1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
|
Since autofs-3.1.4:
-------------------
Since autofs-3.1.3:
-------------------
Since autofs-3.1.2:
-------------------
* Added README.options; missing since 3.1.0... oops!
Since autofs-3.1.1:
-------------------
* NULL pointer fix in mount_smbfs (from Rene H�jbjerg Larsen).
Since autofs-3.1.0:
-------------------
* The -s option (SLOPPYOPT) was put in the wrong place in mount_nfs.c!
Oops! Now fixed.
Since autofs-0.3.14:
--------------------
* "const-ipated" the source at all module interface points, to keep me
from doing stupid things like calling strtok() on a string constant.
* Minor fix to the autofs(5) man page.
* Put entire source under CVS, so added ID tags to as many files as
practical; also cleaned up comment preambles.
* AFS/Hesiod support by Nalin Dahyabhai.
* NIS+ support by Thorsten Kukuk.
* mount_ext2 module that runs e2fsck -p on the filesystem first.
* Recursive autofs mounts support by Richard Henderson. For
unmounting to work with nested mounts, kernel 2.1.93 or later is
required.
* Automatic configuration of compile options using
"configure".
* If mount(8) supports the "-s" (sloppy) option, pass it. Have
"configure" test to see if it does or not.
* Fix from David Engels for Sun map multiserver entries. (Also
requires support in mount(8).)
* When determining if an NFS filesystem is really local, don't bail
out if we get an error from connect(). Keep going to the next
address instead.
* New numbering system to declare autofs "officially no longer experimental."
Since autofs-0.3.13:
--------------------
* Simplified the signalling/forking structure.
* When debugging, leave debug-related signals in the default state.
* Fixed broken timeouts.
* Fixed multiple problems involving module destruction.
* "Shouldn't happen" signals now only get caught once, to avoid
looping.
* SIGHUP now causes a "prune" operation, followed by module
reinitialization.
Since autofs-0.3.12:
--------------------
* Added -p, --pid-file option to daemon.
* Fixed aestetic bug in new spawn() logic.
* Added new cleanup_exit() function in the daemon.
* Added signal handling to some cases that were improperly dealt with.
* Integrated new sample stuff from Miquel van Smoorenburg and
Christoph Lameter.
Since autofs-0.3.11:
--------------------
* Rewrote the spawn() logic to not use stdio at all.
Since autofs-0.3.10:
--------------------
* Added conversion from : -> / in Sun maps, to handle maps which
contain "shared mounts". Can be avoided by \-escaping the colon or
by passing the option --no-slashify-colons.
* ${HOST} corrected (was returning the value for ${ARCH}.)
* Don't free the YP domain name; the documentation says to it but the
documentation is wrong.
Since autofs-0.3.9:
-------------------
* Added YPLIBS option to Makefile.rules to support compiling for glibc.
* Distributing include/linux/auto_fs.h with the daemon one again, to
avoid version skew problems.
* Made \-escaping work (hopefully) correctly.
* Added dlerror() output to error message when dlopen() returns an error.
Since autofs-0.3.8:
-------------------
* Missed at least one place where a SIGCHLD gunked up a waitpid() in
0.3.8. Fixed.
* More documentation and init script fixes from Christoph (with some
edits from yours truly, since I'm an opininated bastard who likes
things done my way.)
* Added check to lookup_program.c: no output at all == error.
Since autofs-0.3.7:
-------------------
* Removed kernel code from distribution, except for 2.0.30 diff (will
remain until 2.0.31 is released.) Kernel code will now only be
included with the main kernel distribution, although Makefiles will
continue to support building the kernel code.
* Tried to eliminate race conditions relating to signals received in
the vicinity of a fork(), as well as multiple SIGCHLD signals
received while SIGCHLD blocked.
* Better sample rc file and man pages from Christoph Lameter of the
Debian project, with edits from hpa.
* In mount_smbfs.c: smb_parse_options() is destructive; make it
operate on a copy of the options string.
