tjh.dev
Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
1Documentation 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>
4
5For general info and legal blurb, please look in README.
6
7==============================================================
8
9This file contains documentation for the sysctl files in
10/proc/sys/fs/ and is valid for Linux kernel version 2.2.
11
12The files in this directory can be used to tune and monitor
13miscellaneous and general things in the operation of the Linux
14kernel. Since some of the files _can_ be used to screw up your
15system, it is advisable to read both documentation and source
16before actually making adjustments.
17
181. /proc/sys/fs
19----------------------------------------------------------
20
21Currently, these files are in /proc/sys/fs:
22- aio-max-nr
23- aio-nr
24- dentry-state
25- dquot-max
26- dquot-nr
27- file-max
28- file-nr
29- inode-max
30- inode-nr
31- inode-state
32- nr_open
33- overflowuid
34- overflowgid
35- suid_dumpable
36- super-max
37- super-nr
38
39==============================================================
40
41aio-nr & aio-max-nr:
42
43aio-nr is the running total of the number of events specified on the
44io_setup system call for all currently active aio contexts. If aio-nr
45reaches aio-max-nr then io_setup will fail with EAGAIN. Note that
46raising aio-max-nr does not result in the pre-allocation or re-sizing
47of any kernel data structures.
48
49==============================================================
50
51dentry-state:
52
53From linux/fs/dentry.c:
54--------------------------------------------------------------
55struct {
56 int nr_dentry;
57 int nr_unused;
58 int age_limit; /* age in seconds */
59 int want_pages; /* pages requested by system */
60 int dummy[2];
61} dentry_stat = {0, 0, 45, 0,};
62--------------------------------------------------------------
63
64Dentries are dynamically allocated and deallocated, and
65nr_dentry seems to be 0 all the time. Hence it's safe to
66assume that only nr_unused, age_limit and want_pages are
67used. Nr_unused seems to be exactly what its name says.
68Age_limit is the age in seconds after which dcache entries
69can be reclaimed when memory is short and want_pages is
70nonzero when shrink_dcache_pages() has been called and the
71dcache isn't pruned yet.
72
73==============================================================
74
75dquot-max & dquot-nr:
76
77The file dquot-max shows the maximum number of cached disk
78quota entries.
79
80The file dquot-nr shows the number of allocated disk quota
81entries and the number of free disk quota entries.
82
83If the number of free cached disk quotas is very low and
84you have some awesome number of simultaneous system users,
85you might want to raise the limit.
86
87==============================================================
88
89file-max & file-nr:
90
91The kernel allocates file handles dynamically, but as yet it
92doesn't free them again.
93
94The value in file-max denotes the maximum number of file-
95handles that the Linux kernel will allocate. When you get lots
96of error messages about running out of file handles, you might
97want to increase this limit.
98
99Historically, the three values in file-nr denoted the number of
100allocated file handles, the number of allocated but unused file
101handles, and the maximum number of file handles. Linux 2.6 always
102reports 0 as the number of free file handles -- this is not an
103error, it just means that the number of allocated file handles
104exactly matches the number of used file handles.
105
106Attempts to allocate more file descriptors than file-max are
107reported with printk, look for "VFS: file-max limit <number>
108reached".
109==============================================================
110
111nr_open:
112
113This denotes the maximum number of file-handles a process can
114allocate. Default value is 1024*1024 (1048576) which should be
115enough for most machines. Actual limit depends on RLIMIT_NOFILE
116resource limit.
117
118==============================================================
119
120inode-max, inode-nr & inode-state:
121
122As with file handles, the kernel allocates the inode structures
123dynamically, but can't free them yet.
124
125The value in inode-max denotes the maximum number of inode
126handlers. This value should be 3-4 times larger than the value
127in file-max, since stdin, stdout and network sockets also
128need an inode struct to handle them. When you regularly run
129out of inodes, you need to increase this value.
130
131The file inode-nr contains the first two items from
132inode-state, so we'll skip to that file...
133
134Inode-state contains three actual numbers and four dummies.
135The actual numbers are, in order of appearance, nr_inodes,
136nr_free_inodes and preshrink.
