Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1 | ToDo/Notes: |
| 2 | - Find and fix bugs. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 3 | - In between ntfs_prepare/commit_write, need exclusion between |
Anton Altaparmakov | 66129f8 | 2004-11-11 12:34:00 +0000 | [diff] [blame] | 4 | simultaneous file extensions. This is given to us by holding i_sem |
| 5 | on the inode. The only places in the kernel when a file is resized |
| 6 | are prepare/commit write and truncate for both of which i_sem is |
| 7 | held. Just have to be careful in readpage/writepage and all other |
| 8 | helpers not running under i_sem that we play nice... |
| 9 | Also need to be careful with initialized_size extention in |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 10 | ntfs_prepare_write. Basically, just be _very_ careful in this code... |
Anton Altaparmakov | 66129f8 | 2004-11-11 12:34:00 +0000 | [diff] [blame] | 11 | UPDATE: The only things that need to be checked are read/writepage |
| 12 | which do not hold i_sem. Note writepage cannot change i_size but it |
| 13 | needs to cope with a concurrent i_size change, just like readpage. |
| 14 | Also both need to cope with concurrent changes to the other sizes, |
| 15 | i.e. initialized/allocated/compressed size, as well. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 16 | - Implement mft.c::sync_mft_mirror_umount(). We currently will just |
| 17 | leave the volume dirty on umount if the final iput(vol->mft_ino) |
| 18 | causes a write of any mirrored mft records due to the mft mirror |
| 19 | inode having been discarded already. Whether this can actually ever |
| 20 | happen is unclear however so it is worth waiting until someone hits |
| 21 | the problem. |
| 22 | - Enable the code for setting the NT4 compatibility flag when we start |
| 23 | making NTFS 1.2 specific modifications. |
| 24 | |
Anton Altaparmakov | fd9d636 | 2005-10-04 13:44:48 +0100 | [diff] [blame] | 25 | 2.1.25-WIP |
| 26 | |
Anton Altaparmakov | 511bea5 | 2005-10-04 14:24:21 +0100 | [diff] [blame] | 27 | - Change ntfs_map_runlist_nolock(), ntfs_attr_find_vcn_nolock() and |
| 28 | {__,}ntfs_cluster_free() to also take an optional attribute search |
| 29 | context as argument. This allows calling these functions with the |
| 30 | mft record mapped. Update all callers. |
| 31 | - Fix potential deadlock in ntfs_mft_data_extend_allocation_nolock() |
| 32 | error handling by passing in the active search context when calling |
| 33 | ntfs_cluster_free(). |
Anton Altaparmakov | fc0fa7d | 2005-10-04 14:36:56 +0100 | [diff] [blame] | 34 | - Change ntfs_cluster_alloc() to take an extra boolean parameter |
| 35 | specifying whether the cluster are being allocated to extend an |
| 36 | attribute or to fill a hole. |
| 37 | - Change ntfs_attr_make_non_resident() to call ntfs_cluster_alloc() |
| 38 | with @is_extension set to TRUE and remove the runlist terminator |
| 39 | fixup code as this is now done by ntfs_cluster_alloc(). |
Anton Altaparmakov | 8925d4f | 2005-10-04 14:48:20 +0100 | [diff] [blame] | 40 | - Change ntfs_attr_make_non_resident to take the attribute value size |
| 41 | as an extra parameter. This is needed since we need to know the size |
| 42 | before we can map the mft record and our callers always know it. The |
| 43 | reason we cannot simply read the size from the vfs inode i_size is |
| 44 | that this is not necessarily uptodate. This happens when |
| 45 | ntfs_attr_make_non_resident() is called in the ->truncate call path. |
Anton Altaparmakov | 2a6fc4e1 | 2005-10-04 14:57:15 +0100 | [diff] [blame] | 46 | - Fix ntfs_attr_make_non_resident() to update the vfs inode i_blocks |
| 47 | which is zero for a resident attribute but should no longer be zero |
| 48 | once the attribute is non-resident as it then has real clusters |
| 49 | allocated. |
Anton Altaparmakov | 2d86829 | 2005-10-04 15:18:56 +0100 | [diff] [blame^] | 50 | - Add fs/ntfs/attrib.[hc]::ntfs_attr_extend_allocation(), a function to |
| 51 | extend the allocation of an attributes. Optionally, the data size, |
| 52 | but not the initialized size can be extended, too. |
Anton Altaparmakov | fd9d636 | 2005-10-04 13:44:48 +0100 | [diff] [blame] | 53 | |
Anton Altaparmakov | 7d333d6 | 2005-09-08 23:01:16 +0100 | [diff] [blame] | 54 | 2.1.24 - Lots of bug fixes and support more clean journal states. |
Anton Altaparmakov | e7a1033 | 2005-09-08 16:12:28 +0100 | [diff] [blame] | 55 | |
| 56 | - Support journals ($LogFile) which have been modified by chkdsk. This |
| 57 | means users can boot into Windows after we marked the volume dirty. |
| 58 | The Windows boot will run chkdsk and then reboot. The user can then |
| 59 | immediately boot into Linux rather than having to do a full Windows |
| 60 | boot first before rebooting into Linux and we will recognize such a |
Anton Altaparmakov | 5a8c0cc | 2005-09-26 10:48:54 +0100 | [diff] [blame] | 61 | journal and empty it as it is clean by definition. Note, this only |
| 62 | works if chkdsk left the journal in an obviously clean state. |
Anton Altaparmakov | e7a1033 | 2005-09-08 16:12:28 +0100 | [diff] [blame] | 63 | - Support journals ($LogFile) with only one restart page as well as |
| 64 | journals with two different restart pages. We sanity check both and |
| 65 | either use the only sane one or the more recent one of the two in the |
| 66 | case that both are valid. |
Anton Altaparmakov | 06d0e3c | 2005-09-08 16:28:25 +0100 | [diff] [blame] | 67 | - Add fs/ntfs/malloc.h::ntfs_malloc_nofs_nofail() which is analogous to |
| 68 | ntfs_malloc_nofs() but it performs allocations with __GFP_NOFAIL and |
| 69 | hence cannot fail. |
Anton Altaparmakov | 9529d46 | 2005-09-08 16:33:12 +0100 | [diff] [blame] | 70 | - Use ntfs_malloc_nofs_nofail() in the two critical regions in |
| 71 | fs/ntfs/runlist.c::ntfs_runlists_merge(). This means we no longer |
| 72 | need to panic() if the allocation fails as it now cannot fail. |
Anton Altaparmakov | 84d6ebe | 2005-09-08 16:46:55 +0100 | [diff] [blame] | 73 | - Fix two nasty runlist merging bugs that had gone unnoticed so far. |
| 74 | Thanks to Stefano Picerno for the bug report. |
Anton Altaparmakov | 8bb7352 | 2005-09-08 16:48:28 +0100 | [diff] [blame] | 75 | - Remove two bogus BUG_ON()s from fs/ntfs/mft.c. |
Anton Altaparmakov | 2b0ada2 | 2005-09-08 16:52:31 +0100 | [diff] [blame] | 76 | - Fix handling of valid but empty mapping pairs array in |
| 77 | fs/ntfs/runlist.c::ntfs_mapping_pairs_decompress(). |
Anton Altaparmakov | f94ad38 | 2005-09-08 17:04:11 +0100 | [diff] [blame] | 78 | - Report unrepresentable inodes during ntfs_readdir() as KERN_WARNING |
| 79 | messages and include the inode number. Thanks to Yura Pakhuchiy for |
| 80 | pointing this out. |
Anton Altaparmakov | 3ffc5a4 | 2005-09-08 20:23:06 +0100 | [diff] [blame] | 81 | - Change ntfs_rl_truncate_nolock() to throw away the runlist if the new |
| 82 | length is zero. |
Anton Altaparmakov | 6e48321 | 2005-09-08 20:26:34 +0100 | [diff] [blame] | 83 | - Add runlist.[hc]::ntfs_rl_punch_nolock() which punches a caller |
| 84 | specified hole into a runlist. |
Anton Altaparmakov | 8e08cea | 2005-09-08 20:29:50 +0100 | [diff] [blame] | 85 | - Fix a bug in fs/ntfs/index.c::ntfs_index_lookup(). When the returned |
| 86 | index entry is in the index root, we forgot to set the @ir pointer in |
| 87 | the index context. Thanks to Yura Pakhuchiy for finding this bug. |
Anton Altaparmakov | f25dfb5 | 2005-09-08 20:35:33 +0100 | [diff] [blame] | 88 | - Remove bogus setting of PageError in ntfs_read_compressed_block(). |
Anton Altaparmakov | 0aaccea | 2005-09-08 20:40:32 +0100 | [diff] [blame] | 89 | - Add fs/ntfs/attrib.[hc]::ntfs_resident_attr_value_resize(). |
Anton Altaparmakov | 2983d1b | 2005-09-08 20:56:09 +0100 | [diff] [blame] | 90 | - Fix a bug in ntfs_map_runlist_nolock() where we forgot to protect |
| 91 | access to the allocated size in the ntfs inode with the size lock. |
| 92 | - Fix ntfs_attr_vcn_to_lcn_nolock() and ntfs_attr_find_vcn_nolock() to |
| 93 | return LCN_ENOENT when there is no runlist and the allocated size is |
| 94 | zero. |
| 95 | - Fix load_attribute_list() to handle the case of a NULL runlist. |
Anton Altaparmakov | 807c453 | 2005-09-08 21:01:17 +0100 | [diff] [blame] | 96 | - Fix handling of sparse attributes in ntfs_attr_make_non_resident(). |
| 97 | - Add BUG() checks to ntfs_attr_make_non_resident() and ntfs_attr_set() |
| 98 | to ensure that these functions are never called for compressed or |
| 99 | encrypted attributes. |
Anton Altaparmakov | bbf1813 | 2005-09-08 21:09:06 +0100 | [diff] [blame] | 100 | - Fix cluster (de)allocators to work when the runlist is NULL and more |
| 101 | importantly to take a locked runlist rather than them locking it |
| 102 | which leads to lock reversal. |
Anton Altaparmakov | 1c7d469 | 2005-09-08 21:15:09 +0100 | [diff] [blame] | 103 | - Truncate {a,c,m}time to the ntfs supported time granularity when |
| 104 | updating the times in the inode in ntfs_setattr(). |
Anton Altaparmakov | 67bb103 | 2005-09-08 21:19:45 +0100 | [diff] [blame] | 105 | - Fixup handling of sparse, compressed, and encrypted attributes in |
Anton Altaparmakov | 311120e | 2005-09-08 22:04:20 +0100 | [diff] [blame] | 106 | fs/ntfs/inode.c::ntfs_read_locked_{,attr_,index_}inode(), |
| 107 | fs/ntfs/aops.c::ntfs_{read,write}page(). |
Anton Altaparmakov | 8dcdeba | 2005-09-08 21:25:48 +0100 | [diff] [blame] | 108 | - Make ntfs_write_block() not instantiate sparse blocks if they contain |
| 109 | only zeroes. |
Anton Altaparmakov | 54b02eb | 2005-09-08 21:43:47 +0100 | [diff] [blame] | 110 | - Optimize fs/ntfs/aops.c::ntfs_write_block() by extending the page |
| 111 | lock protection over the buffer submission for i/o which allows the |
| 112 | removal of the get_bh()/put_bh() pairs for each buffer. |
Anton Altaparmakov | 8273d5d | 2005-09-08 22:00:33 +0100 | [diff] [blame] | 113 | - Fix fs/ntfs/aops.c::ntfs_{read,write}_block() to handle the case |
| 114 | where a concurrent truncate has truncated the runlist under our feet. |
Anton Altaparmakov | a01ac53 | 2005-09-08 22:08:11 +0100 | [diff] [blame] | 115 | - Fix page_has_buffers()/page_buffers() handling in fs/ntfs/aops.c. |
Anton Altaparmakov | e604635 | 2005-09-08 22:13:02 +0100 | [diff] [blame] | 116 | - In fs/ntfs/aops.c::ntfs_end_buffer_async_read(), use a bit spin lock |
| 117 | in the first buffer head instead of a driver global spin lock to |
| 118 | improve scalability. |
Anton Altaparmakov | 7d333d6 | 2005-09-08 23:01:16 +0100 | [diff] [blame] | 119 | - Minor fix to error handling and error message display in |
Anton Altaparmakov | 5d46770 | 2005-09-12 14:33:47 +0100 | [diff] [blame] | 120 | fs/ntfs/aops.c::ntfs_prepare_nonresident_write(). |
| 121 | - Change the mount options {u,f,d}mask to always parse the number as |
| 122 | an octal number to conform to how chmod(1) works, too. Thanks to |
| 123 | Giuseppe Bilotta and Horst von Brand for pointing out the errors of |
| 124 | my ways. |
Anton Altaparmakov | 5c9f6de | 2005-09-19 09:33:40 +0100 | [diff] [blame] | 125 | - Fix various bugs in the runlist merging code. (Based on libntfs |
| 126 | changes by Richard Russon.) |
Anton Altaparmakov | 5a8c0cc | 2005-09-26 10:48:54 +0100 | [diff] [blame] | 127 | - Fix sparse warnings that have crept in over time. |
| 128 | - Change ntfs_cluster_free() to require a write locked runlist on entry |
| 129 | since we otherwise get into a lock reversal deadlock if a read locked |
| 130 | runlist is passed in. In the process also change it to take an ntfs |
| 131 | inode instead of a vfs inode as parameter. |
| 132 | - Fix the definition of the CHKD ntfs record magic. It had an off by |
| 133 | two error causing it to be CHKB instead of CHKD. |
Anton Altaparmakov | 18efefa | 2005-10-04 13:06:00 +0100 | [diff] [blame] | 134 | - Fix a stupid bug in __ntfs_bitmap_set_bits_in_run() which caused the |
| 135 | count to become negative and hence we had a wild memset() scribbling |
| 136 | all over the system's ram. |
Anton Altaparmakov | e7a1033 | 2005-09-08 16:12:28 +0100 | [diff] [blame] | 137 | |
Anton Altaparmakov | af859a4 | 2005-06-25 21:07:27 +0100 | [diff] [blame] | 138 | 2.1.23 - Implement extension of resident files and make writing safe as well as |
| 139 | many bug fixes, cleanups, and enhancements... |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 140 | |
| 141 | - Add printk rate limiting for ntfs_warning() and ntfs_error() when |
| 142 | compiled without debug. This avoids a possible denial of service |
| 143 | attack. Thanks to Carl-Daniel Hailfinger from SuSE for pointing this |
| 144 | out. |
Randy Dunlap | 8907547 | 2005-03-03 11:19:53 +0000 | [diff] [blame] | 145 | - Fix compilation warnings on ia64. (Randy Dunlap) |
Anton Altaparmakov | af859a4 | 2005-06-25 21:07:27 +0100 | [diff] [blame] | 146 | - Use i_size_{read,write}() instead of reading i_size by hand and cache |
| 147 | the value where apropriate. |
Anton Altaparmakov | 3676367 | 2004-11-18 13:46:45 +0000 | [diff] [blame] | 148 | - Add size_lock to the ntfs_inode structure. This is an rw spinlock |
| 149 | and it locks against access to the inode sizes. Note, ->size_lock |
| 150 | is also accessed from irq context so you must use the _irqsave and |
Anton Altaparmakov | af859a4 | 2005-06-25 21:07:27 +0100 | [diff] [blame] | 151 | _irqrestore lock and unlock functions, respectively. Protect all |
| 152 | accesses to allocated_size, initialized_size, and compressed_size. |
Anton Altaparmakov | 218357f | 2004-11-18 20:34:59 +0000 | [diff] [blame] | 153 | - Minor optimization to fs/ntfs/super.c::ntfs_statfs() and its helpers. |
Anton Altaparmakov | 946929d | 2005-01-13 15:26:29 +0000 | [diff] [blame] | 154 | - Implement extension of resident files in the regular file write code |