Since autofs-0.3.6:
-------------------
* Kernel code should now compile on 2.1.43 or higher.
* Minor cleanup of stdout/stderr redirect.
* Fixed this file!
Since autofs-0.3.5:
-------------------
* stdout/stderr of subprograms (only stderr for program maps, of
course) now is automatically redirected to the syslog. As a result,
the spawn[lv]() interface has changed somewhat.
* kernel: minor bug fixes.
* daemon: minor cleanup.
* Now compiles (and runs) on SPARC.
* daemon: Should now handle extremely short timeouts, or no timeouts,
correctly.
Since autofs-0.3.4:
-------------------
* kernel: *Hopefully* fixed the wait queue list corruption (kernel
NULL pointer dereference) bug.
Since autofs-0.3.3:
-------------------
* kernel: *Hopefully* fixed that annoying "filesystem busy" bug.
Since autofs-0.3.2:
-------------------
* Added "mount_generic" and "mount_smbfs" modules.
* Support ":" to escape a mount point beginning with "/" in a Sun map.
* Support -fstype= option in Sun maps.
Since autofs-0.3.1:
-------------------
* kernel: effectivized handling of waits interrupted by a signal.
* kernel: closed some extremely theoretical security holes.
* kernel: fixed two problems that could lead to kernel oopses.
Since autofs-0.3.0:
-------------------
* The kernel code should once again work on kernel 2.0.30 and 2.1.x
for x < 36.
* No need to fork an expiry process if there is nothing to expire!
* Attempting to mount a nonexistent directory should no longer leave a
pad directory around, causing renewed mount attempts on "ls".
* Timeout is now adjustible (set to 0 to turn off timeouts completely.)
Since autofs-0.2.2:
-------------------
* Support expiry of idle mounts; currently the timeout is fixed at 5
minutes.
* Ship kernel source that can be built either as a part of the autofs
or the kernel source trees.
* Ought to compile on the Alpha, although I cannot test it here.
Since autofs-0.2.1:
-------------------
* Can now read options and ignore whitespace and comments in Sun maps.
* Does closing the file descriptors after forking help avoid hung
filesystems?
Since autofs-0.2.0:
-------------------
* Cleaned up the interface between modules (not perfect yet; I was
even quite tempted to rewrite the interface using C++ classes,
but...) Do expect the interface to still change, for example:
expect a umount method to be added to the mount module. Strings
passed are now managed by the calling module, and can be allocated
in any desirable way.
* Moved mount operation to its own module.
* Handle options passed on the command line; not options passed in the map yet.
* Handle $-escapes in Sun style maps, including the Sun predefined
variables.
* New "program" map type: the map is an executable which is passed the
key on the command line, and writes the map entry to stdout on
success and exit 0, or exit 1 on failure.
Since autofs-970417:
--------------------
* Use a real versioning scheme :)
* Fixed bugs that would make file maps unmountable. Still don't quite
understand *why*, which bugs me, though...
* Man page by B. James Phillippe.
Since autofs-970409:
--------------------
* Modified the .so interface slightly, in order to make it possible
for future versions to have multiple lookup sessions.
* Added "file" map type.
Since autofs-970402:
--------------------
* The parsing module has been factored out into a separate dynamic
module. This should make writing other lookup modules much easier.
* The automount daemon now accepts kill -USR1 as a "unmount all
filesystems and exit if everything could be unmounted" and kill
-USR2 "unmount all non-busy filesystems". Prelude to handling
timeouts, I hope.
* The autofs code is now integrated in kernels 2.1.31 and above (a
patch for the 2.1.31 code is included, however; 2.1.32 is correct);
a patch is included for 2.0.30 and has been submitted to Linus for
inclusion in 2.0.31.
* include/linux/auto_fs.h is no longer shipped in the tools
distribution; the one from the kernel should be used, so you need
the (appropriately patched) kernel sources in /usr/src/linux.
* added NEWS and README files.
|