1===============================
2Documentation for /proc/sys/fs/
3===============================
4
5Copyright (c) 1998, 1999,  Rik van Riel <riel@nl.linux.org>
6
7Copyright (c) 2009,        Shen Feng<shen@cn.fujitsu.com>
8
9For general info and legal blurb, please look in intro.rst.
10
11------------------------------------------------------------------------------
12
13This file contains documentation for the sysctl files and directories
14in ``/proc/sys/fs/``.
15
16The files in this directory can be used to tune and monitor
17miscellaneous and general things in the operation of the Linux
18kernel. Since some of the files *can* be used to screw up your
19system, it is advisable to read both documentation and source
20before actually making adjustments.
21
221. /proc/sys/fs
23===============
24
25Currently, these files might (depending on your configuration)
26show up in ``/proc/sys/fs``:
27
28.. contents:: :local:
29
30
31aio-nr & aio-max-nr
32-------------------
33
34``aio-nr`` shows the current system-wide number of asynchronous io
35requests.  ``aio-max-nr`` allows you to change the maximum value
36``aio-nr`` can grow to.  If ``aio-nr`` reaches ``aio-nr-max`` then
37``io_setup`` will fail with ``EAGAIN``.  Note that raising
38``aio-max-nr`` does not result in the
39pre-allocation or re-sizing of any kernel data structures.
40
41
42dentry-state
43------------
44
45This file shows the values in ``struct dentry_stat``, as defined in
46``linux/include/linux/dcache.h``::
47
48  struct dentry_stat_t dentry_stat {
49        int nr_dentry;
50        int nr_unused;
51        int age_limit;         /* age in seconds */
52        int want_pages;        /* pages requested by system */
53        int nr_negative;       /* # of unused negative dentries */
54        int dummy;             /* Reserved for future use */
55  };
56
57Dentries are dynamically allocated and deallocated.
58
59``nr_dentry`` shows the total number of dentries allocated (active
60+ unused). ``nr_unused shows`` the number of dentries that are not
61actively used, but are saved in the LRU list for future reuse.
62
63``age_limit`` is the age in seconds after which dcache entries
64can be reclaimed when memory is short and ``want_pages`` is
65nonzero when ``shrink_dcache_pages()`` has been called and the
66dcache isn't pruned yet.
67
68``nr_negative`` shows the number of unused dentries that are also
69negative dentries which do not map to any files. Instead,
70they help speeding up rejection of non-existing files provided
71by the users.
72
73
74file-max & file-nr
75------------------
76
77The value in ``file-max`` denotes the maximum number of file-
78handles that the Linux kernel will allocate. When you get lots
79of error messages about running out of file handles, you might
80want to increase this limit.
81
82Historically,the kernel was able to allocate file handles
83dynamically, but not to free them again. The three values in
84``file-nr`` denote the number of allocated file handles, the number
85of allocated but unused file handles, and the maximum number of
86file handles. Linux 2.6 and later always reports 0 as the number of free
87file handles -- this is not an error, it just means that the
88number of allocated file handles exactly matches the number of
89used file handles.
90
91Attempts to allocate more file descriptors than ``file-max`` are
92reported with ``printk``, look for::
93
94  VFS: file-max limit <number> reached
95
96in the kernel logs.
97
98
99inode-nr & inode-state
100----------------------
101
102As with file handles, the kernel allocates the inode structures
103dynamically, but can't free them yet.
104
105The file ``inode-nr`` contains the first two items from
106``inode-state``, so we'll skip to that file...
107
108``inode-state`` contains three actual numbers and four dummies.
109The actual numbers are, in order of appearance, ``nr_inodes``,
110``nr_free_inodes`` and ``preshrink``.
111
112``nr_inodes`` stands for the number of inodes the system has
113allocated.
114
115``nr_free_inodes`` represents the number of free inodes (?) and
116preshrink is nonzero when the
117system needs to prune the inode list instead of allocating
118more.