137
138Nr_inodes stands for the number of inodes the system has
139allocated, this can be slightly more than inode-max because
140Linux allocates them one pageful at a time.
141
142Nr_free_inodes represents the number of free inodes (?) and
143preshrink is nonzero when the nr_inodes > inode-max and the
144system needs to prune the inode list instead of allocating
145more.
146
147==============================================================
148
149overflowgid & overflowuid:
150
151Some filesystems only support 16-bit UIDs and GIDs, although in Linux
152UIDs and GIDs are 32 bits. When one of these filesystems is mounted
153with writes enabled, any UID or GID that would exceed 65535 is translated
154to a fixed value before being written to disk.
155
156These sysctls allow you to change the value of the fixed UID and GID.
157The default is 65534.
158
159==============================================================
160
161suid_dumpable:
162
163This value can be used to query and set the core dump mode for setuid
164or otherwise protected/tainted binaries. The modes are
165
1660 - (default) - traditional behaviour. Any process which has changed
167 privilege levels or is execute only will not be dumped
1681 - (debug) - all processes dump core when possible. The core dump is
169 owned by the current user and no security is applied. This is
170 intended for system debugging situations only. Ptrace is unchecked.
1712 - (suidsafe) - any binary which normally would not be dumped is dumped
172 readable by root only. This allows the end user to remove
173 such a dump but not access it directly. For security reasons
174 core dumps in this mode will not overwrite one another or
175 other files. This mode is appropriate when administrators are
176 attempting to debug problems in a normal environment.
177
178==============================================================
179
180super-max & super-nr:
181
182These numbers control the maximum number of superblocks, and
183thus the maximum number of mounted filesystems the kernel
184can have. You only need to increase super-max if you need to
185mount more filesystems than the current value in super-max
186allows you to.
187
188==============================================================
189
190aio-nr & aio-max-nr:
191
192aio-nr shows the current system-wide number of asynchronous io
193requests. aio-max-nr allows you to change the maximum value
194aio-nr can grow to.
195
196==============================================================
197
198
1992. /proc/sys/fs/binfmt_misc
200----------------------------------------------------------
201
202Documentation for the files in /proc/sys/fs/binfmt_misc is
203in Documentation/binfmt_misc.txt.
204
205
2063. /proc/sys/fs/mqueue - POSIX message queues filesystem
207----------------------------------------------------------
208
209The "mqueue" filesystem provides the necessary kernel features to enable the
210creation of a user space library that implements the POSIX message queues
211API (as noted by the MSG tag in the POSIX 1003.1-2001 version of the System
212Interfaces specification.)
213
214The "mqueue" filesystem contains values for determining/setting the amount of
215resources used by the file system.
216
217/proc/sys/fs/mqueue/queues_max is a read/write file for setting/getting the
218maximum number of message queues allowed on the system.
219
220/proc/sys/fs/mqueue/msg_max is a read/write file for setting/getting the
221maximum number of messages in a queue value. In fact it is the limiting value
222for another (user) limit which is set in mq_open invocation. This attribute of
223a queue must be less or equal then msg_max.
224
225/proc/sys/fs/mqueue/msgsize_max is a read/write file for setting/getting the
226maximum message size value (it is every message queue's attribute set during
227its creation).
228
229
2304. /proc/sys/fs/epoll - Configuration options for the epoll interface
231--------------------------------------------------------
232
233This directory contains configuration options for the epoll(7) interface.
234
235max_user_instances
236------------------
237
238This is the maximum number of epoll file descriptors that a single user can
239have open at a given time. The default value is 128, and should be enough
240for normal users.
241
242max_user_watches
243----------------
244
245Every epoll file descriptor can store a number of files to be monitored
246for event readiness. Each one of these monitored files constitutes a "watch".
247This configuration option sets the maximum number of "watches" that are
248allowed for each user.
249Each "watch" costs roughly 90 bytes on a 32bit kernel, and roughly 160 bytes
250on a 64bit one.
251The current default value for max_user_watches is the 1/32 of the available
252low memory, divided for the "watch" cost in bytes.
253