| 155 | paths (fs/ntfs/aops.c::ntfs_{prepare,commit}_write()). At present |
| 156 | this only works until the data attribute becomes too big for the mft |
| 157 | record after which we abort the write returning -EOPNOTSUPP from |
| 158 | ntfs_prepare_write(). |
Anton Altaparmakov | c002f42 | 2005-02-03 12:02:56 +0000 | [diff] [blame] | 159 | - Add disable_sparse mount option together with a per volume sparse |
| 160 | enable bit which is set appropriately and a per inode sparse disable |
| 161 | bit which is preset on some system file inodes as appropriate. |
| 162 | - Enforce that sparse support is disabled on NTFS volumes pre 3.0. |
Anton Altaparmakov | 1a0df15 | 2005-02-03 12:04:36 +0000 | [diff] [blame] | 163 | - Fix a bug in fs/ntfs/runlist.c::ntfs_mapping_pairs_decompress() in |
| 164 | the creation of the unmapped runlist element for the base attribute |
| 165 | extent. |
Anton Altaparmakov | b6ad6c5 | 2005-02-15 10:08:43 +0000 | [diff] [blame] | 166 | - Split ntfs_map_runlist() into ntfs_map_runlist() and a non-locking |
| 167 | helper ntfs_map_runlist_nolock() which is used by ntfs_map_runlist(). |
| 168 | This allows us to map runlist fragments with the runlist lock already |
| 169 | held without having to drop and reacquire it around the call. Adapt |
| 170 | all callers. |
| 171 | - Change ntfs_find_vcn() to ntfs_find_vcn_nolock() which takes a locked |
| 172 | runlist. This allows us to find runlist elements with the runlist |
| 173 | lock already held without having to drop and reacquire it around the |
| 174 | call. Adapt all callers. |
Anton Altaparmakov | 37e4c13 | 2005-02-18 10:03:13 +0000 | [diff] [blame] | 175 | - Change time to u64 in time.h::ntfs2utc() as it otherwise generates a |
Anton Altaparmakov | d8ec785 | 2005-02-18 09:23:39 +0000 | [diff] [blame] | 176 | warning in the do_div() call on sparc32. Thanks to Meelis Roos for |
| 177 | the report and analysis of the warning. |
Anton Altaparmakov | 37e4c13 | 2005-02-18 10:03:13 +0000 | [diff] [blame] | 178 | - Fix a nasty runlist merge bug when merging two holes. |
Anton Altaparmakov | 5ae9fcf | 2005-03-02 17:03:24 +0000 | [diff] [blame] | 179 | - Set the ntfs_inode->allocated_size to the real allocated size in the |
| 180 | mft record for resident attributes (fs/ntfs/inode.c). |
| 181 | - Small readability cleanup to use "a" instead of "ctx->attr" |
| 182 | everywhere (fs/ntfs/inode.c). |
Anton Altaparmakov | 4138268 | 2005-03-03 13:44:15 +0000 | [diff] [blame] | 183 | - Make fs/ntfs/namei.c::ntfs_get_{parent,dentry} static and move the |
| 184 | definition of ntfs_export_ops from fs/ntfs/super.c to namei.c. Also, |
| 185 | declare ntfs_export_ops in fs/ntfs/ntfs.h. |
Anton Altaparmakov | 9451f85 | 2005-03-03 14:43:43 +0000 | [diff] [blame] | 186 | - Correct sparse file handling. The compressed values need to be |
| 187 | checked and set in the ntfs inode as done for compressed files and |
| 188 | the compressed size needs to be used for vfs inode->i_blocks instead |
| 189 | of the allocated size, again, as done for compressed files. |
Anton Altaparmakov | 271849a | 2005-03-07 21:36:18 +0000 | [diff] [blame] | 190 | - Add AT_EA in addition to AT_DATA to whitelist for being allowed to be |
| 191 | non-resident in fs/ntfs/attrib.c::ntfs_attr_can_be_non_resident(). |
| 192 | - Add fs/ntfs/attrib.c::ntfs_attr_vcn_to_lcn_nolock() used by the new |
| 193 | write code. |
Anton Altaparmakov | c0c1cc0 | 2005-03-07 21:43:38 +0000 | [diff] [blame] | 194 | - Fix bug in fs/ntfs/attrib.c::ntfs_find_vcn_nolock() where after |
| 195 | dropping the read lock and taking the write lock we were not checking |
| 196 | whether someone else did not already do the work we wanted to do. |
| 197 | - Rename fs/ntfs/attrib.c::ntfs_find_vcn_nolock() to |
| 198 | ntfs_attr_find_vcn_nolock() and update all callers. |
Anton Altaparmakov | 2bfb4ff | 2005-03-09 15:15:06 +0000 | [diff] [blame] | 199 | - Add fs/ntfs/attrib.[hc]::ntfs_attr_make_non_resident(). |
Anton Altaparmakov | 43b01fd | 2005-03-09 15:18:43 +0000 | [diff] [blame] | 200 | - Fix sign of various error return values to be negative in |
| 201 | fs/ntfs/lcnalloc.c. |
Anton Altaparmakov | 905685f | 2005-03-10 11:06:19 +0000 | [diff] [blame] | 202 | - Modify ->readpage and ->writepage (fs/ntfs/aops.c) so they detect and |
| 203 | handle the case where an attribute is converted from resident to |
| 204 | non-resident by a concurrent file write. |
Jesper Juhl | 251c842 | 2005-04-04 14:59:56 +0100 | [diff] [blame] | 205 | - Remove checks for NULL before calling kfree() since kfree() does the |
| 206 | checking itself. (Jesper Juhl) |
Anton Altaparmakov | b0d2374 | 2005-04-04 16:20:14 +0100 | [diff] [blame] | 207 | - Some utilities modify the boot sector but do not update the checksum. |
| 208 | Thus, relax the checking in fs/ntfs/super.c::is_boot_sector_ntfs() to |
| 209 | only emit a warning when the checksum is incorrect rather than |
| 210 | refusing the mount. Thanks to Bernd Casimir for pointing this |
| 211 | problem out. |
Anton Altaparmakov | bb3cf33 | 2005-04-06 13:34:31 +0100 | [diff] [blame] | 212 | - Update attribute definition handling. |
Anton Altaparmakov | 7fafb8b | 2005-04-06 16:09:21 +0100 | [diff] [blame] | 213 | - Add NTFS_MAX_CLUSTER_SIZE and NTFS_MAX_PAGES_PER_CLUSTER constants. |
| 214 | - Use NTFS_MAX_CLUSTER_SIZE in super.c instead of hard coding 0x10000. |
Anton Altaparmakov | d53ee32 | 2005-04-06 16:11:20 +0100 | [diff] [blame] | 215 | - Use MAX_BUF_PER_PAGE instead of variable sized array allocation for |
| 216 | better code generation and one less sparse warning in fs/ntfs/aops.c. |
Pekka Enberg | 2fb21db | 2005-05-25 21:15:34 +0300 | [diff] [blame] | 217 | - Remove spurious void pointer casts from fs/ntfs/. (Pekka Enberg) |
Anton Altaparmakov | 442d207 | 2005-05-27 16:42:56 +0100 | [diff] [blame] | 218 | - Use C99 style structure initialization after memory allocation where |
| 219 | possible (fs/ntfs/{attrib.c,index.c,super.c}). Thanks to Al Viro and |
| 220 | Pekka Enberg. |
Anton Altaparmakov | 3f2faef | 2005-06-25 15:28:56 +0100 | [diff] [blame] | 221 | - Stamp the transaction log ($UsnJrnl), aka user space journal, if it |
| 222 | is active on the volume and we are mounting read-write or remounting |
| 223 | from read-only to read-write. |
Anton Altaparmakov | 9f993fe | 2005-06-25 16:15:36 +0100 | [diff] [blame] | 224 | - Fix a bug in address space operations error recovery code paths where |
| 225 | if the runlist was not mapped at all and a mapping error occured we |
| 226 | would leave the runlist locked on exit to the function so that the |
| 227 | next access to the same file would try to take the lock and deadlock. |
Anton Altaparmakov | ca8fd7a | 2005-06-25 16:31:27 +0100 | [diff] [blame] | 228 | - Detect the case when Windows has been suspended to disk on the volume |
| 229 | to be mounted and if this is the case do not allow (re)mounting |
| 230 | read-write. This is done by parsing hiberfil.sys if present. |
Anton Altaparmakov | 3bd1f4a | 2005-06-25 16:51:58 +0100 | [diff] [blame] | 231 | - Fix several occurences of a bug where we would perform 'var & ~const' |
| 232 | with a 64-bit variable and a int, i.e. 32-bit, constant. This causes |
| 233 | the higher order 32-bits of the 64-bit variable to be zeroed. To fix |
| 234 | this cast the 'const' to the same 64-bit type as 'var'. |
Anton Altaparmakov | 1d58b27 | 2005-06-25 17:04:55 +0100 | [diff] [blame] | 235 | - Change the runlist terminator of the newly allocated cluster(s) to |
| 236 | LCN_ENOENT in ntfs_attr_make_non_resident(). Otherwise the runlist |
| 237 | code gets confused. |
Anton Altaparmakov | fa3be92 | 2005-06-25 17:15:36 +0100 | [diff] [blame] | 238 | - Add an extra parameter @last_vcn to ntfs_get_size_for_mapping_pairs() |
| 239 | and ntfs_mapping_pairs_build() to allow the runlist encoding to be |
| 240 | partial which is desirable when filling holes in sparse attributes. |
| 241 | Update all callers. |
Anton Altaparmakov | 4757d7df | 2005-06-25 17:24:08 +0100 | [diff] [blame] | 242 | - Change ntfs_map_runlist_nolock() to only decompress the mapping pairs |
| 243 | if the requested vcn is inside it. Otherwise we get into problems |
| 244 | when we try to map an out of bounds vcn because we then try to map |
| 245 | the already mapped runlist fragment which causes |
| 246 | ntfs_mapping_pairs_decompress() to fail and return error. Update |
| 247 | ntfs_attr_find_vcn_nolock() accordingly. |
Anton Altaparmakov | ba6d237 | 2005-06-26 22:12:02 +0100 | [diff] [blame] | 248 | - Fix a nasty deadlock that appeared in recent kernels. |
| 249 | The situation: VFS inode X on a mounted ntfs volume is dirty. For |
| 250 | same inode X, the ntfs_inode is dirty and thus corresponding on-disk |
| 251 | inode, i.e. mft record, which is in a dirty PAGE_CACHE_PAGE belonging |
| 252 | to the table of inodes, i.e. $MFT, inode 0. |
| 253 | What happens: |
| 254 | Process 1: sys_sync()/umount()/whatever... calls |
| 255 | __sync_single_inode() for $MFT -> do_writepages() -> write_page for |
| 256 | the dirty page containing the on-disk inode X, the page is now locked |
| 257 | -> ntfs_write_mst_block() which clears PageUptodate() on the page to |
| 258 | prevent anyone else getting hold of it whilst it does the write out. |
| 259 | This is necessary as the on-disk inode needs "fixups" applied before |
| 260 | the write to disk which are removed again after the write and |
| 261 | PageUptodate is then set again. It then analyses the page looking |
| 262 | for dirty on-disk inodes and when it finds one it calls |
| 263 | ntfs_may_write_mft_record() to see if it is safe to write this |
| 264 | on-disk inode. This then calls ilookup5() to check if the |
| 265 | corresponding VFS inode is in icache(). This in turn calls ifind() |
| 266 | which waits on the inode lock via wait_on_inode whilst holding the |
| 267 | global inode_lock. |
| 268 | Process 2: pdflush results in a call to __sync_single_inode for the |
| 269 | same VFS inode X on the ntfs volume. This locks the inode (I_LOCK) |
| 270 | then calls write-inode -> ntfs_write_inode -> map_mft_record() -> |
| 271 | read_cache_page() for the page (in page cache of table of inodes |
| 272 | $MFT, inode 0) containing the on-disk inode. This page has |
| 273 | PageUptodate() clear because of Process 1 (see above) so |
| 274 | read_cache_page() blocks when it tries to take the page lock for the |
| 275 | page so it can call ntfs_read_page(). |
| 276 | Thus Process 1 is holding the page lock on the page containing the |
| 277 | on-disk inode X and it is waiting on the inode X to be unlocked in |
| 278 | ifind() so it can write the page out and then unlock the page. |
| 279 | And Process 2 is holding the inode lock on inode X and is waiting for |
| 280 | the page to be unlocked so it can call ntfs_readpage() or discover |
| 281 | that Process 1 set PageUptodate() again and use the page. |
| 282 | Thus we have a deadlock due to ifind() waiting on the inode lock. |
| 283 | The solution: The fix is to use the newly introduced |
| 284 | ilookup5_nowait() which does not wait on the inode's lock and hence |
| 285 | avoids the deadlock. This is safe as we do not care about the VFS |
| 286 | inode and only use the fact that it is in the VFS inode cache and the |
| 287 | fact that the vfs and ntfs inodes are one struct in memory to find |
| 288 | the ntfs inode in memory if present. Also, the ntfs inode has its |
| 289 | own locking so it does not matter if the vfs inode is locked. |
Anton Altaparmakov | e74589a | 2005-08-16 16:38:28 +0100 | [diff] [blame] | 290 | - Fix bug in mft record writing where we forgot to set the device in |
Anton Altaparmakov | 481d037 | 2005-08-16 19:42:56 +0100 | [diff] [blame] | 291 | the buffers when mapping them after the VM had discarded them. |
Anton Altaparmakov | e74589a | 2005-08-16 16:38:28 +0100 | [diff] [blame] | 292 | Thanks to Martin MOKREJÅ for the bug report. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 293 | |
| 294 | 2.1.22 - Many bug and race fixes and error handling improvements. |
| 295 | |
| 296 | - Improve error handling in fs/ntfs/inode.c::ntfs_truncate(). |
| 297 | - Change fs/ntfs/inode.c::ntfs_truncate() to return an error code |
| 298 | instead of void and provide a helper ntfs_truncate_vfs() for the |
| 299 | vfs ->truncate method. |
| 300 | - Add a new ntfs inode flag NInoTruncateFailed() and modify |
| 301 | fs/ntfs/inode.c::ntfs_truncate() to set and clear it appropriately. |
| 302 | - Fix min_size and max_size definitions in ATTR_DEF structure in |
| 303 | fs/ntfs/layout.h to be signed. |
| 304 | - Add attribute definition handling helpers to fs/ntfs/attrib.[hc]: |
| 305 | ntfs_attr_size_bounds_check(), ntfs_attr_can_be_non_resident(), and |
| 306 | ntfs_attr_can_be_resident(), which in turn use the new private helper |
| 307 | ntfs_attr_find_in_attrdef(). |
| 308 | - In fs/ntfs/aops.c::mark_ntfs_record_dirty(), take the |
| 309 | mapping->private_lock around the dirtying of the buffer heads |
| 310 | analagous to the way it is done in __set_page_dirty_buffers(). |
| 311 | - Ensure the mft record size does not exceed the PAGE_CACHE_SIZE at |
| 312 | mount time as this cannot work with the current implementation. |
| 313 | - Check for location of attribute name and improve error handling in |
| 314 | general in fs/ntfs/inode.c::ntfs_read_locked_inode() and friends. |
| 315 | - In fs/ntfs/aops.c::ntfs_writepage(), if the page is fully outside |
| 316 | i_size, i.e. race with truncate, invalidate the buffers on the page |
| 317 | so that they become freeable and hence the page does not leak. |
| 318 | - Remove unused function fs/ntfs/runlist.c::ntfs_rl_merge(). (Adrian |
| 319 | Bunk) |
