Root/fs/mini_fo/ChangeLog

12006-01-24 Markus Klotzbuecher <mk@mary.denx.de>
2
3    * Add tons of ugly ifdefs to Ed L. Cashin's mutex patch to
4          retain backwards compatibility.
5
62006-01-24 Ed L. Cashin <ecashin@coraid.com>
7
8    * Support for the new mutex infrastructure
9    (7892f2f48d165a34b0b8130c8a195dfd807b8cb6)
10
112005-10-15 Markus Klotzbuecher <mk@localhost.localdomain>
12
13    * Bugfix for a serious memory leak in mini_fo_follow_link.
14
152005-09-21 Markus Klotzbuecher <mk@mary>
16
17    * new release 0.6.1
18
19    * fix of a compiler warning due to changes in 2.6.13
20
212005-09-21 Klaus Wenninger <klaus.wenninger@siemens.com>
22
23    * file.c: readdir: fix for a bug that caused directory entries
24          to show up twice when using storage filesystems such as
25          minixfs or pramfs.
26
272005-06-30 Eric Lammerts <eric@lammerts.org>
28
29    * fix for an oops when overwriting a binary thats beeing
30          executed.
31
322005-06-09 <mk@mary>
33
34    * Renamed overlay to mini_fo-overlay.
35
36    * Added mini_fo-merge script to allow merging of storage and base
37    after making modifications.
38
392005-05-22 root <mk@mary>
40
41    * Added overlay script that allows to easily mount mini_fo ontop
42    of a given base directory
43
442005-05-10 <mk@mary>
45
46    * inode.c: xattr functions return -EOPNOSUPP instead of
47          -ENOSUPP, what confuses "ls -l"
48
49    * Changed license from LGPL to GPL.
50
512005-05-08 root <mk@mary>
52
53    * Makefile: clean it up and added make install and make
54          uninstall.
55
562005-05-06 <mk@mary>
57
58    * merged devel branch back to main. [v0-6-0-pre3]
59
60    * removed unused files print.c and fist_ioctl. [devel-0-0-18]
61
62    * ioctl: removed fist_ioctl stuff, that is not needed for
63          now.
64
652005-05-03 <mk@mary>
66
67    * file.c: simplified mini_fo_open and mini_fo_setattr using
68          new state changing functions. [devel-0-0-17]
69
70    * inode.c: Fixed getattr state bug (see below) in 2.4 function
71          mini_fo_inode revalidate.
72
73    * inode.c: found an other bug in mini_fo_getattr. States are not
74      reliable in this function, as a file can be opened, unlinked and
75      the getattr function called. This results in a deleted dentry
76      with an inode. Fix is to ignore states and simply use the inode
77      available.
78
792005-04-29 <mk@mary>
80
81    * file.c: Bugfix and cleanup in fasync and fsync. [devel-0-0-16]
82
83    * file.c: do not use mini_fo_lock so the generic version is
84          used (I guess).
85
86    * inode.c: getattr, never call getattr on lower files, as this
87          will cause the inum to change.
88
89    * inode.c: rename_reg_file renamed to rename_nondir, as it
90          doesn't matter as long it't not a dir. Removed all
91          rename_dev_file etc.
92
93    * tagged as devel-0-0-15
94
95    * inode.c: added support for chosing support for extended
96          attrs at compile time by XATTR define in mini_fo.h .
97
98    * inode.c: fixed mini_fo_getattr to use mini_fo inode and not
99          lower again, what avoids inode number changes that confused
100          rm again. This is the proper solution.
101
1022005-04-24 <mk@mary>
103
104    * all files: updated Copyright notive to 2005. [devel-0-0-14]
105
106    * inode.c: fixed mini_fo_getattr to not change the inode
107          number, even if lower files change.
108
109    * super.c: fixed a bug that caused deleted base file to show
110          up suddenly after some time, or after creating a special
111          file. The problem was that after some time or after special
112          file creating sync_sb_inodes is called by the vfs, that
113          called our mini_fo_put_inode. There was (wrongly) called
114          __meta_put_lists, that nuked the lists, although the inode
115          was going to continue its life. Moving __meta_put_lists to
116          mini_fo_clear_inode, where an inode is really destroyed,
117          solved the problem.
118
119
1202005-04-23 <mk@mary>
121
122    * state.c, aux.c: more cleaning up and
123          simplifications. [devel-0-0-13]
124
125    * inode.c: implemented mini_fo_getattr, that was required for
126          2.6 because inode_revalidate has been remove there, and the
127      old "du" bug returned.
128
129
1302005-04-20 <mk@mary>
131
132    * aux.c: get_neg_sto_dentry(): allow to be called for dentries
133          in state UNMODIFIED, NON_EXISTANT _and_ DELETED.
134
1352005-04-19 <mk@mary>
136
137    * Fixed a bug under 2.6 that caused files deleted via mini_fo
138          not to be deleted properly and therefore the fs filled up
139          untill no memory was left. [devel-0-0-12]
140
141    * Added basic hard link support. This means that creating
