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:
35 - pipe-user-pages-hard
36 - pipe-user-pages-soft
45 ==============================================================
49 aio-nr is the running total of the number of events specified on the
50 io_setup system call for all currently active aio contexts. If aio-nr
51 reaches aio-max-nr then io_setup will fail with EAGAIN. Note that
52 raising aio-max-nr does not result in the pre-allocation or re-sizing
53 of any kernel data structures.
55 ==============================================================
59 From linux/include/linux/dcache.h:
60 --------------------------------------------------------------
61 struct dentry_stat_t dentry_stat {
64 int age_limit; /* age in seconds */
65 int want_pages; /* pages requested by system */
66 int nr_negative; /* # of unused negative dentries */
67 int dummy; /* Reserved for future use */
69 --------------------------------------------------------------
71 Dentries are dynamically allocated and deallocated.
73 nr_dentry shows the total number of dentries allocated (active
74 + unused). nr_unused shows the number of dentries that are not
75 actively used, but are saved in the LRU list for future reuse.
77 Age_limit is the age in seconds after which dcache entries
78 can be reclaimed when memory is short and want_pages is
79 nonzero when shrink_dcache_pages() has been called and the
80 dcache isn't pruned yet.
82 nr_negative shows the number of unused dentries that are also
83 negative dentries which do not mapped to actual files.
85 ==============================================================
89 The file dquot-max shows the maximum number of cached disk
92 The file dquot-nr shows the number of allocated disk quota
93 entries and the number of free disk quota entries.
95 If the number of free cached disk quotas is very low and
96 you have some awesome number of simultaneous system users,
97 you might want to raise the limit.
99 ==============================================================
103 The value in file-max denotes the maximum number of file-
104 handles that the Linux kernel will allocate. When you get lots
105 of error messages about running out of file handles, you might
106 want to increase this limit.
108 Historically,the kernel was able to allocate file handles
109 dynamically, but not to free them again. The three values in
110 file-nr denote the number of allocated file handles, the number
111 of allocated but unused file handles, and the maximum number of
112 file handles. Linux 2.6 always reports 0 as the number of free
113 file handles -- this is not an error, it just means that the
114 number of allocated file handles exactly matches the number of
117 Attempts to allocate more file descriptors than file-max are
118 reported with printk, look for "VFS: file-max limit <number>
120 ==============================================================
124 This denotes the maximum number of file-handles a process can
125 allocate. Default value is 1024*1024 (1048576) which should be
126 enough for most machines. Actual limit depends on RLIMIT_NOFILE
129 ==============================================================
131 inode-max, inode-nr & inode-state:
133 As with file handles, the kernel allocates the inode structures
134 dynamically, but can't free them yet.
136 The value in inode-max denotes the maximum number of inode
137 handlers. This value should be 3-4 times larger than the value
138 in file-max, since stdin, stdout and network sockets also
139 need an inode struct to handle them. When you regularly run
140 out of inodes, you need to increase this value.
142 The file inode-nr contains the first two items from
143 inode-state, so we'll skip to that file...
145 Inode-state contains three actual numbers and four dummies.
146 The actual numbers are, in order of appearance, nr_inodes,
147 nr_free_inodes and preshrink.
149 Nr_inodes stands for the number of inodes the system has
150 allocated, this can be slightly more than inode-max because
151 Linux allocates them one pageful at a time.
153 Nr_free_inodes represents the number of free inodes (?) and
154 preshrink is nonzero when the nr_inodes > inode-max and the
155 system needs to prune the inode list instead of allocating
158 ==============================================================
160 overflowgid & overflowuid:
162 Some filesystems only support 16-bit UIDs and GIDs, although in Linux
163 UIDs and GIDs are 32 bits. When one of these filesystems is mounted
164 with writes enabled, any UID or GID that would exceed 65535 is translated
165 to a fixed value before being written to disk.
167 These sysctls allow you to change the value of the fixed UID and GID.
168 The default is 65534.
170 ==============================================================
172 pipe-user-pages-hard:
174 Maximum total number of pages a non-privileged user may allocate for pipes.
175 Once this limit is reached, no new pipes may be allocated until usage goes
176 below the limit again. When set to 0, no limit is applied, which is the default
179 ==============================================================
181 pipe-user-pages-soft:
183 Maximum total number of pages a non-privileged user may allocate for pipes
184 before the pipe size gets limited to a single page. Once this limit is reached,
185 new pipes will be limited to a single page in size for this user in order to
186 limit total memory usage, and trying to increase them using fcntl() will be
187 denied until usage goes below the limit again. The default value allows to
188 allocate up to 1024 pipes at their default size. When set to 0, no limit is
191 ==============================================================
195 The intent of this protection is to avoid unintentional writes to
196 an attacker-controlled FIFO, where a program expected to create a regular
199 When set to "0", writing to FIFOs is unrestricted.
201 When set to "1" don't allow O_CREAT open on FIFOs that we don't own
202 in world writable sticky directories, unless they are owned by the
203 owner of the directory.
205 When set to "2" it also applies to group writable sticky directories.
207 This protection is based on the restrictions in Openwall.
209 ==============================================================
213 A long-standing class of security issues is the hardlink-based
214 time-of-check-time-of-use race, most commonly seen in world-writable
215 directories like /tmp. The common method of exploitation of this flaw
216 is to cross privilege boundaries when following a given hardlink (i.e. a
217 root process follows a hardlink created by another user). Additionally,
218 on systems without separated partitions, this stops unauthorized users
219 from "pinning" vulnerable setuid/setgid files against being upgraded by
220 the administrator, or linking to special files.