| 320 | - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_find() that resulted in |
| 321 | a NULL pointer dereference in the error code path when a corrupt |
| 322 | attribute was found. (Thanks to Domen Puncer for the bug report.) |
| 323 | - Add MODULE_VERSION() to fs/ntfs/super.c. |
| 324 | - Make several functions and variables static. (Adrian Bunk) |
| 325 | - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() so it allocates |
| 326 | buffers for the page if they are not present and then marks the |
| 327 | buffers belonging to the ntfs record dirty. This causes the buffers |
| 328 | to become busy and hence they are safe from removal until the page |
| 329 | has been written out. |
| 330 | - Fix stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find() in the |
| 331 | error handling code path that resulted in a BUG() due to trying to |
| 332 | unmap an extent mft record when the mapping of it had failed and it |
| 333 | thus was not mapped. (Thanks to Ken MacFerrin for the bug report.) |
| 334 | - Drop the runlist lock after the vcn has been read in |
| 335 | fs/ntfs/lcnalloc.c::__ntfs_cluster_free(). |
| 336 | - Rewrite handling of multi sector transfer errors. We now do not set |
| 337 | PageError() when such errors are detected in the async i/o handler |
| 338 | fs/ntfs/aops.c::ntfs_end_buffer_async_read(). All users of mst |
| 339 | protected attributes now check the magic of each ntfs record as they |
| 340 | use it and act appropriately. This has the effect of making errors |
| 341 | granular per ntfs record rather than per page which solves the case |
| 342 | where we cannot access any of the ntfs records in a page when a |
| 343 | single one of them had an mst error. (Thanks to Ken MacFerrin for |
| 344 | the bug report.) |
| 345 | - Fix error handling in fs/ntfs/quota.c::ntfs_mark_quotas_out_of_date() |
| 346 | where we failed to release i_sem on the $Quota/$Q attribute inode. |
| 347 | - Fix bug in handling of bad inodes in fs/ntfs/namei.c::ntfs_lookup(). |
| 348 | - Add mapping of unmapped buffers to all remaining code paths, i.e. |
| 349 | fs/ntfs/aops.c::ntfs_write_mst_block(), mft.c::ntfs_sync_mft_mirror(), |
| 350 | and write_mft_record_nolock(). From now on we require that the |
| 351 | complete runlist for the mft mirror is always mapped into memory. |
| 352 | - Add creation of buffers to fs/ntfs/mft.c::ntfs_sync_mft_mirror(). |
| 353 | - Improve error handling in fs/ntfs/aops.c::ntfs_{read,write}_block(). |
| 354 | - Cleanup fs/ntfs/aops.c::ntfs_{read,write}page() since we know that a |
| 355 | resident attribute will be smaller than a page which makes the code |
| 356 | simpler. Also make the code more tolerant to concurrent ->truncate. |
| 357 | |
| 358 | 2.1.21 - Fix some races and bugs, rewrite mft write code, add mft allocator. |
| 359 | |
| 360 | - Implement extent mft record deallocation |
| 361 | fs/ntfs/mft.c::ntfs_extent_mft_record_free(). |
| 362 | - Splitt runlist related functions off from attrib.[hc] to runlist.[hc]. |
| 363 | - Add vol->mft_data_pos and initialize it at mount time. |
| 364 | - Rename init_runlist() to ntfs_init_runlist(), ntfs_vcn_to_lcn() to |
| 365 | ntfs_rl_vcn_to_lcn(), decompress_mapping_pairs() to |
| 366 | ntfs_mapping_pairs_decompress(), ntfs_merge_runlists() to |
| 367 | ntfs_runlists_merge() and adapt all callers. |
| 368 | - Add fs/ntfs/runlist.[hc]::ntfs_get_nr_significant_bytes(), |
| 369 | ntfs_get_size_for_mapping_pairs(), ntfs_write_significant_bytes(), |
| 370 | and ntfs_mapping_pairs_build(), adapted from libntfs. |
| 371 | - Make fs/ntfs/lcnalloc.c::ntfs_cluster_free_from_rl_nolock() not |
| 372 | static and add a declaration for it to lcnalloc.h. |
| 373 | - Add fs/ntfs/lcnalloc.h::ntfs_cluster_free_from_rl() which is a static |
| 374 | inline wrapper for ntfs_cluster_free_from_rl_nolock() which takes the |
| 375 | cluster bitmap lock for the duration of the call. |
| 376 | - Add fs/ntfs/attrib.[hc]::ntfs_attr_record_resize(). |
| 377 | - Implement the equivalent of memset() for an ntfs attribute in |
| 378 | fs/ntfs/attrib.[hc]::ntfs_attr_set() and switch |
| 379 | fs/ntfs/logfile.c::ntfs_empty_logfile() to using it. |
| 380 | - Remove unnecessary casts from LCN_* constants. |
| 381 | - Implement fs/ntfs/runlist.c::ntfs_rl_truncate_nolock(). |
| 382 | - Add MFT_RECORD_OLD as a copy of MFT_RECORD in fs/ntfs/layout.h and |
| 383 | change MFT_RECORD to contain the NTFS 3.1+ specific fields. |
| 384 | - Add a helper function fs/ntfs/aops.c::mark_ntfs_record_dirty() which |
| 385 | marks all buffers belonging to an ntfs record dirty, followed by |
| 386 | marking the page the ntfs record is in dirty and also marking the vfs |
| 387 | inode containing the ntfs record dirty (I_DIRTY_PAGES). |
| 388 | - Switch fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to using the |
| 389 | new helper fs/ntfs/aops.c::mark_ntfs_record_dirty() and remove the no |
| 390 | longer needed fs/ntfs/index.[hc]::__ntfs_index_entry_mark_dirty(). |
| 391 | - Move ntfs_{un,}map_page() from ntfs.h to aops.h and fix resulting |
| 392 | include errors. |
| 393 | - Move the typedefs for runlist_element and runlist from types.h to |
| 394 | runlist.h and fix resulting include errors. |
| 395 | - Remove unused {__,}format_mft_record() from fs/ntfs/mft.c. |
| 396 | - Modify fs/ntfs/mft.c::__mark_mft_record_dirty() to use the helper |
| 397 | mark_ntfs_record_dirty() which also changes the behaviour in that we |
| 398 | now set the buffers belonging to the mft record dirty as well as the |
| 399 | page itself. |
| 400 | - Update fs/ntfs/mft.c::write_mft_record_nolock() and sync_mft_mirror() |
| 401 | to cope with the fact that there now are dirty buffers in mft pages. |
| 402 | - Update fs/ntfs/inode.c::ntfs_write_inode() to also use the helper |
| 403 | mark_ntfs_record_dirty() and thus to set the buffers belonging to the |
| 404 | mft record dirty as well as the page itself. |
| 405 | - Fix compiler warnings on x86-64 in fs/ntfs/dir.c. (Randy Dunlap, |
| 406 | slightly modified by me) |
| 407 | - Add fs/ntfs/mft.c::try_map_mft_record() which fails with -EALREADY if |
| 408 | the mft record is already locked and otherwise behaves the same way |
| 409 | as fs/ntfs/mft.c::map_mft_record(). |
| 410 | - Modify fs/ntfs/mft.c::write_mft_record_nolock() so that it only |
| 411 | writes the mft record if the buffers belonging to it are dirty. |
| 412 | Otherwise we assume that it was written out by other means already. |
| 413 | - Attempting to write outside initialized size is _not_ a bug so remove |
| 414 | the bug check from fs/ntfs/aops.c::ntfs_write_mst_block(). It is in |
| 415 | fact required to write outside initialized size when preparing to |
| 416 | extend the initialized size. |
| 417 | - Map the page instead of using page_address() before writing to it in |
| 418 | fs/ntfs/aops.c::ntfs_mft_writepage(). |
| 419 | - Provide exclusion between opening an inode / mapping an mft record |
| 420 | and accessing the mft record in fs/ntfs/mft.c::ntfs_mft_writepage() |
| 421 | by setting the page not uptodate throughout ntfs_mft_writepage(). |
| 422 | - Clear the page uptodate flag in fs/ntfs/aops.c::ntfs_write_mst_block() |
| 423 | to ensure noone can see the page whilst the mst fixups are applied. |
| 424 | - Add the helper fs/ntfs/mft.c::ntfs_may_write_mft_record() which |
| 425 | checks if an mft record may be written out safely obtaining any |
| 426 | necessary locks in the process. This is used by |
| 427 | fs/ntfs/aops.c::ntfs_write_mst_block(). |
| 428 | - Modify fs/ntfs/aops.c::ntfs_write_mst_block() to also work for |
| 429 | writing mft records and improve its error handling in the process. |
| 430 | Now if any of the records in the page fail to be written out, all |
| 431 | other records will be written out instead of aborting completely. |
| 432 | - Remove ntfs_mft_aops and update all users to use ntfs_mst_aops. |
| 433 | - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to set the |
| 434 | ntfs_mst_aops for all inodes which are NInoMstProtected() and |
| 435 | ntfs_aops for all other inodes. |
| 436 | - Rename fs/ntfs/mft.c::sync_mft_mirror{,_umount}() to |
| 437 | ntfs_sync_mft_mirror{,_umount}() and change their parameters so they |
| 438 | no longer require an ntfs inode to be present. Update all callers. |
| 439 | - Cleanup the error handling in fs/ntfs/mft.c::ntfs_sync_mft_mirror(). |
| 440 | - Clear the page uptodate flag in fs/ntfs/mft.c::ntfs_sync_mft_mirror() |
| 441 | to ensure noone can see the page whilst the mst fixups are applied. |
| 442 | - Remove the no longer needed fs/ntfs/mft.c::ntfs_mft_writepage() and |
| 443 | fs/ntfs/mft.c::try_map_mft_record(). |
| 444 | - Fix callers of fs/ntfs/aops.c::mark_ntfs_record_dirty() to call it |
| 445 | with the ntfs inode which contains the page rather than the ntfs |
| 446 | inode the mft record of which is in the page. |
| 447 | - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by moving the |
| 448 | index inode bitmap inode release code from there to |
| 449 | fs/ntfs/inode.c::ntfs_clear_big_inode(). (Thanks to Christoph |
| 450 | Hellwig for spotting this.) |
| 451 | - Fix race condition in fs/ntfs/inode.c::ntfs_put_inode() by taking the |
| 452 | inode semaphore around the code that sets ni->itype.index.bmp_ino to |
| 453 | NULL and reorganize the code to optimize it a bit. (Thanks to |
| 454 | Christoph Hellwig for spotting this.) |
| 455 | - Modify fs/ntfs/aops.c::mark_ntfs_record_dirty() to no longer take the |
| 456 | ntfs inode as a parameter as this is confusing and misleading and the |
| 457 | needed ntfs inode is available via NTFS_I(page->mapping->host). |
| 458 | Adapt all callers to this change. |
| 459 | - Modify fs/ntfs/mft.c::write_mft_record_nolock() and |
| 460 | fs/ntfs/aops.c::ntfs_write_mst_block() to only check the dirty state |
| 461 | of the first buffer in a record and to take this as the ntfs record |
| 462 | dirty state. We cannot look at the dirty state for subsequent |
| 463 | buffers because we might be racing with |
| 464 | fs/ntfs/aops.c::mark_ntfs_record_dirty(). |
| 465 | - Move the static inline ntfs_init_big_inode() from fs/ntfs/inode.c to |
| 466 | inode.h and make fs/ntfs/inode.c::__ntfs_init_inode() non-static and |
| 467 | add a declaration for it to inode.h. Fix some compilation issues |
| 468 | that resulted due to #includes and header file interdependencies. |
| 469 | - Simplify setup of i_mode in fs/ntfs/inode.c::ntfs_read_locked_inode(). |
| 470 | - Add helpers fs/ntfs/layout.h::MK_MREF() and MK_LE_MREF(). |
| 471 | - Modify fs/ntfs/mft.c::map_extent_mft_record() to only verify the mft |
| 472 | record sequence number if it is specified (i.e. not zero). |
| 473 | - Add fs/ntfs/mft.[hc]::ntfs_mft_record_alloc() and various helper |
| 474 | functions used by it. |
| 475 | - Update Documentation/filesystems/ntfs.txt with instructions on how to |
| 476 | use the Device-Mapper driver with NTFS ftdisk/LDM raid. This removes |
| 477 | the linear raid problem with the Software RAID / MD driver when one |
| 478 | or more of the devices has an odd number of sectors. |
| 479 | |
| 480 | 2.1.20 - Fix two stupid bugs introduced in 2.1.18 release. |
| 481 | |
| 482 | - Fix stupid bug in fs/ntfs/attrib.c::ntfs_attr_reinit_search_ctx() |
| 483 | where we did not clear ctx->al_entry but it was still set due to |
| 484 | changes in ntfs_attr_lookup() and ntfs_external_attr_find() in |
| 485 | particular. |
| 486 | - Fix another stupid bug in fs/ntfs/attrib.c::ntfs_external_attr_find() |
| 487 | where we forgot to unmap the extent mft record when we had finished |
| 488 | enumerating an attribute which caused a bug check to trigger when the |
| 489 | VFS calls ->clear_inode. |
| 490 | |
| 491 | 2.1.19 - Many cleanups, improvements, and a minor bug fix. |
| 492 | |
| 493 | - Update ->setattr (fs/ntfs/inode.c::ntfs_setattr()) to refuse to |
| 494 | change the uid, gid, and mode of an inode as we do not support NTFS |
| 495 | ACLs yet. |
| 496 | - Remove BKL use from ntfs_setattr() syncing up with the rest of the |
| 497 | kernel. |
| 498 | - Get rid of the ugly transparent union in fs/ntfs/dir.c::ntfs_readdir() |
| 499 | and ntfs_filldir() as per suggestion from Al Viro. |
| 500 | - Change '\0' and L'\0' to simply 0 as per advice from Linus Torvalds. |
| 501 | - Update ->truncate (fs/ntfs/inode.c::ntfs_truncate()) to check if the |
| 502 | inode size has changed and to only output an error if so. |
| 503 | - Rename fs/ntfs/attrib.h::attribute_value_length() to ntfs_attr_size(). |
| 504 | - Add le{16,32,64} as well as sle{16,32,64} data types to |
| 505 | fs/ntfs/types.h. |
| 506 | - Change ntfschar to be le16 instead of u16 in fs/ntfs/types.h. |
| 507 | - Add le versions of VCN, LCN, and LSN called leVCN, leLCN, and leLSN, |
| 508 | respectively, to fs/ntfs/types.h. |
| 509 | - Update endianness conversion macros in fs/ntfs/endian.h to use the |
| 510 | new types as appropriate. |
| 511 | - Do proper type casting when using sle64_to_cpup() in fs/ntfs/dir.c |
| 512 | and index.c. |
| 513 | - Add leMFT_REF data type to fs/ntfs/layout.h. |
| 514 | - Update all NTFS header files with the new little endian data types. |
| 515 | Affected files are fs/ntfs/layout.h, logfile.h, and time.h. |
| 516 | - Do proper type casting when using ntfs_is_*_recordp() in |
| 517 | fs/ntfs/logfile.c, mft.c, and super.c. |
| 518 | - Fix all the sparse bitwise warnings. Had to change all the typedef |
| 519 | enums storing little endian values to simple enums plus a typedef for |
| 520 | the datatype to make sparse happy. |
| 521 | - Fix a bug found by the new sparse bitwise warnings where the default |
| 522 | upcase table was defined as a pointer to wchar_t rather than ntfschar |
