if /proc is enabled) or uses sysctl(2) to effect the same (if sysctl
739 interface is enabled). Inspired by old ntfs driver.
740 - Add debug_msgs insmod/kernel boot parameter to set whether debug
741 messages are {dis,en}abled. This is useful to enable debug messages
742 during ntfs initialization and is the only way to activate debugging
743 when the sysctl interface is not enabled.
744 - Cleanup debug output in various places.
745 - Remove all dollar signs ($) from the source (except comments) to
746 enable compilation on architectures whose gcc compiler does not
747 support dollar signs in the names of variables/constants. Attribute
748 types now start with AT_ instead of $ and $I30 is now just I30.
749 - Cleanup ntfs_lookup() and add consistency check of sequence numbers.
750 - Load complete run list for $MFT/$BITMAP during mount and cleanup
751 access functions. This means we now cope with $MFT/$BITMAP being
752 spread accross several mft records.
753 - Disable modification of mft_zone_multiplier on remount. We can always
754 reenable this later on if we really want to, but we will need to make
755 sure we readjust the mft_zone size / layout accordingly.
757 tng-0.0.5 - Modernize for 2.5.x and further in line-ing with Al Viro's comments.
759 - Use sb_set_blocksize() instead of set_blocksize() and verify the
761 - Use sb_bread() instead of bread() throughout.
762 - Add index_vcn_size{_bits} to ntfs_inode structure to store the size
763 of a directory index block vcn. Apply resulting simplifications in
765 - Fix a small bug somewhere (but forgot what it was).
766 - Change ntfs_{debug,error,warning} to enable gcc to do type checking
767 on the printf-format parameter list and fix bugs reported by gcc
768 as a result. (Richard Russon)
769 - Move inode allocation strategy to Al's new stuff but maintain the
770 divorce of ntfs_inode from struct inode. To achieve this we have two
771 separate slab caches, one for big ntfs inodes containing a struct
772 inode and pure ntfs inodes and at the same time fix some faulty
773 error code paths in ntfs_read_inode().
774 - Show mount options in proc (inode.c::ntfs_show_options()).
776 tng-0.0.4 - Big changes, getting in line with Al Viro's comments.
778 - Modified (un)map_mft_record functions to be common for read and write
779 case. To specify which is which, added extra parameter at front of
780 parameter list. Pass either READ or WRITE to this, each has the
782 - General cleanups to allow for easier folding in vi.
783 - attrib.c::decompress_mapping_pairs() now accepts the old run list
784 argument, and invokes attrib.c::merge_run_lists() to merge the old
785 and the new run lists.
786 - Removed attrib.c::find_first_attr().
787 - Implemented loading of attribute list and complete run list for $MFT.
788 This means we now cope with $MFT being spread across several mft
790 - Adapt to 2.5.2-pre9 and the changed create_empty_buffers() syntax.
791 - Adapt major/minor/kdev_t/[bk]devname stuff to new 2.5.x kernels.
792 - Make ntfs_volume be allocated via kmalloc() instead of using a slab
793 cache. There are too little ntfs_volume structures at any one time
794 to justify a private slab cache.
795 - Fix bogus kmap() use in async io completion. Now use kmap_atomic().
796 Use KM_BIO_IRQ on advice from IRC/kernel...
797 - Use ntfs_map_page() in map_mft_record() and create ->readpage method
798 for reading $MFT (ntfs_mft_readpage). In the process create dedicated
799 address space operations (ntfs_mft_aops) for $MFT inode mapping. Also
800 removed the now superfluous exports from the kernel core patch.
801 - Fix a bug where kfree() was used insted of ntfs_free().
802 - Change map_mft_record() to take ntfs_inode as argument instead of
803 vfs inode. Dito for unmap_mft_record(). Adapt all callers.
804 - Add pointer to ntfs_volume to ntfs_inode.
805 - Add mft record number and sequence number to ntfs_inode. Stop using
806 i_ino and i_generation for in-driver purposes.
807 - Implement attrib.c::merge_run_lists(). (Richard Russon)
808 - Remove use of proper inodes by extent inodes. Move i_ino and
809 i_generation to ntfs_inode to do this. Apply simplifications that
810 result and remove iget_no_wait(), etc.
811 - Pass ntfs_inode everywhere in the driver (used to be struct inode).
812 - Add reference counting in ntfs_inode for the ntfs inode itself and
813 for the mapped mft record.
814 - Extend mft record mapping so we can (un)map extent mft records (new
815 functions (un)map_extent_mft_record), and so mappings are reference
816 counted and don't have to happen twice if already mapped - just ref
818 - Add -o iocharset as alias to -o nls for backwards compatibility.