119
120
121mount-max
122---------
123
124This denotes the maximum number of mounts that may exist
125in a mount namespace.
126
127
128nr_open
129-------
130
131This denotes the maximum number of file-handles a process can
132allocate. Default value is 1024*1024 (1048576) which should be
133enough for most machines. Actual limit depends on ``RLIMIT_NOFILE``
134resource limit.
135
136
137overflowgid & overflowuid
138-------------------------
139
140Some filesystems only support 16-bit UIDs and GIDs, although in Linux
141UIDs and GIDs are 32 bits. When one of these filesystems is mounted
142with writes enabled, any UID or GID that would exceed 65535 is translated
143to a fixed value before being written to disk.
144
145These sysctls allow you to change the value of the fixed UID and GID.
146The default is 65534.
147
148
149pipe-user-pages-hard
150--------------------
151
152Maximum total number of pages a non-privileged user may allocate for pipes.
153Once this limit is reached, no new pipes may be allocated until usage goes
154below the limit again. When set to 0, no limit is applied, which is the default
155setting.
156
157
158pipe-user-pages-soft
159--------------------
160
161Maximum total number of pages a non-privileged user may allocate for pipes
162before the pipe size gets limited to a single page. Once this limit is reached,
163new pipes will be limited to a single page in size for this user in order to
164limit total memory usage, and trying to increase them using ``fcntl()`` will be
165denied until usage goes below the limit again. The default value allows to
166allocate up to 1024 pipes at their default size. When set to 0, no limit is
167applied.
168
169
170protected_fifos
171---------------
172
173The intent of this protection is to avoid unintentional writes to
174an attacker-controlled FIFO, where a program expected to create a regular
175file.
176
177When set to "0", writing to FIFOs is unrestricted.
178
179When set to "1" don't allow ``O_CREAT`` open on FIFOs that we don't own
180in world writable sticky directories, unless they are owned by the
181owner of the directory.
182
183When set to "2" it also applies to group writable sticky directories.
184
185This protection is based on the restrictions in Openwall.
186
187
188protected_hardlinks
189--------------------
190
191A long-standing class of security issues is the hardlink-based
192time-of-check-time-of-use race, most commonly seen in world-writable
193directories like ``/tmp``. The common method of exploitation of this flaw
194is to cross privilege boundaries when following a given hardlink (i.e. a
195root process follows a hardlink created by another user). Additionally,
196on systems without separated partitions, this stops unauthorized users
197from "pinning" vulnerable setuid/setgid files against being upgraded by
198the administrator, or linking to special files.
199
200When set to "0", hardlink creation behavior is unrestricted.
201
202When set to "1" hardlinks cannot be created by users if they do not
203already own the source file, or do not have read/write access to it.
204
205This protection is based on the restrictions in Openwall and grsecurity.
206
207
208protected_regular
209-----------------
210
211This protection is similar to `protected_fifos`_, but it
212avoids writes to an attacker-controlled regular file, where a program
213expected to create one.
214
215When set to "0", writing to regular files is unrestricted.
216
217When set to "1" don't allow ``O_CREAT`` open on regular files that we
218don't own in world writable sticky directories, unless they are
219owned by the owner of the directory.
220
221When set to "2" it also applies to group writable sticky directories.
222
223
224protected_symlinks
225------------------
226
227A long-standing class of security issues is the symlink-based
228time-of-check-time-of-use race, most commonly seen in world-writable
229directories like ``/tmp``. The common method of exploitation of this flaw
230is to cross privilege boundaries when following a given symlink (i.e. a
231root process follows a symlink belonging to another user). For a likely
232incomplete list of hundreds of examples across the years, please see:
233https://cve.mitre.org/cgi-bin/cvekey.cgi?keyword=/tmp
234
235When set to "0", symlink following behavior is unrestricted.
236
237When set to "1" symlinks are permitted to be followed only when outside
238a sticky world-writable directory, or when the uid of the symlink and
239follower match, or when the directory owner matches the symlink's owner.