| 523 | in fs/ntfs/ntfs.h and super.c. |
| 524 | - Change {const_,}cpu_to_le{16,32}(0) to just 0 as suggested by Al Viro. |
| 525 | |
| 526 | 2.1.18 - Fix scheduling latencies at mount time as well as an endianness bug. |
| 527 | |
| 528 | - Remove vol->nr_mft_records as it was pretty meaningless and optimize |
| 529 | the calculation of total/free inodes as used by statfs(). |
| 530 | - Fix scheduling latencies in ntfs_fill_super() by dropping the BKL |
| 531 | because the code itself is using the ntfs_lock semaphore which |
| 532 | provides safe locking. (Ingo Molnar) |
| 533 | - Fix a potential bug in fs/ntfs/mft.c::map_extent_mft_record() that |
| 534 | could occur in the future for when we start closing/freeing extent |
| 535 | inodes if we don't set base_ni->ext.extent_ntfs_inos to NULL after |
| 536 | we free it. |
| 537 | - Rename {find,lookup}_attr() to ntfs_attr_{find,lookup}() as well as |
| 538 | find_external_attr() to ntfs_external_attr_find() to cleanup the |
| 539 | namespace a bit and to be more consistent with libntfs. |
| 540 | - Rename {{re,}init,get,put}_attr_search_ctx() to |
| 541 | ntfs_attr_{{re,}init,get,put}_search_ctx() as well as the type |
| 542 | attr_search_context to ntfs_attr_search_ctx. |
| 543 | - Force use of ntfs_attr_find() in ntfs_attr_lookup() when searching |
| 544 | for the attribute list attribute itself. |
| 545 | - Fix endianness bug in ntfs_external_attr_find(). |
| 546 | - Change ntfs_{external_,}attr_find() to return 0 on success, -ENOENT |
| 547 | if the attribute is not found, and -EIO on real error. In the case |
| 548 | of -ENOENT, the search context is updated to describe the attribute |
| 549 | before which the attribute being searched for would need to be |
| 550 | inserted if such an action were to be desired and in the case of |
| 551 | ntfs_external_attr_find() the search context is also updated to |
| 552 | indicate the attribute list entry before which the attribute list |
| 553 | entry of the attribute being searched for would need to be inserted |
| 554 | if such an action were to be desired. Also make ntfs_find_attr() |
| 555 | static and remove its prototype from attrib.h as it is not used |
| 556 | anywhere other than attrib.c. Update ntfs_attr_lookup() and all |
| 557 | callers of ntfs_{external,}attr_{find,lookup}() for the new return |
| 558 | values. |
| 559 | - Minor cleanup of fs/ntfs/inode.c::ntfs_init_locked_inode(). |
| 560 | |
| 561 | 2.1.17 - Fix bugs in mount time error code paths and other updates. |
| 562 | |
| 563 | - Implement bitmap modification code (fs/ntfs/bitmap.[hc]). This |
| 564 | includes functions to set/clear a single bit or a run of bits. |
| 565 | - Add fs/ntfs/attrib.[hc]::ntfs_find_vcn() which returns the locked |
| 566 | runlist element containing a particular vcn. It also takes care of |
| 567 | mapping any needed runlist fragments. |
| 568 | - Implement cluster (de-)allocation code (fs/ntfs/lcnalloc.[hc]). |
| 569 | - Load attribute definition table from $AttrDef at mount time. |
| 570 | - Fix bugs in mount time error code paths involving (de)allocation of |
| 571 | the default and volume upcase tables. |
| 572 | - Remove ntfs_nr_mounts as it is no longer used. |
| 573 | |
| 574 | 2.1.16 - Implement access time updates, file sync, async io, and read/writev. |
| 575 | |
| 576 | - Add support for readv/writev and aio_read/aio_write (fs/ntfs/file.c). |
| 577 | This is done by setting the appropriate file operations pointers to |
| 578 | the generic helper functions provided by mm/filemap.c. |
| 579 | - Implement fsync, fdatasync, and msync both for files (fs/ntfs/file.c) |
| 580 | and directories (fs/ntfs/dir.c). |
| 581 | - Add support for {a,m,c}time updates to inode.c::ntfs_write_inode(). |
| 582 | Note, except for the root directory and any other system files opened |
| 583 | by the user, the system files will not have their access times |
| 584 | updated as they are only accessed at the inode level an hence the |
| 585 | file level functions which cause the times to be updated are never |
| 586 | invoked. |
| 587 | |
| 588 | 2.1.15 - Invalidate quotas when (re)mounting read-write. |
| 589 | |
| 590 | - Add new element itype.index.collation_rule to the ntfs inode |
| 591 | structure and set it appropriately in ntfs_read_locked_inode(). |
| 592 | - Implement a new inode type "index" to allow efficient access to the |
| 593 | indices found in various system files and adapt inode handling |
| 594 | accordingly (fs/ntfs/inode.[hc]). An index inode is essentially an |
| 595 | attribute inode (NInoAttr() is true) with an attribute type of |
| 596 | AT_INDEX_ALLOCATION. As such, it is no longer allowed to call |
| 597 | ntfs_attr_iget() with an attribute type of AT_INDEX_ALLOCATION as |
| 598 | there would be no way to distinguish between normal attribute inodes |
| 599 | and index inodes. The function to obtain an index inode is |
| 600 | ntfs_index_iget() and it uses the helper function |
| 601 | ntfs_read_locked_index_inode(). Note, we do not overload |
| 602 | ntfs_attr_iget() as indices consist of multiple attributes so using |
| 603 | ntfs_attr_iget() to obtain an index inode would be confusing. |
| 604 | - Ensure that there is no overflow when doing page->index << |
| 605 | PAGE_CACHE_SHIFT by casting page->index to s64 in fs/ntfs/aops.c. |
| 606 | - Use atomic kmap instead of kmap() in fs/ntfs/aops.c::ntfs_read_page() |
| 607 | and ntfs_read_block(). |
| 608 | - Use case sensitive attribute lookups instead of case insensitive ones. |
| 609 | - Lock all page cache pages belonging to mst protected attributes while |
| 610 | accessing them to ensure we never see corrupt data while the page is |
| 611 | under writeout. |
| 612 | - Add framework for generic ntfs collation (fs/ntfs/collation.[hc]). |
| 613 | We have ntfs_is_collation_rule_supported() to check if the collation |
| 614 | rule you want to use is supported and ntfs_collation() which actually |
| 615 | collates two data items. We currently only support COLLATION_BINARY |
| 616 | and COLLATION_NTOFS_ULONG but support for other collation rules will |
| 617 | be added as the need arises. |
| 618 | - Add a new type, ntfs_index_context, to allow retrieval of an index |
| 619 | entry using the corresponding index key. To get an index context, |
| 620 | use ntfs_index_ctx_get() and to release it, use ntfs_index_ctx_put(). |
| 621 | This also adds a new slab cache for the index contexts. To lookup a |
| 622 | key in an index inode, use ntfs_index_lookup(). After modifying an |
| 623 | index entry, call ntfs_index_entry_flush_dcache_page() followed by |
| 624 | ntfs_index_entry_mark_dirty() to ensure the changes are written out |
| 625 | to disk. For details see fs/ntfs/index.[hc]. Note, at present, if |
| 626 | an index entry is in the index allocation attribute rather than the |
| 627 | index root attribute it will not be written out (you will get a |
| 628 | warning message about discarded changes instead). |
| 629 | - Load the quota file ($Quota) and check if quota tracking is enabled |
| 630 | and if so, mark the quotas out of date. This causes windows to |
| 631 | rescan the volume on boot and update all quota entries. |
| 632 | - Add a set_page_dirty address space operation for ntfs_m[fs]t_aops. |
| 633 | It is simply set to __set_page_dirty_nobuffers() to make sure that |
| 634 | running set_page_dirty() on a page containing mft/ntfs records will |
| 635 | not affect the dirty state of the page buffers. |
| 636 | - Add fs/ntfs/index.c::__ntfs_index_entry_mark_dirty() which sets all |
| 637 | buffers that are inside the ntfs record in the page dirty after which |
| 638 | it sets the page dirty. This allows ->writepage to only write the |
| 639 | dirty index records rather than having to write all the records in |
| 640 | the page. Modify fs/ntfs/index.h::ntfs_index_entry_mark_dirty() to |
| 641 | use this rather than __set_page_dirty_nobuffers(). |
| 642 | - Implement fs/ntfs/aops.c::ntfs_write_mst_block() which enables the |
| 643 | writing of page cache pages belonging to mst protected attributes |
| 644 | like the index allocation attribute in directory indices and other |
| 645 | indices like $Quota/$Q, etc. This means that the quota is now marked |
| 646 | out of date on all volumes rather than only on ones where the quota |
| 647 | defaults entry is in the index root attribute of the $Quota/$Q index. |
| 648 | |
| 649 | 2.1.14 - Fix an NFSd caused deadlock reported by several users. |
| 650 | |
| 651 | - Modify fs/ntfs/ntfs_readdir() to copy the index root attribute value |
| 652 | to a buffer so that we can put the search context and unmap the mft |
| 653 | record before calling the filldir() callback. We need to do this |
| 654 | because of NFSd which calls ->lookup() from its filldir callback() |
| 655 | and this causes NTFS to deadlock as ntfs_lookup() maps the mft record |
| 656 | of the directory and since ntfs_readdir() has got it mapped already |
| 657 | ntfs_lookup() deadlocks. |
| 658 | |
| 659 | 2.1.13 - Enable overwriting of resident files and housekeeping of system files. |
| 660 | |
| 661 | - Implement writing of mft records (fs/ntfs/mft.[hc]), which includes |
| 662 | keeping the mft mirror in sync with the mft when mirrored mft records |
| 663 | are written. The functions are write_mft_record{,_nolock}(). The |
| 664 | implementation is quite rudimentary for now with lots of things not |
| 665 | implemented yet but I am not sure any of them can actually occur so |
| 666 | I will wait for people to hit each one and only then implement it. |
| 667 | - Commit open system inodes at umount time. This should make it |
| 668 | virtually impossible for sync_mft_mirror_umount() to ever be needed. |
| 669 | - Implement ->write_inode (fs/ntfs/inode.c::ntfs_write_inode()) for the |
| 670 | ntfs super operations. This gives us inode writing via the VFS inode |
| 671 | dirty code paths. Note: Access time updates are not implemented yet. |
| 672 | - Implement fs/ntfs/mft.[hc]::{,__}mark_mft_record_dirty() and make |
| 673 | fs/ntfs/aops.c::ntfs_writepage() and ntfs_commit_write() use it, thus |
| 674 | finally enabling resident file overwrite! (-8 This also includes a |
| 675 | placeholder for ->writepage (ntfs_mft_writepage()), which for now |
| 676 | just redirties the page and returns. Also, at umount time, we for |
| 677 | now throw away all mft data page cache pages after the last call to |
| 678 | ntfs_commit_inode() in the hope that all inodes will have been |
| 679 | written out by then and hence no dirty (meta)data will be lost. We |
| 680 | also check for this case and emit an error message telling the user |
| 681 | to run chkdsk. |
| 682 | - Use set_page_writeback() and end_page_writeback() in the resident |
| 683 | attribute code path of fs/ntfs/aops.c::ntfs_writepage() otherwise |
| 684 | the radix-tree tag PAGECACHE_TAG_DIRTY remains set even though the |
| 685 | page is clean. |
| 686 | - Implement ntfs_mft_writepage() so it now checks if any of the mft |
| 687 | records in the page are dirty and if so redirties the page and |
| 688 | returns. Otherwise it just returns (after doing set_page_writeback(), |
| 689 | unlock_page(), end_page_writeback() or the radix-tree tag |
| 690 | PAGECACHE_TAG_DIRTY remains set even though the page is clean), thus |
| 691 | alowing the VM to do with the page as it pleases. Also, at umount |
| 692 | time, now only throw away dirty mft (meta)data pages if dirty inodes |
| 693 | are present and ask the user to email us if they see this happening. |
| 694 | - Add functions ntfs_{clear,set}_volume_flags(), to modify the volume |
| 695 | information flags (fs/ntfs/super.c). |
| 696 | - Mark the volume dirty when (re)mounting read-write and mark it clean |
| 697 | when unmounting or remounting read-only. If any volume errors are |
| 698 | found, the volume is left marked dirty to force chkdsk to run. |
| 699 | - Add code to set the NT4 compatibility flag when (re)mounting |
| 700 | read-write for newer NTFS versions but leave it commented out for now |
| 701 | since we do not make any modifications that are NTFS 1.2 specific yet |
| 702 | and since setting this flag breaks Captive-NTFS which is not nice. |
| 703 | This code must be enabled once we start writing NTFS 1.2 specific |
| 704 | changes otherwise Windows NTFS driver might crash / cause corruption. |
| 705 | |
| 706 | 2.1.12 - Fix the second fix to the decompression engine and some cleanups. |
| 707 | |
| 708 | - Add a new address space operations struct, ntfs_mst_aops, for mst |
| 709 | protected attributes. This is because the default ntfs_aops do not |
| 710 | make sense with mst protected data and were they to write anything to |
| 711 | such an attribute they would cause data corruption so we provide |
| 712 | ntfs_mst_aops which does not have any write related operations set. |
| 713 | - Cleanup dirty ntfs inode handling (fs/ntfs/inode.[hc]) which also |
| 714 | includes an adapted ntfs_commit_inode() and an implementation of |
| 715 | ntfs_write_inode() which for now just cleans dirty inodes without |
| 716 | writing them (it does emit a warning that this is happening). |
| 717 | - Undo the second decompression engine fix (see 2.1.9 release ChangeLog |
| 718 | entry) as it was only fixing a theoretical bug but at the same time |
| 719 | it badly broke the handling of sparse and uncompressed compression |
| 720 | blocks. |
| 721 | |
| 722 | 2.1.11 - Driver internal cleanups. |
| 723 | |
| 724 | - Only build logfile.o if building the driver with read-write support. |
| 725 | - Really final white space cleanups. |
| 726 | - Use generic_ffs() instead of ffs() in logfile.c which allows the |
| 727 | log_page_size variable to be optimized by gcc into a constant. |
