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 map to any files. Instead,
84 they help speeding up rejection of non-existing files provided
87 ==============================================================
91 The file dquot-max shows the maximum number of cached disk
94 The file dquot-nr shows the number of allocated disk quota
95 entries and the number of free disk quota entries.
97 If the number of free cached disk quotas is very low and
98 you have some awesome number of simultaneous system users,
99 you might want to raise the limit.
101 ==============================================================
105 The value in file-max denotes the maximum number of file-
106 handles that the Linux kernel will allocate. When you get lots
107 of error messages about running out of file handles, you might
108 want to increase this limit.
110 Historically,the kernel was able to allocate file handles
111 dynamically, but not to free them again. The three values in
112 file-nr denote the number of allocated file handles, the number
113 of allocated but unused file handles, and the maximum number of
114 file handles. Linux 2.6 always reports 0 as the number of free
115 file handles -- this is not an error, it just means that the
116 number of allocated file handles exactly matches the number of
119 Attempts to allocate more file descriptors than file-max are
120 reported with printk, look for "VFS: file-max limit <number>
122 ==============================================================
126 This denotes the maximum number of file-handles a process can
127 allocate. Default value is 1024*1024 (1048576) which should be
128 enough for most machines. Actual limit depends on RLIMIT_NOFILE
131 ==============================================================
133 inode-max, inode-nr & inode-state:
135 As with file handles, the kernel allocates the inode structures
136 dynamically, but can't free them yet.
138 The value in inode-max denotes the maximum number of inode
139 handlers. This value should be 3-4 times larger than the value
140 in file-max, since stdin, stdout and network sockets also
141 need an inode struct to handle them. When you regularly run
142 out of inodes, you need to increase this value.
144 The file inode-nr contains the first two items from
145 inode-state, so we'll skip to that file...
147 Inode-state contains three actual numbers and four dummies.
148 The actual numbers are, in order of appearance, nr_inodes,
149 nr_free_inodes and preshrink.
151 Nr_inodes stands for the number of inodes the system has
152 allocated, this can be slightly more than inode-max because
153 Linux allocates them one pageful at a time.
155 Nr_free_inodes represents the number of free inodes (?) and
156 preshrink is nonzero when the nr_inodes > inode-max and the
157 system needs to prune the inode list instead of allocating
160 ==============================================================
162 overflowgid & overflowuid:
164 Some filesystems only support 16-bit UIDs and GIDs, although in Linux
165 UIDs and GIDs are 32 bits. When one of these filesystems is mounted
166 with writes enabled, any UID or GID that would exceed 65535 is translated
167 to a fixed value before being written to disk.
169 These sysctls allow you to change the value of the fixed UID and GID.
170 The default is 65534.
172 ==============================================================
174 pipe-user-pages-hard:
176 Maximum total number of pages a non-privileged user may allocate for pipes.
177 Once this limit is reached, no new pipes may be allocated until usage goes
178 below the limit again. When set to 0, no limit is applied, which is the default
181 ==============================================================
183 pipe-user-pages-soft:
185 Maximum total number of pages a non-privileged user may allocate for pipes
186 before the pipe size gets limited to a single page. Once this limit is reached,
187 new pipes will be limited to a single page in size for this user in order to
188 limit total memory usage, and trying to increase them using fcntl() will be
189 denied until usage goes below the limit again. The default value allows to
190 allocate up to 1024 pipes at their default size. When set to 0, no limit is
193 ==============================================================
197 The intent of this protection is to avoid unintentional writes to
198 an attacker-controlled FIFO, where a program expected to create a regular
201 When set to "0", writing to FIFOs is unrestricted.
203 When set to "1" don't allow O_CREAT open on FIFOs that we don't own
204 in world writable sticky directories, unless they are owned by the
205 owner of the directory.
207 When set to "2" it also applies to group writable sticky directories.
209 This protection is based on the restrictions in Openwall.
211 ==============================================================
215 A long-standing class of security issues is the hardlink-based
216 time-of-check-time-of-use race, most commonly seen in world-writable
217 directories like /tmp. The common method of exploitation of this flaw
218 is to cross privilege boundaries when following a given hardlink (i.e. a
219 root process follows a hardlink created by another user). Additionally,
220 on systems without separated partitions, this stops unauthorized users
221 from "pinning" vulnerable setuid/setgid files against being upgraded by
222 the administrator, or linking to special files.
224 When set to "0", hardlink creation behavior is unrestricted.
226 When set to "1" hardlinks cannot be created by users if they do not
227 already own the source file, or do not have read/write access to it.
229 This protection is based on the restrictions in Openwall and grsecurity.
231 ==============================================================
235 This protection is similar to protected_fifos, but it
236 avoids writes to an attacker-controlled regular file, where a program
237 expected to create one.