240
241This protection is based on the restrictions in Openwall and grsecurity.
242
243
244suid_dumpable
245-------------
246
247This value can be used to query and set the core dump mode for setuid
248or otherwise protected/tainted binaries. The modes are
249
250=   ==========  ===============================================================
2510   (default)	Traditional behaviour. Any process which has changed
252		privilege levels or is execute only will not be dumped.
2531   (debug)	All processes dump core when possible. The core dump is
254		owned by the current user and no security is applied. This is
255		intended for system debugging situations only.
256		Ptrace is unchecked.
257		This is insecure as it allows regular users to examine the
258		memory contents of privileged processes.
2592   (suidsafe)	Any binary which normally would not be dumped is dumped
260		anyway, but only if the ``core_pattern`` kernel sysctl (see
261		:ref:`Documentation/admin-guide/sysctl/kernel.rst <core_pattern>`)
262		is set to
263		either a pipe handler or a fully qualified path. (For more
264		details on this limitation, see CVE-2006-2451.) This mode is
265		appropriate when administrators are attempting to debug
266		problems in a normal environment, and either have a core dump
267		pipe handler that knows to treat privileged core dumps with
268		care, or specific directory defined for catching core dumps.
269		If a core dump happens without a pipe handler or fully
270		qualified path, a message will be emitted to syslog warning
271		about the lack of a correct setting.
272=   ==========  ===============================================================
273
274
275
2762. /proc/sys/fs/binfmt_misc
277===========================
278
279Documentation for the files in ``/proc/sys/fs/binfmt_misc`` is
280in Documentation/admin-guide/binfmt-misc.rst.
281
282
2833. /proc/sys/fs/mqueue - POSIX message queues filesystem
284========================================================
285
286
287The "mqueue"  filesystem provides  the necessary kernel features to enable the
288creation of a  user space  library that  implements  the  POSIX message queues
289API (as noted by the  MSG tag in the  POSIX 1003.1-2001 version  of the System
290Interfaces specification.)
291
292The "mqueue" filesystem contains values for determining/setting the
293amount of resources used by the file system.
294
295``/proc/sys/fs/mqueue/queues_max`` is a read/write file for
296setting/getting the maximum number of message queues allowed on the
297system.
298
299``/proc/sys/fs/mqueue/msg_max`` is a read/write file for
300setting/getting the maximum number of messages in a queue value.  In
301fact it is the limiting value for another (user) limit which is set in
302``mq_open`` invocation.  This attribute of a queue must be less than
303or equal to ``msg_max``.
304
305``/proc/sys/fs/mqueue/msgsize_max`` is a read/write file for
306setting/getting the maximum message size value (it is an attribute of
307every message queue, set during its creation).
308
309``/proc/sys/fs/mqueue/msg_default`` is a read/write file for
310setting/getting the default number of messages in a queue value if the
311``attr`` parameter of ``mq_open(2)`` is ``NULL``. If it exceeds
312``msg_max``, the default value is initialized to ``msg_max``.
313
314``/proc/sys/fs/mqueue/msgsize_default`` is a read/write file for
315setting/getting the default message size value if the ``attr``
316parameter of ``mq_open(2)`` is ``NULL``. If it exceeds
317``msgsize_max``, the default value is initialized to ``msgsize_max``.
318
3194. /proc/sys/fs/epoll - Configuration options for the epoll interface
320=====================================================================
321
322This directory contains configuration options for the epoll(7) interface.
323
324max_user_watches
325----------------
326
327Every epoll file descriptor can store a number of files to be monitored
328for event readiness. Each one of these monitored files constitutes a "watch".
329This configuration option sets the maximum number of "watches" that are
330allowed for each user.
331Each "watch" costs roughly 90 bytes on a 32-bit kernel, and roughly 160 bytes
332on a 64-bit one.
333The current default value for ``max_user_watches`` is 4% of the
334available low memory, divided by the "watch" cost in bytes.
335