| 728 | - Rename uchar_t to ntfschar everywhere as uchar_t is unsigned 1-byte |
| 729 | char as defined by POSIX and as found on some systems. |
| 730 | |
| 731 | 2.1.10 - Force read-only (re)mounting of volumes with unsupported volume flags. |
| 732 | |
| 733 | - Finish off the white space cleanups (remove trailing spaces, etc). |
| 734 | - Clean up ntfs_fill_super() and ntfs_read_inode_mount() by removing |
| 735 | the kludges around the first iget(). Instead of (re)setting ->s_op |
| 736 | we have the $MFT inode set up by explicit new_inode() / set ->i_ino / |
| 737 | insert_inode_hash() / call ntfs_read_inode_mount() directly. This |
| 738 | kills the need for second super_operations and allows to return error |
| 739 | from ntfs_read_inode_mount() without resorting to ugly "poisoning" |
| 740 | tricks. (Al Viro) |
| 741 | - Force read-only (re)mounting if any of the following bits are set in |
| 742 | the volume information flags: |
| 743 | VOLUME_IS_DIRTY, VOLUME_RESIZE_LOG_FILE, |
| 744 | VOLUME_UPGRADE_ON_MOUNT, VOLUME_DELETE_USN_UNDERWAY, |
| 745 | VOLUME_REPAIR_OBJECT_ID, VOLUME_MODIFIED_BY_CHKDSK |
| 746 | To make this easier we define VOLUME_MUST_MOUNT_RO_MASK with all the |
| 747 | above bits set so the test is made easy. |
| 748 | |
| 749 | 2.1.9 - Fix two bugs in decompression engine. |
| 750 | |
| 751 | - Fix a bug where we would not always detect that we have reached the |
| 752 | end of a compression block because we were ending at minus one byte |
| 753 | which is effectively the same as being at the end. The fix is to |
| 754 | check whether the uncompressed buffer has been fully filled and if so |
| 755 | we assume we have reached the end of the compression block. A big |
| 756 | thank you to Marcin Gibuła for the bug report, the assistance in |
| 757 | tracking down the bug and testing the fix. |
| 758 | - Fix a possible bug where when a compressed read is truncated to the |
| 759 | end of the file, the offset inside the last page was not truncated. |
| 760 | |
| 761 | 2.1.8 - Handle $MFT mirror and $LogFile, improve time handling, and cleanups. |
| 762 | |
| 763 | - Use get_bh() instead of manual atomic_inc() in fs/ntfs/compress.c. |
| 764 | - Modify fs/ntfs/time.c::ntfs2utc(), get_current_ntfs_time(), and |
| 765 | utc2ntfs() to work with struct timespec instead of time_t on the |
| 766 | Linux UTC time side thus preserving the full precision of the NTFS |
| 767 | time and only loosing up to 99 nano-seconds in the Linux UTC time. |
| 768 | - Move fs/ntfs/time.c to fs/ntfs/time.h and make the time functions |
| 769 | static inline. |
| 770 | - Remove unused ntfs_dirty_inode(). |
| 771 | - Cleanup super operations declaration in fs/ntfs/super.c. |
| 772 | - Wrap flush_dcache_mft_record_page() in #ifdef NTFS_RW. |
| 773 | - Add NInoTestSetFoo() and NInoTestClearFoo() macro magic to |
| 774 | fs/ntfs/inode.h and use it to declare NInoTest{Set,Clear}Dirty. |
| 775 | - Move typedefs for ntfs_attr and test_t from fs/ntfs/inode.c to |
| 776 | fs/ntfs/inode.h so they can be used elsewhere. |
| 777 | - Determine the mft mirror size as the number of mirrored mft records |
| 778 | and store it in ntfs_volume->mftmirr_size (fs/ntfs/super.c). |
| 779 | - Load the mft mirror at mount time and compare the mft records stored |
| 780 | in it to the ones in the mft. Force a read-only mount if the two do |
| 781 | not match (fs/ntfs/super.c). |
| 782 | - Fix type casting related warnings on 64-bit architectures. Thanks |
| 783 | to Meelis Roos for reporting them. |
| 784 | - Move %L to %ll as %L is floating point and %ll is integer which is |
| 785 | what we want. |
| 786 | - Read the journal ($LogFile) and determine if the volume has been |
| 787 | shutdown cleanly and force a read-only mount if not (fs/ntfs/super.c |
| 788 | and fs/ntfs/logfile.c). This is a little bit of a crude check in |
| 789 | that we only look at the restart areas and not at the actual log |
| 790 | records so that there will be a very small number of cases where we |
| 791 | think that a volume is dirty when in fact it is clean. This should |
| 792 | only affect volumes that have not been shutdown cleanly and did not |
| 793 | have any pending, non-check-pointed i/o. |
| 794 | - If the $LogFile indicates a clean shutdown and a read-write (re)mount |
| 795 | is requested, empty $LogFile by overwriting it with 0xff bytes to |
| 796 | ensure that Windows cannot cause data corruption by replaying a stale |
| 797 | journal after Linux has written to the volume. |
| 798 | |
| 799 | 2.1.7 - Enable NFS exporting of mounted NTFS volumes. |
| 800 | |
| 801 | - Set i_generation in the VFS inode from the seq_no of the NTFS inode. |
| 802 | - Make ntfs_lookup() NFS export safe, i.e. use d_splice_alias(), etc. |
| 803 | - Implement ->get_dentry() in fs/ntfs/namei.c::ntfs_get_dentry() as the |
| 804 | default doesn't allow inode number 0 which is a valid inode on NTFS |
| 805 | and even if it did allow that it uses iget() instead of ntfs_iget() |
| 806 | which makes it useless for us. |
| 807 | - Implement ->get_parent() in fs/ntfs/namei.c::ntfs_get_parent() as the |
| 808 | default just returns -EACCES which is not very useful. |
| 809 | - Define export operations (->s_export_op) for NTFS (ntfs_export_ops) |
| 810 | and set them up in the super block at mount time (super.c) this |
| 811 | allows mounted NTFS volumes to be exported via NFS. |
| 812 | - Add missing return -EOPNOTSUPP; in |
| 813 | fs/ntfs/aops.c::ntfs_commit_nonresident_write(). |
| 814 | - Enforce no atime and no dir atime updates at mount/remount time as |
| 815 | they are not implemented yet anyway. |
| 816 | - Move a few assignments in fs/ntfs/attrib.c::load_attribute_list() to |
| 817 | after a NULL check. Thanks to Dave Jones for pointing this out. |
| 818 | |
| 819 | 2.1.6 - Fix minor bug in handling of compressed directories. |
| 820 | |
| 821 | - Fix bug in handling of compressed directories. A compressed |
| 822 | directory is not really compressed so when we set the ->i_blocks |
| 823 | field of a compressed directory inode we were setting it from the |
| 824 | non-existing field ni->itype.compressed.size which gave random |
| 825 | results... For directories we now always use ni->allocated_size. |
| 826 | |
| 827 | 2.1.5 - Fix minor bug in attribute list attribute handling. |
| 828 | |
| 829 | - Fix bug in attribute list handling. Actually it is not as much a bug |
| 830 | as too much protection in that we were not allowing attribute lists |
| 831 | which waste space on disk while Windows XP clearly allows it and in |
| 832 | fact creates such attribute lists so our driver was failing. |
| 833 | - Update NTFS documentation ready for 2.6 kernel release. |
| 834 | |
| 835 | 2.1.4 - Reduce compiler requirements. |
| 836 | |
| 837 | - Remove all uses of unnamed structs and unions in the driver to make |
| 838 | old and newer gcc versions happy. Makes it a bit uglier IMO but at |
| 839 | least people will stop hassling me about it. |
| 840 | |
| 841 | 2.1.3 - Important bug fixes in corner cases. |
| 842 | |
| 843 | - super.c::parse_ntfs_boot_sector(): Correct the check for 64-bit |
| 844 | clusters. (Philipp Thomas) |
| 845 | - attrib.c::load_attribute_list(): Fix bug when initialized_size is a |
| 846 | multiple of the block_size but not the cluster size. (Szabolcs |
| 847 | Szakacsits <szaka@sienet.hu>) |
| 848 | |
| 849 | 2.1.2 - Important bug fixes aleviating the hangs in statfs. |
| 850 | |
| 851 | - Fix buggy free cluster and free inode determination logic. |
| 852 | |
| 853 | 2.1.1 - Minor updates. |
| 854 | |
| 855 | - Add handling for initialized_size != data_size in compressed files. |
| 856 | - Reduce function local stack usage from 0x3d4 bytes to just noise in |
| 857 | fs/ntfs/upcase.c. (Randy Dunlap <rddunlap@osdl.ord>) |
| 858 | - Remove compiler warnings for newer gcc. |
| 859 | - Pages are no longer kmapped by mm/filemap.c::generic_file_write() |
| 860 | around calls to ->{prepare,commit}_write. Adapt NTFS appropriately |
| 861 | in fs/ntfs/aops.c::ntfs_prepare_nonresident_write() by using |
| 862 | kmap_atomic(KM_USER0). |
| 863 | |
| 864 | 2.1.0 - First steps towards write support: implement file overwrite. |
| 865 | |
| 866 | - Add configuration option for developmental write support with an |
| 867 | appropriately scary configuration help text. |
| 868 | - Initial implementation of fs/ntfs/aops.c::ntfs_writepage() and its |
| 869 | helper fs/ntfs/aops.c::ntfs_write_block(). This enables mmap(2) based |
| 870 | overwriting of existing files on ntfs. Note: Resident files are |
| 871 | only written into memory, and not written out to disk at present, so |
| 872 | avoid writing to files smaller than about 1kiB. |
| 873 | - Initial implementation of fs/ntfs/aops.c::ntfs_prepare_write(), its |
| 874 | helper fs/ntfs/aops.c::ntfs_prepare_nonresident_write() and their |
| 875 | counterparts, fs/ntfs/aops.c::ntfs_commit_write(), and |
| 876 | fs/ntfs/aops.c::ntfs_commit_nonresident_write(), respectively. Also, |
| 877 | add generic_file_write() to the ntfs file operations (fs/ntfs/file.c). |
| 878 | This enables write(2) based overwriting of existing files on ntfs. |
| 879 | Note: As with mmap(2) based overwriting, resident files are only |
| 880 | written into memory, and not written out to disk at present, so avoid |
| 881 | writing to files smaller than about 1kiB. |
| 882 | - Implement ->truncate (fs/ntfs/inode.c::ntfs_truncate()) and |
| 883 | ->setattr() (fs/ntfs/inode.c::ntfs_setattr()) inode operations for |
| 884 | files with the purpose of intercepting and aborting all i_size |
| 885 | changes which we do not support yet. ntfs_truncate() actually only |
| 886 | emits a warning message but AFAICS our interception of i_size changes |
| 887 | elsewhere means ntfs_truncate() never gets called for i_size changes. |
| 888 | It is only called from generic_file_write() when we fail in |
| 889 | ntfs_prepare_{,nonresident_}write() in order to discard any |
| 890 | instantiated buffers beyond i_size. Thus i_size is not actually |
| 891 | changed so our warning message is enough. Unfortunately it is not |
| 892 | possible to easily determine if i_size is being changed or not hence |
| 893 | we just emit an appropriately worded error message. |
| 894 | |
| 895 | 2.0.25 - Small bug fixes and cleanups. |
| 896 | |
| 897 | - Unlock the page in an out of memory error code path in |
| 898 | fs/ntfs/aops.c::ntfs_read_block(). |
| 899 | - If fs/ntfs/aops.c::ntfs_read_page() is called on an uptodate page, |
| 900 | just unlock the page and return. (This can happen due to ->writepage |
| 901 | clearing PageUptodate() during write out of MstProtected() |
| 902 | attributes. |
| 903 | - Remove leaked write code again. |
| 904 | |
| 905 | 2.0.24 - Cleanups. |
| 906 | |
| 907 | - Treat BUG_ON() as ASSERT() not VERIFY(), i.e. do not use side effects |
| 908 | inside BUG_ON(). (Adam J. Richter) |
| 909 | - Split logical OR expressions inside BUG_ON() into individual BUG_ON() |
| 910 | calls for improved debugging. (Adam J. Richter) |
| 911 | - Add errors flag to the ntfs volume state, accessed via |
| 912 | NVol{,Set,Clear}Errors(vol). |
| 913 | - Do not allow read-write remounts of read-only volumes with errors. |
| 914 | - Clarify comment for ntfs file operation sendfile which was added by |
| 915 | Christoph Hellwig a while ago (just using generic_file_sendfile()) |
| 916 | to say that ntfs ->sendfile is only used for the case where the |
| 917 | source data is on the ntfs partition and the destination is |
| 918 | somewhere else, i.e. nothing we need to concern ourselves with. |
| 919 | - Add generic_file_write() as our ntfs file write operation. |
| 920 | |
| 921 | 2.0.23 - Major bug fixes (races, deadlocks, non-i386 architectures). |
| 922 | |
| 923 | - Massive internal locking changes to mft record locking. Fixes lock |
| 924 | recursion and replaces the mrec_lock read/write semaphore with a |
| 925 | mutex. Also removes the now superfluous mft_count. This fixes several |
| 926 | race conditions and deadlocks, especially in the future write code. |
| 927 | - Fix ntfs over loopback for compressed files by adding an |
| 928 | optimization barrier. (gcc was screwing up otherwise ?) |
| 929 | - Miscellaneous cleanups all over the code and a fix or two in error |
| 930 | handling code paths. |
| 931 | Thanks go to Christoph Hellwig for pointing out the following two: |
| 932 | - Remove now unused function fs/ntfs/malloc.h::vmalloc_nofs(). |
| 933 | - Fix ntfs_free() for ia64 and parisc by checking for VMALLOC_END, too. |
| 934 | |
| 935 | 2.0.22 - Cleanups, mainly to ntfs_readdir(), and use C99 initializers. |
| 936 | |
| 937 | - Change fs/ntfs/dir.c::ntfs_reddir() to only read/write ->f_pos once |
| 938 | at entry/exit respectively. |
| 939 | - Use C99 initializers for structures. |
| 940 | - Remove unused variable blocks from fs/ntfs/aops.c::ntfs_read_block(). |
| 941 | |
| 942 | 2.0.21 - Check for, and refuse to work with too large files/directories/volumes. |
| 943 | |
| 944 | - Limit volume size at mount time to 2TiB on architectures where |
| 945 | unsigned long is 32-bits (fs/ntfs/super.c::parse_ntfs_boot_sector()). |
| 946 | This is the most we can do without overflowing the 32-bit limit of |
| 947 | the block device size imposed on us by sb_bread() and sb_getblk() |
| 948 | for the time being. |
| 949 | - Limit file/directory size at open() time to 16TiB on architectures |
| 950 | where unsigned long is 32-bits (fs/ntfs/file.c::ntfs_file_open() and |
| 951 | fs/ntfs/dir.c::ntfs_dir_open()). This is the most we can do without |
| 952 | overflowing the page cache page index. |
| 953 | |
| 954 | 2.0.20 - Support non-resident directory index bitmaps, fix page leak in readdir. |
