1 Documentation for /proc/sys/fs/* kernel version 2.2.10
2 (c) 1998, 1999, Rik van Riel <riel@nl.linux.org>
3 (c) 2009, Shen Feng<shen@cn.fujitsu.com>
5 For general info and legal blurb, please look in README.
7 ==============================================================
9 This file contains documentation for the sysctl files in
10 /proc/sys/fs/ and is valid for Linux kernel version 2.2.
12 The files in this directory can be used to tune and monitor
13 miscellaneous and general things in the operation of the Linux
14 kernel. Since some of the files _can_ be used to screw up your
15 system, it is advisable to read both documentation and source
16 before actually making adjustments.
19 ----------------------------------------------------------
21 Currently, these files are in /proc/sys/fs:
41 ==============================================================
45 aio-nr is the running total of the number of events specified on the
46 io_setup system call for all currently active aio contexts. If aio-nr
47 reaches aio-max-nr then io_setup will fail with EAGAIN. Note that
48 raising aio-max-nr does not result in the pre-allocation or re-sizing
49 of any kernel data structures.
51 ==============================================================
55 From linux/fs/dentry.c:
56 --------------------------------------------------------------
60 int age_limit; /* age in seconds */
61 int want_pages; /* pages requested by system */
63 } dentry_stat = {0, 0, 45, 0,};
64 --------------------------------------------------------------
66 Dentries are dynamically allocated and deallocated, and
67 nr_dentry seems to be 0 all the time. Hence it's safe to
68 assume that only nr_unused, age_limit and want_pages are
69 used. Nr_unused seems to be exactly what its name says.
70 Age_limit is the age in seconds after which dcache entries
71 can be reclaimed when memory is short and want_pages is
72 nonzero when shrink_dcache_pages() has been called and the
73 dcache isn't pruned yet.
75 ==============================================================
79 The file dquot-max shows the maximum number of cached disk
82 The file dquot-nr shows the number of allocated disk quota
83 entries and the number of free disk quota entries.
85 If the number of free cached disk quotas is very low and
86 you have some awesome number of simultaneous system users,
87 you might want to raise the limit.
89 ==============================================================
93 The value in file-max denotes the maximum number of file-
94 handles that the Linux kernel will allocate. When you get lots
95 of error messages about running out of file handles, you might
96 want to increase this limit.
98 Historically,the kernel was able to allocate file handles
99 dynamically, but not to free them again. The three values in
100 file-nr denote the number of allocated file handles, the number
101 of allocated but unused file handles, and the maximum number of
102 file handles. Linux 2.6 always reports 0 as the number of free
103 file handles -- this is not an error, it just means that the
104 number of allocated file handles exactly matches the number of
107 Attempts to allocate more file descriptors than file-max are
108 reported with printk, look for "VFS: file-max limit <number>
110 ==============================================================
114 This denotes the maximum number of file-handles a process can
115 allocate. Default value is 1024*1024 (1048576) which should be
116 enough for most machines. Actual limit depends on RLIMIT_NOFILE
119 ==============================================================
121 inode-max, inode-nr & inode-state:
123 As with file handles, the kernel allocates the inode structures
124 dynamically, but can't free them yet.
126 The value in inode-max denotes the maximum number of inode
127 handlers. This value should be 3-4 times larger than the value
128 in file-max, since stdin, stdout and network sockets also
129 need an inode struct to handle them. When you regularly run
130 out of inodes, you need to increase this value.
132 The file inode-nr contains the first two items from
133 inode-state, so we'll skip to that file...
135 Inode-state contains three actual numbers and four dummies.
136 The actual numbers are, in order of appearance, nr_inodes,
137 nr_free_inodes and preshrink.
139 Nr_inodes stands for the number of inodes the system has
140 allocated, this can be slightly more than inode-max because
141 Linux allocates them one pageful at a time.
143 Nr_free_inodes represents the number of free inodes (?) and
144 preshrink is nonzero when the nr_inodes > inode-max and the
145 system needs to prune the inode list instead of allocating
148 ==============================================================
150 overflowgid & overflowuid:
152 Some filesystems only support 16-bit UIDs and GIDs, although in Linux
153 UIDs and GIDs are 32 bits. When one of these filesystems is mounted
154 with writes enabled, any UID or GID that would exceed 65535 is translated
155 to a fixed value before being written to disk.
157 These sysctls allow you to change the value of the fixed UID and GID.
158 The default is 65534.
160 ==============================================================
164 A long-standing class of security issues is the hardlink-based
165 time-of-check-time-of-use race, most commonly seen in world-writable
166 directories like /tmp. The common method of exploitation of this flaw
167 is to cross privilege boundaries when following a given hardlink (i.e. a
168 root process follows a hardlink created by another user). Additionally,
169 on systems without separated partitions, this stops unauthorized users
170 from "pinning" vulnerable setuid/setgid files against being upgraded by
171 the administrator, or linking to special files.