142          hardlinks will work, but existing ones will be treated as
143          individual files. [devel-0-0-11]
144
1452005-04-17 <mk@mary>
146
147    * Bugfixes
148
1492005-04-13 root <mk@mary>
150
151    * Added file state.c for the state transition
152          functions. Doesn't work very well yet, though...
153
1542005-04-12 <mk@mary>
155
156    * Porting to 2.6 started, which is easier than expected, also
157          due to Olivier previous work.
158
1592005-04-08 <mk@mary>
160
161    * Fixed the bug that caused du to return invalid sizes of
162          directory trees. The problem was that
163          mini_fo_inode_revalidate didn't always copy the attributes
164          from the base inode properly.
165
1662005-04-01 Markus Klotzbuecher <mk@chasey>
167
168    * Merged devel branch back to main trunk and updated the
169          RELEASE notes. This will be 0-6-0-pre1.
170
1712005-03-31 Markus Klotzbuecher <mk@chasey>
172
173    * Fixed some bugs in rename_reg_file, that only showed up in
174          the kernel compile test. Kernel compiles cleanly ontop of
175          mini_fo, now also make mrproper etc. work. Seems pretty stable.
176
1772005-03-28 Markus Klotzbuecher <mk@chasey>
178
179    * Many, many directory renaming bugfixes and a lot of other
180          cleanup. Dir renaming seems to work relatively stable.
181
1822005-03-22 Markus Klotzbuecher <mk@chasey>
183
184    * Finished implementing lightweight directory renaming. Some
185          basic testing indicates it works fine.
186      Next is to implement testcases for the testsuite and confirm
187          everything is really working ok.
188
1892005-03-18 Markus Klotzbuecher <mk@chasey>
190
191    * Finished implementing meta.c stuff required for directory
192          renaming.
193
1942005-03-17 Markus Klotzbuecher <mk@chasey>
195
196    * Fixed all compile warnings + an extremly old bug that
197          somehow crept in while reworking the wol stuff to the META
198          system. Turning on -Werror again... :-)
199
200    * Fixed some bugs in the new rename_reg_file function.
201
202    * Rewrote mini_fo rename and split it into several
203          subfunctions, that handle the different types
204          seperately. Rewrote the regular file function aswell, as it
205          was implemented somewhat inefficient.
206
2072005-03-16 Markus Klotzbuecher <mk@chasey>
208
209    * Implemented new META subsystem, removed old WOL stuff in favor
210      if it.
211
212    * After some basic testing everything seems ok...
213
2142005-03-11 Markus Klotzbuecher <mk@chasey>
215
216    * Renaming a non regular file caused trouble because I always
217      tried to copy the contents. Now I only do this for regular
218      files. mini_fo_rename still isn't implemented properly, renaming
219      of device files, symlinks etc. results in a empty regular file
220      instead of the proper type.
221
222    * Directory renaming suddenly works! What a surprise! I guess
223          this is because renaming is implemented as making a copy and
224          removing the original. Still this might not work
225          everywhere...
226
2272005-03-09 Markus Klotzbuecher <mk@chasey>
228
229    * Bugfix, when a mini_fo directory that exists in storage
230        (state: MODIFIED, CREATED and DEL_REWRITTEN) is deleted, a
231        possibly existing WOL file contained in it needs to be
232        deleted too.
233
234    * Starting cleanup: defined state names in order to get rid of
235          the state numbers.
236
2372005-03-08 Markus Klotzbuecher <mk@chasey>
238
239    * Makefile fix, fist_ioctl was built against wrong sources if ARCH=um
240
241    * Fixed a bug in dentry.c, mini_fo_d_hash. In state 4 =
242          DEL_REWRITTEN the hash was calculated from the base dentry,
243          which was wrong and and caused assertions in
244          __mini_fo_hidden_dentry to fail.
245
2462005-02-21 <mk@mary>
247
248    * Implemented directory deleting (inode.c)
249
250    * main.c: made mini_fo_parse_options a little more robust.
251
2522004-12-22 <mk@mary>
253
254    * Makefile cleanup and uml stuff, removed unneccessary files
255
256    * Created a new and hopefully more informative README
257
258    * CHANGELOG: created a new CHANGELOG and added old entries reversely
259
260
2612004-10-24 Gleb Natapov <gleb@nbase.co.il>
262
263    * Fix: owner and group where not correctly copied from base to
264          storage.
265
266
2672004-10-05 Gleb Natapov <gleb@nbase.co.il>
268
269    * Implementation of fsync, fasync and lock mini_fo functions.
270
271
2722004-09-29 Bob Lee <bob@pantasys.com>
273
274    * Fix of a serious pointer bug
275
276
2772004-09-28 Gleb Natapov <gleb@nbase.co.il>
278
279    * Implementation of mini_fo_mknod and mini_fo_rename, support
280          for device files.
281
282

Archive Download this file



interactive