239 When set to "0", writing to regular files is unrestricted.
241 When set to "1" don't allow O_CREAT open on regular files that we
242 don't own in world writable sticky directories, unless they are
243 owned by the owner of the directory.
245 When set to "2" it also applies to group writable sticky directories.
247 ==============================================================
251 A long-standing class of security issues is the symlink-based
252 time-of-check-time-of-use race, most commonly seen in world-writable
253 directories like /tmp. The common method of exploitation of this flaw
254 is to cross privilege boundaries when following a given symlink (i.e. a
255 root process follows a symlink belonging to another user). For a likely
256 incomplete list of hundreds of examples across the years, please see:
257 http://cve.mitre.org/cgi-bin/cvekey.cgi?keyword=/tmp
259 When set to "0", symlink following behavior is unrestricted.
261 When set to "1" symlinks are permitted to be followed only when outside
262 a sticky world-writable directory, or when the uid of the symlink and
263 follower match, or when the directory owner matches the symlink's owner.
265 This protection is based on the restrictions in Openwall and grsecurity.
267 ==============================================================
271 This value can be used to query and set the core dump mode for setuid
272 or otherwise protected/tainted binaries. The modes are
274 0 - (default) - traditional behaviour. Any process which has changed
275 privilege levels or is execute only will not be dumped.
276 1 - (debug) - all processes dump core when possible. The core dump is
277 owned by the current user and no security is applied. This is
278 intended for system debugging situations only. Ptrace is unchecked.
279 This is insecure as it allows regular users to examine the memory
280 contents of privileged processes.
281 2 - (suidsafe) - any binary which normally would not be dumped is dumped
282 anyway, but only if the "core_pattern" kernel sysctl is set to
283 either a pipe handler or a fully qualified path. (For more details
284 on this limitation, see CVE-2006-2451.) This mode is appropriate
285 when administrators are attempting to debug problems in a normal
286 environment, and either have a core dump pipe handler that knows
287 to treat privileged core dumps with care, or specific directory
288 defined for catching core dumps. If a core dump happens without
289 a pipe handler or fully qualifid path, a message will be emitted
290 to syslog warning about the lack of a correct setting.
292 ==============================================================
294 super-max & super-nr:
296 These numbers control the maximum number of superblocks, and
297 thus the maximum number of mounted filesystems the kernel
298 can have. You only need to increase super-max if you need to
299 mount more filesystems than the current value in super-max
302 ==============================================================
306 aio-nr shows the current system-wide number of asynchronous io
307 requests. aio-max-nr allows you to change the maximum value
310 ==============================================================
314 This denotes the maximum number of mounts that may exist
315 in a mount namespace.
317 ==============================================================
320 2. /proc/sys/fs/binfmt_misc
321 ----------------------------------------------------------
323 Documentation for the files in /proc/sys/fs/binfmt_misc is
324 in Documentation/admin-guide/binfmt-misc.rst.
327 3. /proc/sys/fs/mqueue - POSIX message queues filesystem
328 ----------------------------------------------------------
330 The "mqueue" filesystem provides the necessary kernel features to enable the
331 creation of a user space library that implements the POSIX message queues
332 API (as noted by the MSG tag in the POSIX 1003.1-2001 version of the System
333 Interfaces specification.)
335 The "mqueue" filesystem contains values for determining/setting the amount of
336 resources used by the file system.
338 /proc/sys/fs/mqueue/queues_max is a read/write file for setting/getting the
339 maximum number of message queues allowed on the system.
341 /proc/sys/fs/mqueue/msg_max is a read/write file for setting/getting the
342 maximum number of messages in a queue value. In fact it is the limiting value
343 for another (user) limit which is set in mq_open invocation. This attribute of
344 a queue must be less or equal then msg_max.
346 /proc/sys/fs/mqueue/msgsize_max is a read/write file for setting/getting the
347 maximum message size value (it is every message queue's attribute set during
350 /proc/sys/fs/mqueue/msg_default is a read/write file for setting/getting the
351 default number of messages in a queue value if attr parameter of mq_open(2) is
352 NULL. If it exceed msg_max, the default value is initialized msg_max.
354 /proc/sys/fs/mqueue/msgsize_default is a read/write file for setting/getting
355 the default message size value if attr parameter of mq_open(2) is NULL. If it
356 exceed msgsize_max, the default value is initialized msgsize_max.
358 4. /proc/sys/fs/epoll - Configuration options for the epoll interface
359 --------------------------------------------------------
361 This directory contains configuration options for the epoll(7) interface.
366 Every epoll file descriptor can store a number of files to be monitored
367 for event readiness. Each one of these monitored files constitutes a "watch".
368 This configuration option sets the maximum number of "watches" that are
369 allowed for each user.
370 Each "watch" costs roughly 90 bytes on a 32bit kernel, and roughly 160 bytes
372 The current default value for max_user_watches is the 1/32 of the available
373 low memory, divided for the "watch" cost in bytes.