222 When set to "0", hardlink creation behavior is unrestricted.
224 When set to "1" hardlinks cannot be created by users if they do not
225 already own the source file, or do not have read/write access to it.
227 This protection is based on the restrictions in Openwall and grsecurity.
229 ==============================================================
233 This protection is similar to protected_fifos, but it
234 avoids writes to an attacker-controlled regular file, where a program
235 expected to create one.
237 When set to "0", writing to regular files is unrestricted.
239 When set to "1" don't allow O_CREAT open on regular files that we
240 don't own in world writable sticky directories, unless they are
241 owned by the owner of the directory.
243 When set to "2" it also applies to group writable sticky directories.
245 ==============================================================
249 A long-standing class of security issues is the symlink-based
250 time-of-check-time-of-use race, most commonly seen in world-writable
251 directories like /tmp. The common method of exploitation of this flaw
252 is to cross privilege boundaries when following a given symlink (i.e. a
253 root process follows a symlink belonging to another user). For a likely
254 incomplete list of hundreds of examples across the years, please see:
255 http://cve.mitre.org/cgi-bin/cvekey.cgi?keyword=/tmp
257 When set to "0", symlink following behavior is unrestricted.
259 When set to "1" symlinks are permitted to be followed only when outside
260 a sticky world-writable directory, or when the uid of the symlink and
261 follower match, or when the directory owner matches the symlink's owner.
263 This protection is based on the restrictions in Openwall and grsecurity.
265 ==============================================================
269 This value can be used to query and set the core dump mode for setuid
270 or otherwise protected/tainted binaries. The modes are
272 0 - (default) - traditional behaviour. Any process which has changed
273 privilege levels or is execute only will not be dumped.
274 1 - (debug) - all processes dump core when possible. The core dump is
275 owned by the current user and no security is applied. This is
276 intended for system debugging situations only. Ptrace is unchecked.
277 This is insecure as it allows regular users to examine the memory
278 contents of privileged processes.
279 2 - (suidsafe) - any binary which normally would not be dumped is dumped
280 anyway, but only if the "core_pattern" kernel sysctl is set to
281 either a pipe handler or a fully qualified path. (For more details
282 on this limitation, see CVE-2006-2451.) This mode is appropriate
283 when administrators are attempting to debug problems in a normal
284 environment, and either have a core dump pipe handler that knows
285 to treat privileged core dumps with care, or specific directory
286 defined for catching core dumps. If a core dump happens without
287 a pipe handler or fully qualifid path, a message will be emitted
288 to syslog warning about the lack of a correct setting.
290 ==============================================================
292 super-max & super-nr:
294 These numbers control the maximum number of superblocks, and
295 thus the maximum number of mounted filesystems the kernel
296 can have. You only need to increase super-max if you need to
297 mount more filesystems than the current value in super-max
300 ==============================================================
304 aio-nr shows the current system-wide number of asynchronous io
305 requests. aio-max-nr allows you to change the maximum value
308 ==============================================================
312 This denotes the maximum number of mounts that may exist
313 in a mount namespace.
315 ==============================================================
318 2. /proc/sys/fs/binfmt_misc
319 ----------------------------------------------------------
321 Documentation for the files in /proc/sys/fs/binfmt_misc is
322 in Documentation/admin-guide/binfmt-misc.rst.
325 3. /proc/sys/fs/mqueue - POSIX message queues filesystem
326 ----------------------------------------------------------
328 The "mqueue" filesystem provides the necessary kernel features to enable the
329 creation of a user space library that implements the POSIX message queues
330 API (as noted by the MSG tag in the POSIX 1003.1-2001 version of the System
331 Interfaces specification.)
333 The "mqueue" filesystem contains values for determining/setting the amount of
334 resources used by the file system.
336 /proc/sys/fs/mqueue/queues_max is a read/write file for setting/getting the
337 maximum number of message queues allowed on the system.
339 /proc/sys/fs/mqueue/msg_max is a read/write file for setting/getting the
340 maximum number of messages in a queue value. In fact it is the limiting value
341 for another (user) limit which is set in mq_open invocation. This attribute of
342 a queue must be less or equal then msg_max.
344 /proc/sys/fs/mqueue/msgsize_max is a read/write file for setting/getting the
345 maximum message size value (it is every message queue's attribute set during
348 /proc/sys/fs/mqueue/msg_default is a read/write file for setting/getting the
349 default number of messages in a queue value if attr parameter of mq_open(2) is
350 NULL. If it exceed msg_max, the default value is initialized msg_max.
352 /proc/sys/fs/mqueue/msgsize_default is a read/write file for setting/getting
353 the default message size value if attr parameter of mq_open(2) is NULL. If it
354 exceed msgsize_max, the default value is initialized msgsize_max.
356 4. /proc/sys/fs/epoll - Configuration options for the epoll interface
357 --------------------------------------------------------
359 This directory contains configuration options for the epoll(7) interface.
364 Every epoll file descriptor can store a number of files to be monitored
365 for event readiness. Each one of these monitored files constitutes a "watch".
366 This configuration option sets the maximum number of "watches" that are
367 allowed for each user.
368 Each "watch" costs roughly 90 bytes on a 32bit kernel, and roughly 160 bytes
370 The current default value for max_user_watches is the 1/32 of the available
371 low memory, divided for the "watch" cost in bytes.