| 955 | |
| 956 | - Move the directory index bitmap to use an attribute inode instead of |
| 957 | having special fields for it inside the ntfs inode structure. This |
| 958 | means that the index bitmaps now use the page cache for i/o, too, |
| 959 | and also as a side effect we get support for non-resident index |
| 960 | bitmaps for free. |
| 961 | - Simplify/cleanup error handling in fs/ntfs/dir.c::ntfs_readdir() and |
| 962 | fix a page leak that manifested itself in some cases. |
| 963 | - Add fs/ntfs/inode.c::ntfs_put_inode(), which we need to release the |
| 964 | index bitmap inode on the final iput(). |
| 965 | |
| 966 | 2.0.19 - Fix race condition, improvements, and optimizations in i/o interface. |
| 967 | |
| 968 | - Apply block optimization added to fs/ntfs/aops.c::ntfs_read_block() |
| 969 | to fs/ntfs/compress.c::ntfs_file_read_compressed_block() as well. |
| 970 | - Drop the "file" from ntfs_file_read_compressed_block(). |
| 971 | - Rename fs/ntfs/aops.c::ntfs_enb_buffer_read_async() to |
| 972 | ntfs_end_buffer_async_read() (more like the fs/buffer.c counterpart). |
| 973 | - Update ntfs_end_buffer_async_read() with the improved logic from |
| 974 | its updated counterpart fs/buffer.c::end_buffer_async_read(). Apply |
| 975 | further logic improvements to better determine when we set PageError. |
| 976 | - Update submission of buffers in fs/ntfs/aops.c::ntfs_read_block() to |
| 977 | check for the buffers being uptodate first in line with the updated |
| 978 | fs/buffer.c::block_read_full_page(). This plugs a small race |
| 979 | condition. |
| 980 | |
| 981 | 2.0.18 - Fix race condition in reading of compressed files. |
| 982 | |
| 983 | - There was a narrow window between checking a buffer head for being |
| 984 | uptodate and locking it in ntfs_file_read_compressed_block(). We now |
| 985 | lock the buffer and then check whether it is uptodate or not. |
| 986 | |
| 987 | 2.0.17 - Cleanups and optimizations - shrinking the ToDo list. |
| 988 | |
| 989 | - Modify fs/ntfs/inode.c::ntfs_read_locked_inode() to return an error |
| 990 | code and update callers, i.e. ntfs_iget(), to pass that error code |
| 991 | up instead of just using -EIO. |
| 992 | - Modifications to super.c to ensure that both mount and remount |
| 993 | cannot set any write related options when the driver is compiled |
| 994 | read-only. |
| 995 | - Optimize block resolution in fs/ntfs/aops.c::ntfs_read_block() to |
| 996 | cache the current runlist element. This should improve performance |
| 997 | when reading very large and/or very fragmented data. |
| 998 | |
| 999 | 2.0.16 - Convert access to $MFT/$BITMAP to attribute inode API. |
| 1000 | |
| 1001 | - Fix a stupid bug introduced in 2.0.15 where we were unmapping the |
| 1002 | wrong inode in fs/ntfs/inode.c::ntfs_attr_iget(). |
| 1003 | - Fix debugging check in fs/ntfs/aops.c::ntfs_read_block(). |
| 1004 | - Convert $MFT/$BITMAP access to attribute inode API and remove all |
| 1005 | remnants of the ugly mftbmp address space and operations hack. This |
| 1006 | means we finally have only one readpage function as well as only one |
| 1007 | async io completion handler. Yey! The mft bitmap is now just an |
| 1008 | attribute inode and is accessed from vol->mftbmp_ino just as if it |
| 1009 | were a normal file. Fake inodes rule. (-: |
| 1010 | |
| 1011 | 2.0.15 - Fake inodes based attribute i/o via the pagecache, fixes and cleanups. |
| 1012 | |
| 1013 | - Fix silly bug in fs/ntfs/super.c::parse_options() which was causing |
| 1014 | remounts to fail when the partition had an entry in /etc/fstab and |
| 1015 | the entry specified the nls= option. |
| 1016 | - Apply same macro magic used in fs/ntfs/inode.h to fs/ntfs/volume.h to |
| 1017 | expand all the helper functions NVolFoo(), NVolSetFoo(), and |
| 1018 | NVolClearFoo(). |
| 1019 | - Move copyright statement from driver initialisation message to |
| 1020 | module description (fs/super.c). This makes the initialisation |
| 1021 | message fit on one line and fits in better with rest of kernel. |
| 1022 | - Update fs/ntfs/attrib.c::map_run_list() to work on both real and |
| 1023 | attribute inodes, and both for files and directories. |
| 1024 | - Implement fake attribute inodes allowing all attribute i/o to go via |
| 1025 | the page cache and to use all the normal vfs/mm functionality: |
| 1026 | - Add ntfs_attr_iget() and its helper ntfs_read_locked_attr_inode() |
| 1027 | to fs/ntfs/inode.c. |
| 1028 | - Add needed cleanup code to ntfs_clear_big_inode(). |
| 1029 | - Merge address space operations for files and directories (aops.c), |
| 1030 | now just have ntfs_aops: |
| 1031 | - Rename: |
| 1032 | end_buffer_read_attr_async() -> ntfs_end_buffer_read_async(), |
| 1033 | ntfs_attr_read_block() -> ntfs_read_block(), |
| 1034 | ntfs_file_read_page() -> ntfs_readpage(). |
| 1035 | - Rewrite fs/ntfs/aops.c::ntfs_readpage() to work on both real and |
| 1036 | attribute inodes, and both for files and directories. |
| 1037 | - Remove obsolete fs/ntfs/aops.c::ntfs_mst_readpage(). |
| 1038 | |
| 1039 | 2.0.14 - Run list merging code cleanup, minor locking changes, typo fixes. |
| 1040 | |
| 1041 | - Change fs/ntfs/super.c::ntfs_statfs() to not rely on BKL by moving |
| 1042 | the locking out of super.c::get_nr_free_mft_records() and taking and |
| 1043 | dropping the mftbmp_lock rw_semaphore in ntfs_statfs() itself. |
| 1044 | - Bring attribute runlist merging code (fs/ntfs/attrib.c) in sync with |
| 1045 | current userspace ntfs library code. This means that if a merge |
| 1046 | fails the original runlists are always left unmodified instead of |
| 1047 | being silently corrupted. |
| 1048 | - Misc typo fixes. |
| 1049 | |
| 1050 | 2.0.13 - Use iget5_locked() in preparation for fake inodes and small cleanups. |
| 1051 | |
| 1052 | - Remove nr_mft_bits and the now superfluous union with nr_mft_records |
| 1053 | from ntfs_volume structure. |
| 1054 | - Remove nr_lcn_bits and the now superfluous union with nr_clusters |
| 1055 | from ntfs_volume structure. |
| 1056 | - Use iget5_locked() and friends instead of conventional iget(). Wrap |
| 1057 | the call in fs/ntfs/inode.c::ntfs_iget() and update callers of iget() |
| 1058 | to use ntfs_iget(). Leave only one iget() call at mount time so we |
| 1059 | don't need an ntfs_iget_mount(). |
| 1060 | - Change fs/ntfs/inode.c::ntfs_new_extent_inode() to take mft_no as an |
| 1061 | additional argument. |
| 1062 | |
| 1063 | 2.0.12 - Initial cleanup of address space operations following 2.0.11 changes. |
| 1064 | |
| 1065 | - Merge fs/ntfs/aops.c::end_buffer_read_mst_async() and |
| 1066 | fs/ntfs/aops.c::end_buffer_read_file_async() into one function |
| 1067 | fs/ntfs/aops.c::end_buffer_read_attr_async() using NInoMstProtected() |
| 1068 | to determine whether to apply mst fixups or not. |
| 1069 | - Above change allows merging fs/ntfs/aops.c::ntfs_file_read_block() |
| 1070 | and fs/ntfs/aops.c::ntfs_mst_readpage() into one function |
| 1071 | fs/ntfs/aops.c::ntfs_attr_read_block(). Also, create a tiny wrapper |
| 1072 | fs/ntfs/aops.c::ntfs_mst_readpage() to transform the parameters from |
| 1073 | the VFS readpage function prototype to the ntfs_attr_read_block() |
| 1074 | function prototype. |
| 1075 | |
| 1076 | 2.0.11 - Initial preparations for fake inode based attribute i/o. |
| 1077 | |
| 1078 | - Move definition of ntfs_inode_state_bits to fs/ntfs/inode.h and |
| 1079 | do some macro magic (adapted from include/linux/buffer_head.h) to |
| 1080 | expand all the helper functions NInoFoo(), NInoSetFoo(), and |
| 1081 | NInoClearFoo(). |
| 1082 | - Add new flag to ntfs_inode_state_bits: NI_Sparse. |
| 1083 | - Add new fields to ntfs_inode structure to allow use of fake inodes |
| 1084 | for attribute i/o: type, name, name_len. Also add new state bits: |
| 1085 | NI_Attr, which, if set, indicates the inode is a fake inode, and |
| 1086 | NI_MstProtected, which, if set, indicates the attribute uses multi |
| 1087 | sector transfer protection, i.e. fixups need to be applied after |
| 1088 | reads and before/after writes. |
| 1089 | - Rename fs/ntfs/inode.c::ntfs_{new,clear,destroy}_inode() to |
| 1090 | ntfs_{new,clear,destroy}_extent_inode() and update callers. |
| 1091 | - Use ntfs_clear_extent_inode() in fs/ntfs/inode.c::__ntfs_clear_inode() |
| 1092 | instead of ntfs_destroy_extent_inode(). |
| 1093 | - Cleanup memory deallocations in {__,}ntfs_clear_{,big_}inode(). |
| 1094 | - Make all operations on ntfs inode state bits use the NIno* functions. |
| 1095 | - Set up the new ntfs inode fields and state bits in |
| 1096 | fs/ntfs/inode.c::ntfs_read_inode() and add appropriate cleanup of |
| 1097 | allocated memory to __ntfs_clear_inode(). |
| 1098 | - Cleanup ntfs_inode structure a bit for better ordering of elements |
| 1099 | w.r.t. their size to allow better packing of the structure in memory. |
| 1100 | |
| 1101 | 2.0.10 - There can only be 2^32 - 1 inodes on an NTFS volume. |
| 1102 | |
| 1103 | - Add check at mount time to verify that the number of inodes on the |
| 1104 | volume does not exceed 2^32 - 1, which is the maximum allowed for |
| 1105 | NTFS according to Microsoft. |
| 1106 | - Change mft_no member of ntfs_inode structure to be unsigned long. |
| 1107 | Update all users. This makes ntfs_inode->mft_no just a copy of struct |
| 1108 | inode->i_ino. But we can't just always use struct inode->i_ino and |
| 1109 | remove mft_no because extent inodes do not have an attached struct |
| 1110 | inode. |
| 1111 | |
| 1112 | 2.0.9 - Decompression engine now uses a single buffer and other cleanups. |
| 1113 | |
| 1114 | - Change decompression engine to use a single buffer protected by a |
| 1115 | spin lock instead of per-CPU buffers. (Rusty Russell) |
| 1116 | - Do not update cb_pos when handling a partial final page during |
| 1117 | decompression of a sparse compression block, as the value is later |
| 1118 | reset without being read/used. (Rusty Russell) |
| 1119 | - Switch to using the new KM_BIO_SRC_IRQ for atomic kmap()s. (Andrew |
| 1120 | Morton) |
| 1121 | - Change buffer size in ntfs_readdir()/ntfs_filldir() to use |
| 1122 | NLS_MAX_CHARSET_SIZE which makes the buffers almost 1kiB each but |
| 1123 | it also makes everything safer so it is a good thing. |
| 1124 | - Miscellaneous minor cleanups to comments. |
| 1125 | |
| 1126 | 2.0.8 - Major updates for handling of case sensitivity and dcache aliasing. |
| 1127 | |
| 1128 | Big thanks go to Al Viro and other inhabitants of #kernel for investing |
| 1129 | their time to discuss the case sensitivity and dcache aliasing issues. |
| 1130 | |
| 1131 | - Remove unused source file fs/ntfs/attraops.c. |
| 1132 | - Remove show_inodes mount option(s), thus dropping support for |
| 1133 | displaying of short file names. |
| 1134 | - Remove deprecated mount option posix. |
| 1135 | - Restore show_sys_files mount option. |
| 1136 | - Add new mount option case_sensitive, to determine if the driver |
| 1137 | treats file names as case sensitive or not. If case sensitive, create |
| 1138 | file names in the POSIX namespace. Otherwise create file names in the |
| 1139 | LONG/WIN32 namespace. Note, files remain accessible via their short |
| 1140 | file name, if it exists. |
| 1141 | - Remove really dumb logic bug in boot sector recovery code. |
| 1142 | - Fix dcache aliasing issues wrt short/long file names via changes |
| 1143 | to fs/ntfs/dir.c::ntfs_lookup_inode_by_name() and |
| 1144 | fs/ntfs/namei.c::ntfs_lookup(): |
| 1145 | - Add additional argument to ntfs_lookup_inode_by_name() in which we |
| 1146 | return information about the matching file name if the case is not |
| 1147 | matching or the match is a short file name. See comments above the |
| 1148 | function definition for details. |
| 1149 | - Change ntfs_lookup() to only create dcache entries for the correctly |
| 1150 | cased file name and only for the WIN32 namespace counterpart of DOS |
| 1151 | namespace file names. This ensures we have only one dentry per |
| 1152 | directory and also removes all dcache aliasing issues between short |
| 1153 | and long file names once we add write support. See comments above |
| 1154 | function for details. |
| 1155 | - Fix potential 1 byte overflow in fs/ntfs/unistr.c::ntfs_ucstonls(). |
| 1156 | |
| 1157 | 2.0.7 - Minor cleanups and updates for changes in core kernel code. |
| 1158 | |
| 1159 | - Remove much of the NULL struct element initializers. |
| 1160 | - Various updates to make compatible with recent kernels. |
| 1161 | - Remove defines of MAX_BUF_PER_PAGE and include linux/buffer_head.h |
| 1162 | in fs/ntfs/ntfs.h instead. |
| 1163 | - Remove no longer needed KERNEL_VERSION checks. We are now in the |
| 1164 | kernel proper so they are no longer needed. |
| 1165 | |
| 1166 | 2.0.6 - Major bugfix to make compatible with other kernel changes. |
| 1167 | |
| 1168 | - Initialize the mftbmp address space properly now that there are more |
| 1169 | fields in the struct address_space. This was leading to hangs and |
| 1170 | oopses on umount since 2.5.12 because of changes to other parts of |
| 1171 | the kernel. We probably want a kernel generic init_address_space() |
| 1172 | function... |
| 1173 | - Drop BKL from ntfs_readdir() after consultation with Al Viro. The |
| 1174 | only caller of ->readdir() is vfs_readdir() which holds i_sem during |
| 1175 | the call, and i_sem is sufficient protection against changes in the |
| 1176 | directory inode (including ->i_size). |
| 1177 | - Use generic_file_llseek() for directories (as opposed to |
| 1178 | default_llseek()) as this downs i_sem instead of the BKL which is |
| 1179 | what we now need for exclusion against ->f_pos changes considering we |