819 - The latest core patch is now tiny. In fact just a single additional
820 export is necessary over the base kernel.
822 tng-0.0.3 - Cleanups, enhancements, bug fixes.
824 - Work on attrib.c::decompress_mapping_pairs() to detect base extents
825 and setup the run list appropriately using knowledge provided by the
826 sizes in the base attribute record.
827 - Balance the get_/put_attr_search_ctx() calls so we don't leak memory
829 - Introduce ntfs_malloc_nofs() and ntfs_free() to allocate/free a single
830 page or use vmalloc depending on the amount of memory requested.
831 - Cleanup error output. The __FUNCTION__ "(): " is now added
832 automatically. Introduced a new header file debug.h to support this
833 and also moved ntfs_debug() function into it.
834 - Make reading of compressed files more intelligent and especially get
835 rid of the vmalloc_nofs() from readpage(). This now uses per CPU
836 buffers (allocated at first mount with cluster size <= 4kiB and
837 deallocated on last umount with cluster size <= 4kiB), and
838 asynchronous io for the compressed data using a list of buffer heads.
839 Er, we use synchronous io as async io only works on whole pages
840 covered by buffers and not on individual buffer heads...
841 - Bug fix for reading compressed files with sparse compression blocks.
843 tng-0.0.2 - Now handles larger/fragmented/compressed volumes/files/dirs.
845 - Fixed handling of directories when cluster size exceeds index block
847 - Hide DOS only name space directory entries from readdir() but allow
848 them in lookup(). This should fix the problem that Linux doesn't
849 support directory hard links, while still allowing access to entries
850 via their short file name. This also has the benefit of mimicking
851 what Windows users are used to, so it is the ideal solution.
852 - Implemented sync_page everywhere so no more hangs in D state when
854 - Stop using bforget() in favour of brelse().
855 - Stop locking buffers unnecessarily.
856 - Implemented compressed files (inode->mapping contains uncompressed
857 data, raw compressed data is currently bread() into a vmalloc()ed
859 - Enable compressed directories. (Their index root is marked compressed
860 to indicate that new files in that directory should be created
862 - Use vsnprintf rather than vsprintf in the ntfs_error and ntfs_warning
863 functions. (Thanks to Will Dyson for pointing this out.)
864 - Moved the ntfs_inode and ntfs_volume (the former ntfs_inode_info and
865 ntfs_sb_info) out of the common inode and super_block structures and
866 started using the generic_ip and generic_sbp pointers instead. This
867 makes ntfs entirely private with respect to the kernel tree.
868 - Detect compiler version and abort with error message if gcc less than
870 - Fix bug in name comparison function in unistr.c.
871 - Implement attribute lists part 1, the infrastructure: search contexts
872 and operations, find_external_attr(), lookup_attr()) and make the
873 code use the infrastructure.
874 - Fix stupid buffer overflow bug that became apparent on larger run
875 list containing attributes.
876 - Fix bugs in readdir() that became apparent on larger directories.
878 The driver is now really useful and survives the test
879 find . -type f -exec md5sum "{}" \;
880 without any error messages on a over 1GiB sized partition with >16k
881 files on it, including compressed files and directories and many files
882 and directories with attribute lists.
884 tng-0.0.1 - The first useful version.
886 - Added ntfs_lookup().
887 - Added default upcase generation and handling.
888 - Added compile options to be shown on module init.
889 - Many bug fixes that were "hidden" before.
890 - Update to latest kernel.
891 - Added ntfs_readdir().
892 - Added file operations for mmap(), read(), open() and llseek(). We just
893 use the generic ones. The whole point of going through implementing
894 readpage() methods and where possible get_block() call backs is that
895 this allows us to make use of the generic high level methods provided
898 The driver is now actually useful! Yey. (-: It undoubtedly has got bugs
899 though and it doesn't implement accesssing compressed files yet. Also,
900 accessing files with attribute list attributes is not implemented yet
901 either. But for small or simple file systems it should work and allow
902 you to list directories, use stat on directory entries and the file
903 system, open, read, mmap and llseek around in files. A big mile stone
906 tng-0.0.0 - Initial version tag.
908 Initial driver implementation. The driver can mount and umount simple
909 NTFS file systems (i.e. ones without attribute lists in the system
910 files). If the mount fails there might be problems in the error handling
911 code paths, so be warned. Otherwise it seems to be loading the system
912 files nicely and the mft record read mapping/unmapping seems to be
913 working nicely, too. Proof of inode metadata in the page cache and non-
914 resident file unnamed stream data in the page cache concepts is thus