173 When set to "0", hardlink creation behavior is unrestricted.
175 When set to "1" hardlinks cannot be created by users if they do not
176 already own the source file, or do not have read/write access to it.
178 This protection is based on the restrictions in Openwall and grsecurity.
180 ==============================================================
184 A long-standing class of security issues is the symlink-based
185 time-of-check-time-of-use race, most commonly seen in world-writable
186 directories like /tmp. The common method of exploitation of this flaw
187 is to cross privilege boundaries when following a given symlink (i.e. a
188 root process follows a symlink belonging to another user). For a likely
189 incomplete list of hundreds of examples across the years, please see:
190 http://cve.mitre.org/cgi-bin/cvekey.cgi?keyword=/tmp
192 When set to "0", symlink following behavior is unrestricted.
194 When set to "1" symlinks are permitted to be followed only when outside
195 a sticky world-writable directory, or when the uid of the symlink and
196 follower match, or when the directory owner matches the symlink's owner.
198 This protection is based on the restrictions in Openwall and grsecurity.
200 ==============================================================
204 This value can be used to query and set the core dump mode for setuid
205 or otherwise protected/tainted binaries. The modes are
207 0 - (default) - traditional behaviour. Any process which has changed
208 privilege levels or is execute only will not be dumped.
209 1 - (debug) - all processes dump core when possible. The core dump is
210 owned by the current user and no security is applied. This is
211 intended for system debugging situations only. Ptrace is unchecked.
212 This is insecure as it allows regular users to examine the memory
213 contents of privileged processes.
214 2 - (suidsafe) - any binary which normally would not be dumped is dumped
215 anyway, but only if the "core_pattern" kernel sysctl is set to
216 either a pipe handler or a fully qualified path. (For more details
217 on this limitation, see CVE-2006-2451.) This mode is appropriate
218 when administrators are attempting to debug problems in a normal
219 environment, and either have a core dump pipe handler that knows
220 to treat privileged core dumps with care, or specific directory
221 defined for catching core dumps. If a core dump happens without
222 a pipe handler or fully qualifid path, a message will be emitted
223 to syslog warning about the lack of a correct setting.
225 ==============================================================
227 super-max & super-nr:
229 These numbers control the maximum number of superblocks, and
230 thus the maximum number of mounted filesystems the kernel
231 can have. You only need to increase super-max if you need to
232 mount more filesystems than the current value in super-max
235 ==============================================================
239 aio-nr shows the current system-wide number of asynchronous io
240 requests. aio-max-nr allows you to change the maximum value
243 ==============================================================
246 2. /proc/sys/fs/binfmt_misc
247 ----------------------------------------------------------
249 Documentation for the files in /proc/sys/fs/binfmt_misc is
250 in Documentation/binfmt_misc.txt.
253 3. /proc/sys/fs/mqueue - POSIX message queues filesystem
254 ----------------------------------------------------------
256 The "mqueue" filesystem provides the necessary kernel features to enable the
257 creation of a user space library that implements the POSIX message queues
258 API (as noted by the MSG tag in the POSIX 1003.1-2001 version of the System
259 Interfaces specification.)
261 The "mqueue" filesystem contains values for determining/setting the amount of
262 resources used by the file system.
264 /proc/sys/fs/mqueue/queues_max is a read/write file for setting/getting the
265 maximum number of message queues allowed on the system.
267 /proc/sys/fs/mqueue/msg_max is a read/write file for setting/getting the
268 maximum number of messages in a queue value. In fact it is the limiting value
269 for another (user) limit which is set in mq_open invocation. This attribute of
270 a queue must be less or equal then msg_max.
272 /proc/sys/fs/mqueue/msgsize_max is a read/write file for setting/getting the
273 maximum message size value (it is every message queue's attribute set during
276 /proc/sys/fs/mqueue/msg_default is a read/write file for setting/getting the
277 default number of messages in a queue value if attr parameter of mq_open(2) is
278 NULL. If it exceed msg_max, the default value is initialized msg_max.
280 /proc/sys/fs/mqueue/msgsize_default is a read/write file for setting/getting
281 the default message size value if attr parameter of mq_open(2) is NULL. If it
282 exceed msgsize_max, the default value is initialized msgsize_max.
284 4. /proc/sys/fs/epoll - Configuration options for the epoll interface
285 --------------------------------------------------------
287 This directory contains configuration options for the epoll(7) interface.
292 Every epoll file descriptor can store a number of files to be monitored
293 for event readiness. Each one of these monitored files constitutes a "watch".
294 This configuration option sets the maximum number of "watches" that are
295 allowed for each user.
296 Each "watch" costs roughly 90 bytes on a 32bit kernel, and roughly 160 bytes
298 The current default value for max_user_watches is the 1/32 of the available
299 low memory, divided for the "watch" cost in bytes.