| 1180 | no longer take the BKL in ntfs_readdir(). |
| 1181 | |
| 1182 | 2.0.5 - Major bugfix. Buffer overflow in extent inode handling. |
| 1183 | |
| 1184 | - No need to set old blocksize in super.c::ntfs_fill_super() as the |
| 1185 | VFS does so via invocation of deactivate_super() calling |
| 1186 | fs->fill_super() calling block_kill_super() which does it. |
| 1187 | - BKL moved from VFS into dir.c::ntfs_readdir(). (Linus Torvalds) |
| 1188 | -> Do we really need it? I don't think so as we have exclusion on |
| 1189 | the directory ntfs_inode rw_semaphore mrec_lock. We mmight have to |
| 1190 | move the ->f_pos accesses under the mrec_lock though. Check this... |
| 1191 | - Fix really, really, really stupid buffer overflow in extent inode |
| 1192 | handling in mft.c::map_extent_mft_record(). |
| 1193 | |
| 1194 | 2.0.4 - Cleanups and updates for kernel 2.5.11. |
| 1195 | |
| 1196 | - Add documentation on how to use the MD driver to be able to use NTFS |
| 1197 | stripe and volume sets in Linux and generally cleanup documentation |
| 1198 | a bit. |
| 1199 | Remove all uses of kdev_t in favour of struct block_device *: |
| 1200 | - Change compress.c::ntfs_file_read_compressed_block() to use |
| 1201 | sb_getblk() instead of getblk(). |
| 1202 | - Change super.c::ntfs_fill_super() to use bdev_hardsect_size() instead |
| 1203 | of get_hardsect_size(). |
| 1204 | - No need to get old blocksize in super.c::ntfs_fill_super() as |
| 1205 | fs/super.c::get_sb_bdev() already does this. |
| 1206 | - Set bh->b_bdev instead of bh->b_dev throughout aops.c. |
| 1207 | |
| 1208 | 2.0.3 - Small bug fixes, cleanups, and performance improvements. |
| 1209 | |
| 1210 | - Remove some dead code from mft.c. |
| 1211 | - Optimize readpage and read_block functions throughout aops.c so that |
| 1212 | only initialized blocks are read. Non-initialized ones have their |
| 1213 | buffer head mapped, zeroed, and set up to date, without scheduling |
| 1214 | any i/o. Thanks to Al Viro for advice on how to avoid the device i/o. |
| 1215 | Thanks go to Andrew Morton for spotting the below: |
| 1216 | - Fix buglet in allocate_compression_buffers() error code path. |
| 1217 | - Call flush_dcache_page() after modifying page cache page contents in |
| 1218 | ntfs_file_readpage(). |
| 1219 | - Check for existence of page buffers throughout aops.c before calling |
| 1220 | create_empty_buffers(). This happens when an I/O error occurs and the |
| 1221 | read is retried. (It also happens once writing is implemented so that |
| 1222 | needed doing anyway but I had left it for later...) |
| 1223 | - Don't BUG_ON() uptodate and/or mapped buffers throughout aops.c in |
| 1224 | readpage and read_block functions. Reasoning same as above (i.e. I/O |
| 1225 | error retries and future write code paths.) |
| 1226 | |
| 1227 | 2.0.2 - Minor updates and cleanups. |
| 1228 | |
| 1229 | - Cleanup: rename mst.c::__post_read_mst_fixup to post_write_mst_fixup |
| 1230 | and cleanup the code a bit, removing the unused size parameter. |
| 1231 | - Change default fmask to 0177 and update documentation. |
| 1232 | - Change attrib.c::get_attr_search_ctx() to return the search context |
| 1233 | directly instead of taking the address of a pointer. A return value |
| 1234 | of NULL means the allocation failed. Updated all callers |
| 1235 | appropriately. |
| 1236 | - Update to 2.5.9 kernel (preserving backwards compatibility) by |
| 1237 | replacing all occurences of page->buffers with page_buffers(page). |
| 1238 | - Fix minor bugs in runlist merging, also minor cleanup. |
| 1239 | - Updates to bootsector layout and mft mirror contents descriptions. |
| 1240 | - Small bug fix in error detection in unistr.c and some cleanups. |
| 1241 | - Grow name buffer allocations in unistr.c in aligned mutlipled of 64 |
| 1242 | bytes. |
| 1243 | |
| 1244 | 2.0.1 - Minor updates. |
| 1245 | |
| 1246 | - Make default umask correspond to documentation. |
| 1247 | - Improve documentation. |
| 1248 | - Set default mode to include execute bit. The {u,f,d}mask can be used |
| 1249 | to take it away if desired. This allows binaries to be executed from |
| 1250 | a mounted ntfs partition. |
| 1251 | |
| 1252 | 2.0.0 - New version number. Remove TNG from the name. Now in the kernel. |
| 1253 | |
| 1254 | - Add kill_super, just keeping up with the vfs changes in the kernel. |
| 1255 | - Repeat some changes from tng-0.0.8 that somehow got lost on the way |
| 1256 | from the CVS import into BitKeeper. |
| 1257 | - Begin to implement proper handling of allocated_size vs |
| 1258 | initialized_size vs data_size (i.e. i_size). Done are |
| 1259 | mft.c::ntfs_mft_readpage(), aops.c::end_buffer_read_index_async(), |
| 1260 | and attrib.c::load_attribute_list(). |
| 1261 | - Lock the runlist in attrib.c::load_attribute_list() while using it. |
| 1262 | - Fix memory leak in ntfs_file_read_compressed_block() and generally |
| 1263 | clean up compress.c a little, removing some uncommented/unused debug |
| 1264 | code. |
| 1265 | - Tidy up dir.c a little bit. |
| 1266 | - Don't bother getting the runlist in inode.c::ntfs_read_inode(). |
| 1267 | - Merge mft.c::ntfs_mft_readpage() and aops.c::ntfs_index_readpage() |
| 1268 | creating aops.c::ntfs_mst_readpage(), improving the handling of |
| 1269 | holes and overflow in the process and implementing the correct |
| 1270 | equivalent of ntfs_file_get_block() in ntfs_mst_readpage() itself. |
| 1271 | I am aiming for correctness at the moment. Modularisation can come |
| 1272 | later. |
| 1273 | - Rename aops.c::end_buffer_read_index_async() to |
| 1274 | end_buffer_read_mst_async() and optimize the overflow checking and |
| 1275 | handling. |
| 1276 | - Use the host of the mftbmp address space mapping to hold the ntfs |
| 1277 | volume. This is needed so the async i/o completion handler can |
| 1278 | retrieve a pointer to the volume. Hopefully this will not cause |
| 1279 | problems elsewhere in the kernel... Otherwise will need to use a |
| 1280 | fake inode. |
| 1281 | - Complete implementation of proper handling of allocated_size vs |
| 1282 | initialized_size vs data_size (i.e. i_size) in whole driver. |
| 1283 | Basically aops.c is now completely rewritten. |
| 1284 | - Change NTFS driver name to just NTFS and set version number to 2.0.0 |
| 1285 | to make a clear distinction from the old driver which is still on |
| 1286 | version 1.1.22. |
| 1287 | |
| 1288 | tng-0.0.8 - 08/03/2002 - Now using BitKeeper, http://linux-ntfs.bkbits.net/ |
| 1289 | |
| 1290 | - Replace bdevname(sb->s_dev) with sb->s_id. |
| 1291 | - Remove now superfluous new-line characters in all callers of |
| 1292 | ntfs_debug(). |
| 1293 | - Apply kludge in ntfs_read_inode(), setting i_nlink to 1 for |
| 1294 | directories. Without this the "find" utility gets very upset which is |
| 1295 | fair enough as Linux/Unix do not support directory hard links. |
| 1296 | - Further runlist merging work. (Richard Russon) |
| 1297 | - Backwards compatibility for gcc-2.95. (Richard Russon) |
| 1298 | - Update to kernel 2.5.5-pre1 and rediff the now tiny patch. |
Anton Altaparmakov | c002f42 | 2005-02-03 12:02:56 +0000 | [diff] [blame] | 1299 | - Convert to new filesystem declaration using ->ntfs_get_sb() and |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1300 | replacing ntfs_read_super() with ntfs_fill_super(). |
| 1301 | - Set s_maxbytes to MAX_LFS_FILESIZE to avoid page cache page index |
| 1302 | overflow on 32-bit architectures. |
| 1303 | - Cleanup upcase loading code to use ntfs_(un)map_page(). |
| 1304 | - Disable/reenable preemtion in critical sections of compession engine. |
| 1305 | - Replace device size determination in ntfs_fill_super() with |
| 1306 | sb->s_bdev->bd_inode->i_size (in bytes) and remove now superfluous |
| 1307 | function super.c::get_nr_blocks(). |
| 1308 | - Implement a mount time option (show_inodes) allowing choice of which |
| 1309 | types of inode names readdir() returns and modify ntfs_filldir() |
| 1310 | accordingly. There are several parameters to show_inodes: |
| 1311 | system: system files |
| 1312 | win32: long file names (including POSIX file names) [DEFAULT] |
| 1313 | long: same as win32 |
| 1314 | dos: short file names only (excluding POSIX file names) |
| 1315 | short: same as dos |
| 1316 | posix: same as both win32 and dos |
| 1317 | all: all file names |
| 1318 | Note that the options are additive, i.e. specifying: |
| 1319 | -o show_inodes=system,show_inodes=win32,show_inodes=dos |
| 1320 | is the same as specifying: |
| 1321 | -o show_inodes=all |
| 1322 | Note that the "posix" and "all" options will show all directory |
| 1323 | names, BUT the link count on each directory inode entry is set to 1, |
| 1324 | due to Linux not supporting directory hard links. This may well |
| 1325 | confuse some userspace applications, since the directory names will |
| 1326 | have the same inode numbers. Thus it is NOT advisable to use the |
| 1327 | "posix" or "all" options. We provide them only for completeness sake. |
| 1328 | - Add copies of allocated_size, initialized_size, and compressed_size to |
| 1329 | the ntfs inode structure and set them up in |
| 1330 | inode.c::ntfs_read_inode(). These reflect the unnamed data attribute |
| 1331 | for files and the index allocation attribute for directories. |
| 1332 | - Add copies of allocated_size and initialized_size to ntfs inode for |
| 1333 | $BITMAP attribute of large directories and set them up in |
| 1334 | inode.c::ntfs_read_inode(). |
| 1335 | - Add copies of allocated_size and initialized_size to ntfs volume for |
| 1336 | $BITMAP attribute of $MFT and set them up in |
| 1337 | super.c::load_system_files(). |
| 1338 | - Parse deprecated ntfs driver options (iocharset, show_sys_files, |
| 1339 | posix, and utf8) and tell user what the new options to use are. Note |
| 1340 | we still do support them but they will be removed with kernel 2.7.x. |
| 1341 | - Change all occurences of integer long long printf formatting to hex |
| 1342 | as printk() will not support long long integer format if/when the |
| 1343 | div64 patch goes into the kernel. |
| 1344 | - Make slab caches have stable names and change the names to what they |
| 1345 | were intended to be. These changes are required/made possible by the |
| 1346 | new slab cache name handling which removes the length limitation by |
| 1347 | requiring the caller of kmem_cache_create() to supply a stable name |
| 1348 | which is then referenced but not copied. |
| 1349 | - Rename run_list structure to run_list_element and create a new |
| 1350 | run_list structure containing a pointer to a run_list_element |
| 1351 | structure and a read/write semaphore. Adapt all users of runlists |
| 1352 | to new scheme and take and release the lock as needed. This fixes a |
| 1353 | nasty race as the run_list changes even when inodes are locked for |
| 1354 | reading and even when the inode isn't locked at all, so we really |
| 1355 | needed the serialization. We use a semaphore rather than a spinlock |
| 1356 | as memory allocations can sleep and doing everything GFP_ATOMIC |
| 1357 | would be silly. |
| 1358 | - Cleanup read_inode() removing all code checking for lowest_vcn != 0. |
| 1359 | This can never happen due to the nature of lookup_attr() and how we |
| 1360 | support attribute lists. If it did happen it would imply the inode |
| 1361 | being corrupt. |
| 1362 | - Check for lowest_vcn != 0 in ntfs_read_inode() and mark the inode as |
| 1363 | bad if found. |
| 1364 | - Update to 2.5.6-pre2 changes in struct address_space. |
| 1365 | - Use parent_ino() when accessing d_parent inode number in dir.c. |
| 1366 | - Import Sourceforge CVS repository into BitKeeper repository: |
| 1367 | http://linux-ntfs.bkbits.net/ntfs-tng-2.5 |
| 1368 | - Update fs/Makefile, fs/Config.help, fs/Config.in, and |
| 1369 | Documentation/filesystems/ntfs.txt for NTFS TNG. |
| 1370 | - Create kernel configuration option controlling whether debugging |
| 1371 | is enabled or not. |
| 1372 | - Add the required export of end_buffer_io_sync() from the patches |
| 1373 | directory to the kernel code. |
| 1374 | - Update inode.c::ntfs_show_options() with show_inodes mount option. |
| 1375 | - Update errors mount option. |
| 1376 | |
| 1377 | tng-0.0.7 - 13/02/2002 - The driver is now feature complete for read-only! |
| 1378 | |
| 1379 | - Cleanup mft.c and it's debug/error output in particular. Fix a minor |
| 1380 | bug in mapping of extent inodes. Update all the comments to fit all |
| 1381 | the recent code changes. |
| 1382 | - Modify vcn_to_lcn() to cope with entirely unmapped runlists. |
| 1383 | - Cleanups in compress.c, mostly comments and folding help. |
| 1384 | - Implement attrib.c::map_run_list() as a generic helper. |
| 1385 | - Make compress.c::ntfs_file_read_compressed_block() use map_run_list() |
| 1386 | thus making code shorter and enabling attribute list support. |
| 1387 | - Cleanup incorrect use of [su]64 with %L printf format specifier in |
| 1388 | all source files. Type casts to [unsigned] long long added to correct |
| 1389 | the mismatches (important for architectures which have long long not |
| 1390 | being 64 bits). |
| 1391 | - Merge async io completion handlers for directory indexes and $MFT |
| 1392 | data into one by setting the index_block_size{_bits} of the ntfs |
| 1393 | inode for $MFT to the mft_record_size{_bits} of the ntfs_volume. |
| 1394 | - Cleanup aops.c, update comments. |
| 1395 | - Make ntfs_file_get_block() use map_run_list() so all files now |
| 1396 | support attribute lists. |
| 1397 | - Make ntfs_dir_readpage() almost verbatim copy of |
| 1398 | block_read_full_page() by using ntfs_file_get_block() with only real |
| 1399 | difference being the use of our own async io completion handler |
| 1400 | rather than the default one, thus reducing the amount of code and |
| 1401 | automatically enabling attribute list support for directory indices. |
| 1402 | - Fix bug in load_attribute_list() - forgot to call brelse in error |
| 1403 | code path. |
| 1404 | - Change parameters to find_attr() and lookup_attr(). We no longer |
| 1405 | pass in the upcase table and its length. These can be gotten from |
| 1406 | ctx->ntfs_ino->vol->upcase{_len}. Update all callers. |
| 1407 | - Cleanups in attrib.c. |
| 1408 | - Implement merging of runlists, attrib.c::merge_run_lists() and its |
| 1409 | helpers. (Richard Russon) |
| 1410 | - Attribute lists part 2, attribute extents and multi part runlists: |
| 1411 | enable proper support for LCN_RL_NOT_MAPPED and automatic mapping of |
| 1412 | further runlist parts via attrib.c::map_run_list(). |
| 1413 | - Tiny endianness bug fix in decompress_mapping_pairs(). |
| 1414 | |
| 1415 | tng-0.0.6 - Encrypted directories, bug fixes, cleanups, debugging enhancements. |
| 1416 | |
| 1417 | - Enable encrypted directories. (Their index root is marked encrypted |
| 1418 | to indicate that new files in that directory should be created |
| 1419 | encrypted.) |
| 1420 | - Fix bug in NInoBmpNonResident() macro. (Cut and paste error.) |
| 1421 | - Enable $Extend system directory. Most (if not all) extended system |
| 1422 | files do not have unnamed data attributes so ntfs_read_inode() had to |
| 1423 | special case them but that is ok, as the special casing recovery |
| 1424 | happens inside an error code path so there is zero slow down in the |
| 1425 | normal fast path. The special casing is done by introducing a new |
| 1426 | function inode.c::ntfs_is_extended_system_file() which checks if any |
| 1427 | of the hard links in the inode point to $Extend as being their parent |
| 1428 | directory and if they do we assume this is an extended system file. |
| 1429 | - Create a sysctl/proc interface to allow {dis,en}abling of debug output |
| 1430 | when compiled with -DDEBUG. Default is debug messages to be disabled. |
| 1431 | To enable them, one writes a non-zero value to /proc/sys/fs/ntfs-debug |
| 1432 | (if /proc is enabled) or uses sysctl(2) to effect the same (if sysctl |
| 1433 | interface is enabled). Inspired by old ntfs driver. |
| 1434 | - Add debug_msgs insmod/kernel boot parameter to set whether debug |
| 1435 | messages are {dis,en}abled. This is useful to enable debug messages |
| 1436 | during ntfs initialization and is the only way to activate debugging |
| 1437 | when the sysctl interface is not enabled. |
| 1438 | - Cleanup debug output in various places. |
| 1439 | - Remove all dollar signs ($) from the source (except comments) to |
| 1440 | enable compilation on architectures whose gcc compiler does not |
| 1441 | support dollar signs in the names of variables/constants. Attribute |
| 1442 | types now start with AT_ instead of $ and $I30 is now just I30. |
| 1443 | - Cleanup ntfs_lookup() and add consistency check of sequence numbers. |
| 1444 | - Load complete runlist for $MFT/$BITMAP during mount and cleanup |
| 1445 | access functions. This means we now cope with $MFT/$BITMAP being |
| 1446 | spread accross several mft records. |
| 1447 | - Disable modification of mft_zone_multiplier on remount. We can always |
| 1448 | reenable this later on if we really want to, but we will need to make |
| 1449 | sure we readjust the mft_zone size / layout accordingly. |
| 1450 | |
| 1451 | tng-0.0.5 - Modernize for 2.5.x and further in line-ing with Al Viro's comments. |
| 1452 | |
| 1453 | - Use sb_set_blocksize() instead of set_blocksize() and verify the |
| 1454 | return value. |
| 1455 | - Use sb_bread() instead of bread() throughout. |
| 1456 | - Add index_vcn_size{_bits} to ntfs_inode structure to store the size |
| 1457 | of a directory index block vcn. Apply resulting simplifications in |
| 1458 | dir.c everywhere. |
| 1459 | - Fix a small bug somewhere (but forgot what it was). |
| 1460 | - Change ntfs_{debug,error,warning} to enable gcc to do type checking |
| 1461 | on the printf-format parameter list and fix bugs reported by gcc |
| 1462 | as a result. (Richard Russon) |
| 1463 | - Move inode allocation strategy to Al's new stuff but maintain the |
| 1464 | divorce of ntfs_inode from struct inode. To achieve this we have two |
| 1465 | separate slab caches, one for big ntfs inodes containing a struct |
| 1466 | inode and pure ntfs inodes and at the same time fix some faulty |
| 1467 | error code paths in ntfs_read_inode(). |
| 1468 | - Show mount options in proc (inode.c::ntfs_show_options()). |
| 1469 | |
| 1470 | tng-0.0.4 - Big changes, getting in line with Al Viro's comments. |
| 1471 | |
| 1472 | - Modified (un)map_mft_record functions to be common for read and write |
| 1473 | case. To specify which is which, added extra parameter at front of |
| 1474 | parameter list. Pass either READ or WRITE to this, each has the |
| 1475 | obvious meaning. |
| 1476 | - General cleanups to allow for easier folding in vi. |
| 1477 | - attrib.c::decompress_mapping_pairs() now accepts the old runlist |
| 1478 | argument, and invokes attrib.c::merge_run_lists() to merge the old |
| 1479 | and the new runlists. |
| 1480 | - Removed attrib.c::find_first_attr(). |
| 1481 | - Implemented loading of attribute list and complete runlist for $MFT. |
| 1482 | This means we now cope with $MFT being spread across several mft |
| 1483 | records. |
| 1484 | - Adapt to 2.5.2-pre9 and the changed create_empty_buffers() syntax. |
| 1485 | - Adapt major/minor/kdev_t/[bk]devname stuff to new 2.5.x kernels. |
| 1486 | - Make ntfs_volume be allocated via kmalloc() instead of using a slab |
| 1487 | cache. There are too little ntfs_volume structures at any one time |
| 1488 | to justify a private slab cache. |
| 1489 | - Fix bogus kmap() use in async io completion. Now use kmap_atomic(). |
| 1490 | Use KM_BIO_IRQ on advice from IRC/kernel... |
| 1491 | - Use ntfs_map_page() in map_mft_record() and create ->readpage method |
| 1492 | for reading $MFT (ntfs_mft_readpage). In the process create dedicated |
| 1493 | address space operations (ntfs_mft_aops) for $MFT inode mapping. Also |
| 1494 | removed the now superfluous exports from the kernel core patch. |
| 1495 | - Fix a bug where kfree() was used insted of ntfs_free(). |
| 1496 | - Change map_mft_record() to take ntfs_inode as argument instead of |
| 1497 | vfs inode. Dito for unmap_mft_record(). Adapt all callers. |
| 1498 | - Add pointer to ntfs_volume to ntfs_inode. |
| 1499 | - Add mft record number and sequence number to ntfs_inode. Stop using |
| 1500 | i_ino and i_generation for in-driver purposes. |
| 1501 | - Implement attrib.c::merge_run_lists(). (Richard Russon) |
| 1502 | - Remove use of proper inodes by extent inodes. Move i_ino and |
| 1503 | i_generation to ntfs_inode to do this. Apply simplifications that |
| 1504 | result and remove iget_no_wait(), etc. |
| 1505 | - Pass ntfs_inode everywhere in the driver (used to be struct inode). |
| 1506 | - Add reference counting in ntfs_inode for the ntfs inode itself and |
| 1507 | for the mapped mft record. |
| 1508 | - Extend mft record mapping so we can (un)map extent mft records (new |
| 1509 | functions (un)map_extent_mft_record), and so mappings are reference |
| 1510 | counted and don't have to happen twice if already mapped - just ref |
| 1511 | count increases. |
| 1512 | - Add -o iocharset as alias to -o nls for backwards compatibility. |
| 1513 | - The latest core patch is now tiny. In fact just a single additional |
| 1514 | export is necessary over the base kernel. |
| 1515 | |
| 1516 | tng-0.0.3 - Cleanups, enhancements, bug fixes. |
| 1517 | |
| 1518 | - Work on attrib.c::decompress_mapping_pairs() to detect base extents |
| 1519 | and setup the runlist appropriately using knowledge provided by the |
| 1520 | sizes in the base attribute record. |
| 1521 | - Balance the get_/put_attr_search_ctx() calls so we don't leak memory |
| 1522 | any more. |
| 1523 | - Introduce ntfs_malloc_nofs() and ntfs_free() to allocate/free a single |
| 1524 | page or use vmalloc depending on the amount of memory requested. |
| 1525 | - Cleanup error output. The __FUNCTION__ "(): " is now added |
| 1526 | automatically. Introduced a new header file debug.h to support this |
| 1527 | and also moved ntfs_debug() function into it. |
| 1528 | - Make reading of compressed files more intelligent and especially get |
| 1529 | rid of the vmalloc_nofs() from readpage(). This now uses per CPU |
| 1530 | buffers (allocated at first mount with cluster size <= 4kiB and |
| 1531 | deallocated on last umount with cluster size <= 4kiB), and |
| 1532 | asynchronous io for the compressed data using a list of buffer heads. |
| 1533 | Er, we use synchronous io as async io only works on whole pages |
| 1534 | covered by buffers and not on individual buffer heads... |
| 1535 | - Bug fix for reading compressed files with sparse compression blocks. |
| 1536 | |
| 1537 | tng-0.0.2 - Now handles larger/fragmented/compressed volumes/files/dirs. |
| 1538 | |
| 1539 | - Fixed handling of directories when cluster size exceeds index block |
| 1540 | size. |
| 1541 | - Hide DOS only name space directory entries from readdir() but allow |
| 1542 | them in lookup(). This should fix the problem that Linux doesn't |
| 1543 | support directory hard links, while still allowing access to entries |
| 1544 | via their short file name. This also has the benefit of mimicking |
| 1545 | what Windows users are used to, so it is the ideal solution. |
| 1546 | - Implemented sync_page everywhere so no more hangs in D state when |
| 1547 | waiting for a page. |
| 1548 | - Stop using bforget() in favour of brelse(). |
| 1549 | - Stop locking buffers unnecessarily. |
| 1550 | - Implemented compressed files (inode->mapping contains uncompressed |
| 1551 | data, raw compressed data is currently bread() into a vmalloc()ed |
| 1552 | memory buffer). |
| 1553 | - Enable compressed directories. (Their index root is marked compressed |
| 1554 | to indicate that new files in that directory should be created |
| 1555 | compressed.) |
| 1556 | - Use vsnprintf rather than vsprintf in the ntfs_error and ntfs_warning |
| 1557 | functions. (Thanks to Will Dyson for pointing this out.) |
| 1558 | - Moved the ntfs_inode and ntfs_volume (the former ntfs_inode_info and |
| 1559 | ntfs_sb_info) out of the common inode and super_block structures and |
| 1560 | started using the generic_ip and generic_sbp pointers instead. This |
| 1561 | makes ntfs entirely private with respect to the kernel tree. |
| 1562 | - Detect compiler version and abort with error message if gcc less than |
| 1563 | 2.96 is used. |
| 1564 | - Fix bug in name comparison function in unistr.c. |
| 1565 | - Implement attribute lists part 1, the infrastructure: search contexts |
| 1566 | and operations, find_external_attr(), lookup_attr()) and make the |
| 1567 | code use the infrastructure. |
| 1568 | - Fix stupid buffer overflow bug that became apparent on larger run |
| 1569 | list containing attributes. |
| 1570 | - Fix bugs in readdir() that became apparent on larger directories. |
| 1571 | |
| 1572 | The driver is now really useful and survives the test |
| 1573 | find . -type f -exec md5sum "{}" \; |
| 1574 | without any error messages on a over 1GiB sized partition with >16k |
| 1575 | files on it, including compressed files and directories and many files |
| 1576 | and directories with attribute lists. |
| 1577 | |
| 1578 | tng-0.0.1 - The first useful version. |
| 1579 | |
| 1580 | - Added ntfs_lookup(). |
| 1581 | - Added default upcase generation and handling. |
| 1582 | - Added compile options to be shown on module init. |
| 1583 | - Many bug fixes that were "hidden" before. |
| 1584 | - Update to latest kernel. |
| 1585 | - Added ntfs_readdir(). |
| 1586 | - Added file operations for mmap(), read(), open() and llseek(). We just |
| 1587 | use the generic ones. The whole point of going through implementing |
| 1588 | readpage() methods and where possible get_block() call backs is that |
| 1589 | this allows us to make use of the generic high level methods provided |
| 1590 | by the kernel. |
| 1591 | |
| 1592 | The driver is now actually useful! Yey. (-: It undoubtedly has got bugs |
| 1593 | though and it doesn't implement accesssing compressed files yet. Also, |
| 1594 | accessing files with attribute list attributes is not implemented yet |
Anton Altaparmakov | c002f42 | 2005-02-03 12:02:56 +0000 | [diff] [blame] | 1595 | either. But for small or simple filesystems it should work and allow |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1596 | you to list directories, use stat on directory entries and the file |
| 1597 | system, open, read, mmap and llseek around in files. A big mile stone |
| 1598 | has been reached! |
| 1599 | |
| 1600 | tng-0.0.0 - Initial version tag. |
| 1601 | |
| 1602 | Initial driver implementation. The driver can mount and umount simple |
Anton Altaparmakov | c002f42 | 2005-02-03 12:02:56 +0000 | [diff] [blame] | 1603 | NTFS filesystems (i.e. ones without attribute lists in the system |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1604 | files). If the mount fails there might be problems in the error handling |
| 1605 | code paths, so be warned. Otherwise it seems to be loading the system |
| 1606 | files nicely and the mft record read mapping/unmapping seems to be |
| 1607 | working nicely, too. Proof of inode metadata in the page cache and non- |
| 1608 | resident file unnamed stream data in the page cache concepts is thus |
| 1609 | complete. |